Fix Readme
Abstract
This readme provides information for IBM WebSphere MQ for HP NonStop Server, Version 5.3.1, Fix Pack 17.
Content
--------------------------------------------------------------------
CONTENTS
========
Introduction
About this release
Installation, migration, upgrade and configuration information
Uninstallation information
Known limitations, problems and workarounds
Documentation updates
Contacting IBM software support
Notices and Trademarks
INTRODUCTION
============
Welcome to IBM(R) WebSphere(R) MQ for HP NonStop Server, Version 5.3.1.17
This release notes file applies to the latest WebSphere MQ cross-platform
books (for Version 5.3), and to the WebSphere MQ for HP NonStop Server
Version 5.3 specific books (WebSphere MQ for HP NonStop Server System
Administration Guide and WebSphere MQ for HP NonStop Server Quick
Beginnings).
The content of these release notes applies to the WebSphere MQ for HP
NonStop Server product unless otherwise stated.
This release notes file contains information that was not available in
time for our publications. In addition to this file, README,
you can find more information on the WebSphere MQ Web site:
http://www.ibm.com/software/integration/wmq/
For current information on known problems and available fixes,
SupportPacs(TM), product documentation and online versions of this and
other readme files see the Support page of the WebSphere MQ Web site:
http://www.ibm.com/software/integration/wmq/support
ABOUT THIS RELEASE
==================
Nomenclature
------------
The terms "WMQ V5.3.1" and "WMQ V5.3.1.0" both refer to the same WMQ Refresh
Pack without subsequent service installed. Throughout this readme, references
to "WMQ V5.3.1.x" refer to WMQ 5.3.1 with or without subsequent service
installed.
New in this release
-------------------
and is designated V5.3.1.17, with associated APAR IT35715.
performed since WMQ 5.3.1 was released.
following version procedure strings:
identifier)
following VPROC information:
T0085H06_OpenSSL_1_1_1j__16_Feb_2021
The Non-native library distributed with this fixpack has the following
version procedure string:
with the current non native library.
Note that non-native applications do not support MQCONNX, even though the
function can be called from within applications.
since V5317 was released. The current set of recommended solutions is
described later on in this readme.
products are at SPR levels that incorporate these fixes (solutions) as a
preventive measure. IBM has tested WMQ with these fixes in our own
environment before making the recommendations.
Important note about SSL channels
---------------------------------
with known vulnerabilities discovered since the version shipped with V5.3.1.
A new version of the SSLupdate.pdf document, which was first provided with
Fixpack V5.3.1.10, is delivered with this release and is located in the
<install_path>/opt/mqm/READMES/en_US directory. If you use SSL channels
you should review the revised version of SSLupdate.pdf prior to installing
this fixpack.
removed:
SSLv3.0:
DES_SHA_EXPORT1024
RC4_56_SHA_EXPORT1024
RC4_MD5_EXPORT
RC2_MD5_EXPORT
DES_SHA_EXPORT
TLSv1.0:
TLS_RSA_WITH_DES_CBC_SHA
above have been removed and will result in SSL channel failure if being
configured.
consult documentation for any instructions when adding new environment variables.
Important note about WebSphere MQ V5.3 classes for Java and JMS
---------------------------------------------------------------
Fixpack V5.3.1.10 resolved an incompatibility between NonStop Java 7 and the
WMQ product libraries - that fix is also included in V5.3.1.17.
The method used to configure Java in V5.3.1.10 and later differs from
that in releases prior to V5.3.1.10. The Java.pdf document shipped in the
<install_path>/opt/mqm/READMES/en_US directory was updated to reflect the
change. Java/JMS users migrating from versions prior to 5.3.1.10 should
review the updated document.
Important note about PUT library support for V5.3.1
---------------------------------------
Since FixPack V5.3.1.10 Java 7 has been supported. This change introduced
MQ libraries utilizing the PUT user thread model.
As of Fixpack V5.3.1.14 these libraries have been approved for C native PIC
OSS usage.
PUT threaded applications have to be build using -D_PUT_MODEL_ and be linked
against ZPUTDLL as well as /opt/mqm/lib/put/libmqm_r.so.
pthreads have to be created with a stack size of at least 2097152 Bytes
(~2MB).
Important note about instmqm for V5.3.1
---------------------------------------
Since FixPack V5.3.1.5, IBM provided a modified WebSphere MQ product
installation script, instmqm, for any level of V5.3.1. The new installation
script includes a workaround for the OS problem introduced in G06.29/H06.06
where the OSS 'cp' command creates Guardian files in Format-2 form during an
installation rather than Format-1. This change caused problems binding and
compiling non-native and native COBOL applications, as well as wasting a lot of
disk space because of the very large default extents settings for the Format-2
files created by OSS cp.
Instmqm has been modified in FixPack V5.3.1.5 to work around this change in OSS
cp by forcing all Guardian files in an installation to be created as Format-1.
The use of the new installation script is recommended for all new V5.3.1
installations.
Existing installations that are not affected by the application relink or
rebind problems can remain as they are.
Important note about queue files for WebSphere V5.3.1
---------------------------------------
As of FixPack release V5.3.1.15, WebSphere MQ supports format 2 queue files.
Should altmqfls be called with parameters which would result in queue files
greater than 2GB, it will now create format 2 files instead of returning
an error.
Please see APAR IT24742 for further information.
Product fix history
-------------------
CVE-2021-23839
CVE-2021-23840
CVE-2021-23841
APAR IT35713 - Mitigation for the following security vulnerabilities in OpenSSL:
CVE-2020-1971 EDIPARTYNAME NULL pointer de-reference
APAR IT32657 - Mitigation for the following security vulnerabilities in openssl
CVE-2019-1551 rsaz_512_sqr overflow bug on x86_64
APAR IT31015 - Mitigation for the following security vulnerabilities in openssl
CVE-2019-1547 ECDSA remote timing attack
CVE-2019-1549 Fork protection
CVE-2019-1563 Padding Oracle in PKCS7_dataDecode and
CMS_decrypt_set1_pkey
=====================================
The following problems are resolved in FixPack V5.3.1.16:
APAR IT26660 - Mitigation for the following security vulnerabilities in OpenSSL:
Client DoS due to large DH parameter (CVE-2018-0732)
Cache timing vulnerability in RSA Key Generation (CVE-2018-0737)
APAR IT28065 - A queue server reports a FDC and can go down if it receives a message
with garbage content.
APAR IT28437 - After installing WebSphere MQ 5.3.1.15 patch 1 on NonStop, the MQ
version still shows version 5.3.1.15.
APAR IT27853 - Expired entries were not removed from cluster cache.
APAR IT28733 - MQ v5.3 for NonStop wmqtrig script leads to unpredictabe triggering
result due to handling of param file.
APAR IT28993 - Application cannot reconnect to queue manager after cpu goes
down and fails with error 2059 on retry.
APAR IT29055 - The counter of queue depth is not decremented correctly.
APAR IT29452 - Enhanced CPU detection within SDCP script.
APAR IT29514 - Improved error handling during REFRESH CLUSTER command when CLUSNL()
attribute points to a namelist that doesn't exist.
APAR IT29707 - Mitigation for the following security vulnerabilities in OpenSSL:
(CVE-2019-1559)
APAR IT29927 - Queue manager gets unresponsive and even control commands get
hung (amqrfdm, runmqsc, endmqm, amqrrmit, ...).
APAR IT29955 - In a cluster CSQX457I CSQXREPO repository available is issued
then CSQX419I and CSQX448E as the repository manager stops.
The following problems were resolved in FixPack V5.3.1.15:
APAR IT24026 - Mitigation for the following security vulnerabilities in OpenSSL:
Malformed X.509 IPAddressFamily could cause OOB read (CVE-2017-3735).
APAR IT24145 - In a situation where report messages are re-routed to a
dead letter queue they may have an incorrect 'format' field
in their MQDLH header.
APAR IT24388 - MQ for NonStop server, reinitialisation of multiple slave
repository mangers in parallel results in deadlock, FDC
RM220005 with AMQ9511 and AMQ9448 logged.
APAR IT24742 - Enabling of FORMAT2 support for queue files.
APAR IT24999 - If a cluster queue is set to PUT(DISABLED) and messages
are in transit the target queue manager can throw an FDC
"incorrect application process opener context stored in
mqopen for mqput" or the cluster channel to this queue
manager may be terminated.
APAR IT25657 - SDCP terminates without having created output or incomplete
output.
APAR IT25659 - Non-native applications may abend in READX, due to implicit
cast of readcnt to the wrong type.
APAR IT25662 - Messages can be put on PUT(DISABLED) local queues, if their
CLUS() attribute is set.
APAR IT25665 - Errors AMQ9511 and AMQ9448 are logged along with FFST regarding
probeid: RM220005 and probe type: AMQ9511.
APAR IT25666 - Errors AMQ9511 and AMQ9448 are logged along with FFST regarding
probeid: RM220005 and probe type: AMQ9511.
APAR IT25667 - AMQRRMFA processes create redundant opens to other AMQRRMFA
processes, when one reinitializes due to an error
(error logged AMQ9448).
APAR IT25669 - Errors AMQ9511 and AMQ9448 are logged along with FFST regarding
probeid: RM220005 and probe type: AMQ9511.
APAR IT25671 - Errors AMQ9511 and AMQ9448 are logged along with FFST regarding
probeid: RM220005 and probe type: AMQ9511.
APAR IT25677 - When a close message is overtaken by a cpu down message the
cleanup done as a reaction to the close will fail since the
data it tries to access is not valid anymore, resulting in
error 4022 when trying to lock a mutex.
APAR IT25692 - Mitigation for the following security vulnerabilities in OpenSSL:
Constructed ASN.1 types with a recursive definition could
exceed the stack (CVE-2018-0739).
The following problems were resolved in FixPack V5.3.1.14:
APAR IT21769 - IBM MQ 8 for HP Nonstop is delivered with export tools running against
an existing MQ5 installation.
When amqmexpc is run against an queue manager of v5.3.13 and later the
mqchsvr process will terminate and the export will fail. A compatible
binary is delivered with FixPack 14.
APAR IT21970 - Two receivers unexpectedly get the same message using MQGET with
MQGMO_BROWSE option at same time even though specifying MQGMO_LOCK
option.
APAR IT21652 - Under certain situations LQMA processes were using an excessive amount
of dynamic memory, which results in a critical resource level for the
whole system.
APAR IT21112 - Cluster sender channels fail to restart following a CPU failure on a
2 CPU system. This was caused by a racing condition during the takeover
process and has been fixed with this fixpack.
APAR IT19592 - Applications may receive the error 'MQRC_STORAGE_NOT_AVAILABLE' when
issuing a MQCONN call. This was caused due to not freed quick cells
when applications did not issue a MQDISC call, since the required
task needed to cleaned up was not active for WebSphere MQ on HP NonStop.
This has been changed for FixPack 14.
APAR IT20043 - Inconsistent treatment of CLUSTER QMGR ALIAS may lead to error
2087 'MQRC_UNKNOWN_REMOTE_Q_MGR'.
APAR IZ10060 fixed this issue has been back ported to IBM Websphere
MQ for HP NonStop Server with this release.
APAR IT21677 - IBM Websphere MQ5.3.1.13 introduced a regression where runmqsc
failed with error message "AMQ8242: SSLCIPH definition wrong."
if the ALTER CHANNEL command was used with SSLCIPH('') to clear
the used CipherSpec.
This has been fixed with FP14.
APAR IT21856 - LQMA process may be rendered unresponsive after a pthread_mutex_lock
failure. No new connections can be established and applications issuing
a MQCONN call appear to hang when trying to open the LQMA process.
APAR IT21950 - IBM Websphere MQ5.3.1.10 introduced support for Java 7 along with
new libraries. If an installation earlier than v5.3.1.10 was upgraded
using the 'svcmqm' utility, those new libraries were not moved to the
installation.
The following problems were resolved in FixPack V5.3.1.13:
APAR IT18767 - FixPack V5.3.1.12 introduced a regression which caused unthreaded
LQMAs to terminate unexpectedly during startup. This has been
fixed with this release.
APAR IT18769 - Within a threaded environment the name resolution of a starting
remote channel could cause delays. This has been changed as such
that the name resolution is done in a more thread friendly
manner.
APAR IT18770 - In rare cases a racing condition during LQMA startup could cause
a MQCONN request to fail with RC 2195. This was caused due the
starting agent sending its registration request before the
Execution Controller has finished the creation process. The
request is then rejected since the Execution Controller does not
accept requests by unknown agents. This has been fixed.
APAR IT18593 - OSS environment variables which contain '_' were not usable in
the Guardian environment. This has been changed as such that
'_' can be substituted with '^' in a PARAM and will be correctly
interpreted.
APAR IT18625 - SSL support has been changed to accommodate the changes done by
the latest OpenSSL update. This offers an alternative way of
SSL configurations on HP NonStop, based on the way it is done
for IBM MQ on other platforms.
This update deprecates cipherspecs TLS_RSA_WITH_DES_CBC_SHA,
TLS_RSA_WITH_3DES_EDE_CBC_SHA and TLS_RSA_WITH_NULL_SHA256 since
they are considered weak (see SSLUpdate.pdf).
APAR IT18692 - OpenSSL library update to version 1.0.2j. Mitigation for:
Fix handling of OCSP Status Request extension (CVE-2016-6304),
prevention of possible out of bounds write (CVE-2016-2182),
refined limit checking preventing undefined behavior(CVE-2016-2177),
added missing length checks, preventing DoS risk (CVE-2016-6306),
CRL sanity check (CVE-2016-7052).
The following problems were resolved in FixPack V5.3.1.12:
APAR IT08589 - WebSphere MQ V6 or V7 queues become "missing" from clusters. Application
calls to MQOPEN (sometimes MQPUT1, MQPUT) suffer queue name lookup
errors (Examples: 2085, 2082) when they try to access the affected
cluster queues.
APAR IT10388 - When the receiving QMgr has its channel set with SSLCAUTH(OPTIONAL),
the Nonstop Queue Manager insists on sending a certificate to
identify itself. The remote side, however, does not require one
APAR IT11557 - Updated OpenSSL library to version 1.0.2h. This update deprecates sslchiph
cipherspec TLS_RSA_WITH_DES_CBC_SHA and fixes various security
vulnerabilities.
APAR IT12856 - When a node is removed from an MQ cluster and there is an automatically
defined cluster sender channel from a MQ node on NonStop to the removed
node, the sender channel on NonStop can go into state INITIALIZING and
stay there for up to 60 minutes instead of being deleted after the first
retry interval has expired.
APAR IT12875 - When a message is put into a remote/cluster queue within a TMF user
transaction and while the corresponding transmit queue is empty or almost
empty, it can take up to 60 seconds after the end of the transaction
until the message is transmitted. As a side effect automatic channel
starts can also be delayed.
APAR IT12894 - Java 7 SSL clients behave differently than previous versions which
resulted in an RC 2009 error on its connection attempt. This
is fixed with V.5.3.1.12.
APAR IT14169 - When a CPU is stopped, some sender channels were not recovered
and stayed in inactive state.
APAR IT15316 - Memory allocation during MCA and LQMA creation and initialization was
not handled properly and may result in an unresponsive system.
APAR IT15317 - Under some conditions agents stay alive but don't accept new
connections. That is the case if an agent has reached the maximum
number of connections it is allowed to handle during its lifetime
but the agent can't terminate before the last connection has been
closed.
APAR IT15490 - A configured process name rule has not been applied because it
was defined with lower case characters, while the operating
system interface delivered upper case process names. Since the name
comparison then fails the configuration is not applied.
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.12:
IN000100 - The CRTMQM command may fail with error 1017 if queue managers are
created in parallel.
IN000101 - The installer has been enhanced to properly support SMF virtual disks
if the system revision is on H06.26 and newer or J06.15 and newer.
IN000102 - Removed access to uninitialized memory within mqconn which may result in
RC 2059 MQRC_Q_MGR_NOT_AVAILABLE.
The following problems were resolved in FixPack V5.3.1.11:
APAR IT03572 - In V5.3.1.10, the dspmqver -V command displays the string
"VPROC" rather than the VPROC string encoded in the product.
V5.3.1.11 corrects this problem.
APAR IT04083 - Error 30 received by Queue Server in a complex Queue Manager
configuration using MQGMO_SET_SIGNAL. This is a result of
the limit on the number of outstanding messages sent by the
process not being set to the same value as the internal counter.
This results in the Queue Server reporting "Guardian error 30
attempting to send signal notification" in the queue manager
error log.
APAR IT04533 - Cluster sender channels fail to start following a CPU failure
on a 2 CPU system when the Repository Manager and Channel Server
are both in the same CPU. Following the CPU failure, the Queue
Manager error log repeatedly reports message AMQ9422
"Repository manager error, RC=545284114"
APAR IT04876 - Some ssl channels with mismatched sslchiph cipherspecs run
successfully because protocol versions are not compared.
APAR IT05353 - Mitigation for SSLV3 POODLE Attack - CVE-2104-3566
APAR IT07330 - WebSphere MQ V5.3.X Process does not write FDC (First
Failure Symptom Report) records when the current FDC file is
physically full. This can result in processes experiencing error
conditions, but the error condition not being reported in the
FDC file. This fix changes the error handling to identify this
scenario and switch to a new FDC file with an incremented
sequence number
APAR IV19854 - The master repository manager in a partial repository
unnecessarily subscribes to each cluster queue manager that
it is aware of. This causes a significant increase in the number
of subscriptions that are made, and in large clusters this can
cause performance problems.
APAR IY87702 - SIGSEGV resulting from failed getpwnam. In some circumstances
getpwnam can return a success value in errno, but a null
pointer as the user record. This results in a SIGSEGV from
amqoamd. This change adds a check for this condition and handles
it correctly.
APAR IY91357 - Altering a channel definition unintentionally resets the
SSLPEER() attribute for the channel.
When a channel with SSLPEER information is modified to include
either MSGEXIT or MSGDATA attributes, the SSLPEER attribute
becomes blank. V5.3.1.11 resolves this problem.
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.11:
909 - In the GA release, the TFILE was opened with depth 100 in the
Guardian NonStop process pairs. In configurations with large
large numbers of connecting applications/threads, this could
result in probe QS270000 errors from nspPrimary, with an Arith1
value of 83 (0083 Attempt to begin more concurrent transactions
than can be handled). This change increases the TFILE open depth
to 1000. The default value can be changed from 1000 to a value
between 100 and 1000 using MQTFILEDEPTH environment variable.
If a value < 100 or > 1000 is specified, a value of 100 will
be used, and an FDC will be reported with the text
"Invalid MQTFILEDEPTH specified - using default"
1380 - In the cra library, several FFST code sites report an error
without reporting the name of the channel involved in the error
condition. This change adds the channel name to the FFST report
in cases where the channel name is available to the calling code
1471 - In earlier releases of 5.3.1, if a garbage argument string is
supplied to dspmqfls, the command abends rather than reporting a
usage message. V5.3.1.11 resolves this problem
1597 - In earlier releases of 5.3.1, the dspmqfls output shows a
Queue/Status server value. Status servers were part of the V5.1
architecture but are not present in 5.3.1. V5.3.1.11 changes the
message to reflect the fact that all objects are now managed by
Queue Servers.
2230 - In earlier releases of 5.3.1, the runmqsc command
"DISPLAY QSTATUS(*) TYPE(HANDLE) ALL" does not cleanly handle a
scenario where there are more than 500 handles. Attempts to use
the command when there are more than 500 handles to be returned
results in FFSTs with probe NS013000 from nspReply, and probe
QS264002 from qslHandleHandleStatus. See later section on
"KNOWN LIMITATIONS, PROBLEMS AND WORKAROUNDS"
4158 - Slow memory leak in MQDISC in non-native library. In prior
releases, the non-native implementation of MQDISC contained a
slow memory leak. This resulted in applications performing
repeated MQCONN/MQDISC operations during the process lifetime
eventually receiving a MQRC_STORAGE_NOT_AVAILABLE error from
MQCONN, and an FFST for probe ZS219001 from zstInsertPCD. The
problem can also manifest as a probe XC130006 from
xehExceptionHandler inside an MQOPEN call.
4395 - If memory allocation for internal working storage fails during
an attempt to create a dynamic queue, the LQMA assoicated with
the operation will ABEND, rather than returning a completion
code of 2071 (MQRC_STORAGE_NOT_AVAILABLE). This change resolves
the problem.
4421 - If an attempt to start an OAM server detects an existing OAM
server in the CPU, in prior releases, the FFST generated did not
include the process name. This change amends the FFST to
include the process name of the offending process, where that
name can be determined.
4428 - The altmqfls command "resetmeasure" was incorrectly documented
in the original version of the SysAdmin guide. See later section
"DOCUMENTATION UPDATES"
4496 - In earlier versions of this readme file, the instructions on
how to reassign Queue Server for
SYSTEM.CLUSTER.TRANSMISSION.QUEUE are incorrect. Refer to the
"KNOWN LIMITATIONS, PROBLEMS AND WORKAROUNDS" section for the
correct procedure.
4523 - If an OAM server attempts to start in a CPU, and there is
already an OAM server registered, prior to V5.3.1.11, the OAM
subsystem reported that there was a rogue process, but not
the name of the process. This change enhances the error
reporting to report the name of the rogue process.
4726 - Prior to this fixpack, the instmqm script did not check for
saveabend files and FFSTs generated during the validation
phase of the installation. V5.3.1.11 adds this check.
4774 - The cobol binding library, MQMCB, was shipped without symbols in
FixPack V5.3.1.10, which causes applications to fail. V5.3.1.11
resolves this issue.
4800 - During instmqm, the script checks for the UNIX socket server.
In prior releases, in the event that this check failed, the
error message referred to the old UNIX socket server ($ZPMON).
This has been changed to refer to the new UNIX socket subsystem
$ZLSnn, "nn" is the CPU number.
APAR IC85889 - UNEXPECTED IPC MESSAGES RECEIVED FOLLOWING PROCESS
TERMINATION CAUSE EC TO ABEND REPEATEDLY
If a process sends an unexpected message to the MQECSVR
process pair, the primary process will abend resulting in a
takeover by the backup process. The new primary process will
then abend. The only resolution is to restart the queue manager.
APAR IC87007 - PRIMARY QUEUE SERVER ABEND FOLLOWING TRANSACTION ABORT
In some circumstances, aborts of global units of work
involving MQGETs of messages greater than 52k will result in an
abend of the primary queue server responsible for the queue.
This renders the queue manager unresponsive
APAR IC87627 - WEBSPHERE MQ V5.3.1.8 FDC FILES DO NOT CONTAIN ENOUGH ERROR
INFORMATION
APAR IC89128 - LISTENER FAILS TO START FOLLOWING UPGRADE TO
HP J06.14/H06.25 OR T9050J01-AWT/T9050H02-AWS
In some configurations, the listener process will not run
following an upgrade to J06.14/H06.25 or the installation of
T9050J01-AWT/T9050H02-AWS. The failure is dependent on the
number of OSS processes running and their distribution between
the CPUs on the system. This problem also affects endmqlsr.
APAR IC89751 - CLUSSDR Channels do not restart without manual intervention
after CPU crash.
If a CLUSSDR or SDR channel is running in an MCA that is in the
same CPU as the Primary Channel Server, and that channel has
been running for at least 5 minutes, and that CPU crashes, the
channel will not automatically restart.
APAR IC92511 - TERMINATION OF SLAVE REPOSITORY MANAGER RESULTS IN CLUSTER
CHANNEL DELETION. In some cases this can result in a manually
stopped cluster sender channel starting unexpectedly when they
are dynamically recreated.
APAR IC92570 - MULTIPLE CLUSTER RECIEVER CHANNELS OR LONG RUNNING CLUSRCVR
CHANNELS CAUSE METADATA FDC'S
APAR IC93289 - WEBSPHERE MQ R531 ON HP-NSS CLUSTER PUB SUB MESSAGES FAIL
WITH AMQ9538 - COMMIT CONTROL ERROR
APAR IC94647 - Closing dynamic queues with MQCO_DELETE or MQCO_DELETE_PURGE
with an outstanding MQGET with signal results in multiple FDCs
and an orphaned dynamic queue
APAR IC96947 - Backup Queue Server generates "Open handle points to unused
entry in default page" FFST following LQMA termination
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.10:
4607 - Analyze lock logic in queue server takeover does not take account
of temporary dynamic queues
4679 - Exhaustion of message quickcell (MQC) space available to queue
server results in unstable queue manager
4563 - RDF compatible message overflow file setting requires queue
manager restart
4560 - Slave repository managers issue channel manipulation commands
that should be performed only by the master
4557 - Write NonStop specific trace information to product standard
trace files rather than platform-specific file
4556 - Internal Queue manager query array size can cause problems
with clusters containing more than 10 queue managers
The following problems were resolved in FixPack V5.3.1.9:
APAR IC51322 - CODE CHANGES FOR OPTIMIZATION OF DEFAULT CONVERSION
Delay in processing seen during the default conversion between
client and server. The channel attempts normal conversion first,
including attempting to load a conversion table before dropping
through to the default conversion. Once default is required
for a specific set of code pages, this is now remembered
and the attempt to perform normal conversion skipped.
APAR IY76799 - A FAILED CALL TO GETPEERNAME RESULTS IN A LEAKED SOCKET FILE
WMQ amqrmppa (channel) process leaks a file descriptor in
the rare circumstance that a call to the getpeername function
fails. Such a failed call is reported in the WMQ error logs for
the queue manager via an AMQ9213 message reporting that the
getpeername call has failed. The failure of
the getpeername call is the result of preoblems external to
WMQ. This problem could also cause the WMQ listener process
(runmqlsr) to run out of file descriptors, if there is
no queue manager running, and lots of connections come in,
whilst these connections fail with getpeername problems.
APAR IY95706 - SIGSEGV IMMEDIATELY FOLLOWING RETURN FROM ZFUDOESOBJECTEXIST
An amqzlga0 process or amqrmppa process will raise an XC130004
FDC and the amqzlga0 process will end. The abrupt termination of
the amqzlga0/amqrmppa process may cause channels or applications
connected to the queue manager to fail
APAR IZ06131 - RECOVER FROM SYSTEM.AUTH.DATA.QUEUE CORRUPTED BY USER ERROR
Applications receive MQRC_NOT_AUTHORIZED (2035) error
setmqaut returns MQRC_UNKNOWN_OBJECT_NAME (2085) and fails to
set new authorities. Records are missing from the
SYSTEM.AUTH.DATA.QUEUE, determined by a mismatch between the
output from amqoamd -m -s, and the authorities that customers
believe they have set for users and objects. This fix allows
recovery from this situation using setmqaut without the need
to recreate the queue manager
APAR IC74903 - SSLPeer value slash (/) causes SSL handshake to fail
Using slashes in the Distinguished Name name fields such as CN,
O, OU, L, will fail SSLPEER value verification, and as such SSL
Handshake, due to an MQ parsing error. MQ uses slashes (/) to
delimit the distinguished name values when matching the SSLPEER
value with the information contained in the certificate.
APAR IC82919 - xcsAllocateMemBlock returning xecS_E_NO_MEM
Queue manager (receiver channel side) with API exits generate
FDCs containing probe ZF137003 and error log entries for AMQ9518
in the form of - File '/mqs/WMQtest/var/mqm/@ipcc/AMQRFCDA.DAT'
not found.
APAR IC80942 - Authority commands produced by amqoamd command is not usable if
authorization is "+None"
Using the setmqaut command with commands generated by amqoamd
results in a failure e.g
"setmqaut -m qmgrname -n name -t queue -g group +None"
returns:
"AMQ7097: You gave an authorization specification that is not
valid."
APAR IC81429 - MQRC error 2003 received if an overflow file record is missing
If a portion of a message that should be stored the queue
overflow file is missing, the generation of an FFST is
suppressed and a 2003 reason code is returned. All subsequent
non-specific MQGETs from the queue fail, since the partial
message cannot be retrieved.
APAR IC81420 - Queue Server abend during simultaneous GET and BROWSE
When a message larger than 56k is being browsed (MQGET with
one of the MQGMO_BROWSE_* options) from a client application
using message groups, and another process is simultaneously
performing a destructive MQGET on the same message. The primary
queue server terminates, and more or more of the following
probes are generated :QS003002, AO211001, AO200002, ZI074001,
NS026006
APAR IC81367 - FixPack 5318 MQMCB Guardian library is not usable
The COBOL wrapper libraries were changed from type LINKFILE to
DLL in the v5.3.1.8 Fixpack, and the packaging unintentionally
stripped the symbols from the Guardian variant of the library
ZWMQBIN.MQMCB. During the link phase of a cobal program build
the following fatal error is encountered:
"Cannot use file specified in CONSULT or SEARCH directive"
Problem affects building in Guardian only.
APAR IC83299 - CPU Failure during mqget of persistent messages in global
unit of work results in incorrectly deleted records in queue and
overflow files
If a CPU fails where multiple applications are performing FIFO
MQGETs from the same queue, and the following conditions are
true:
* The primary queue server is running in the failing CPU
* Some (but not all) of the applications are running in the
failing CPU
* None of the LQMAs are running in the failing CPU
* The applications are using global units of work
* Applications in the failing CPU have completed MQGET operations
but not committed the transactions
There is a failure window where an application not in the
failing CPU will remove an additional message record from the
queue file.
APAR IC83197 - Queue Server open handle management cannot handle backup
restart cases where handles from non-contiguous pages are
synced by the primary
Queue servers with more than 3000 queue opens do not
correctly handle a NonStop takeover. The queue manager becomes
unresponsive and requires a restart to resolve the error
situation. When the queue server hangs, the backup queue server
produces a large number of FFST entries with the following
probe ids:
QS165004 from qslSetHandle
QS192007 from qslAddOpener
QS190005 from qslHandleOpen
APAR IC83569 - Persistent Reference messages sent over a channel cause Commit
Control Error
When a persistent reference message is put to a queue, FFST's
with probe CS075003 are generated with the following error
information
Major Errorcode :- rrcE_COMMIT_CONTROL_ERROR
Minor Errorcode :- OK
Comment1 :- Error 2232 returned from lpiSPIHPNSSTxInfo
In addition, Sender channels will go into a retry state and
queue manager error logs will contain 'Commit Control' errors
APAR IC83699 - Cluster cache maintenance asynchronous time values result in
cache content divergence
Repository queue managers report FFST with probe RM527002
from rrmHPNSSGetMetaPacket and Comment1 field
"Meta data mis-match Expected: metalen=4". The problem
resolved by this APAR is one of several possible causes of
these symptoms.
APAR IC83328 - Permanent dynamic queues are not deleted in some cases when
MQCO_DELETE is used on MQCLOSE.
Permanent dynamic queues are not deleted as expected after
termination of the last application that has the queue open.
APAR IC83228 - Repository manager/Channel server deadlock
Listener process hangs on queue manager startup for 5
minutes then generates an FFST with probe RM264002
from rfxConnectCache
Comment1 :- Gave up waiting for cache to be initialized
Comment2 :- Tried 300 times at 1 second intervals.
This problem also occurs sometimes when attempting
to use runmqsc while the queue manager is starting
APAR IY90524 - segv in xcsloadfunction for channel exit
WebSphere MQ channel process (amqrmppa) terminates with FFST
showing probe id XC130003 due to a SIGSEGV SIGNAL with
a function stack as follows:
MQM Function Stack
ccxResponder
rrxResponder
rriAcceptSess
rriInitExits
xcsLoadFunction
APAR IC81311 - MQGET implementation masks reason codes in some cases
Attempting to perform an MQGET after a Local Queue Manager
Agent (LQMA) process fails or has been forcibly terminated
results in a MQRC_UNEXPECTED_ERROR (2195). This is incorrect.
The result should be MQRC_CONNECTON_BROKEN (2009)
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.9:
4363 - MQA process opener field is sometimes corrupted in MQGET with
set signal operations
4388 - Guardian sample uses the wrong link directive
4339 - Shared hconn rendered unsuable by changes in 5317
4123 - FASTPATH bound connect processing does not correctly set reason
codes in some cases
4039 - Backup queue server abend during commit after MQGET
3826 - altmqfls --qsize does not report failure correctly
4116 - enhance SDCP to find installed version of OSSLS2 (T8620)
The following problems were resolved in FixPack V5.3.1.8:
APAR IC54121 - Cluster channel in a retrying state will no longer start
automatically if the following commands are issued
STOP CHANNEL() MODE(QUIESCE) STATUS(INACTIVE).
The channel will fail to start once communication
to the remote system is restored. A manual start and
stop of the channel is required to restore normal
channel operation
APAR IC54459 - Channel stays in binding state for a long time when it contains
an invalid conname value. During this period, all requests to
the channel are ignored.
APAR IC60204 - Comamnd server experiences a memory leak when namelist
inquireies are performed with names values.
The leak is observed when a PCF MQCMD_INQUIRE_NAMELIST or
MQCMD_INQUIRE_NAMELIST_NAMES is requested that has one or
more names attributes.
APAR IC70168 - High CPU usage from backup Queue Server performing browse
operations on queues with high queue depth.
When a queue has a very high queue depth (CURDEPTH) a browse
can cause the cpu usage of the backup Queue Server to increase
to 100%. Backup queue server CPU usage becomes significant at
a queue depth of 15000, with CPU use reaching 100% at a queue
depth of approx. 38000 messages.
APAR IC70947 - qmproc.ini file validation does not detect incorrect CPU syntax.
CPU lists in the qmproc.ini file that do not use comma
characters to separate CPU numbers in the list are ignored, but
they are not reported as an error. This can lead to unexpected
CPU assignements for WMQ processes.
APAR IC71839 - RESET QUEUE STATISTICS PCF returns incorrect values.
When using WebSphere MQ v5.3 for HP NonStop Server PCF command
Reset Queue Statistics, intermittently some of the values
associated with a queue will not be returned correctly.
* HighQDepth - The maximum number of messages on the queue
since the statistics were last reset.
(parameter identifier: MQIA_HIGH_Q_DEPTH).
* MsgEnqCount - The number of messages enqueued (the number of
MQPUT calls to the queue), since the
statistics were last reset.
(parameter identifier: MQIA_MSG_ENQ_COUNT).
* MsgDeqCount - The number of messages dequeued (the number of
MQGET calls to the queue), since the
statistics were last reset.
(parameter identifier: MQIA_MSG_DEQ_COUNT).
Incorrect values may also be returned incorrectly for the
command Inquire Queue Status
* LastGetTime - Time at which the last message was
destructively read from the queue
(parameter identifier: MQCACF_LAST_GET_TIME).
* LastPutTime - Time at which the last message was
successfully put to the queue
(parameter identifier: MQCACF_LAST_PUT_TIME).
APAR IC71912 - MQMC Channel menu display shows incorrect channel status.
In some instances, the MQMC Channel Menu display will not
show a change in channel status, and attempts to refresh the
screen or recycle the MQS-MQMSVR Pathway server do not correct
the problem. RUNMQSC continues to show the correct status.
The MQMC channel monitor panel does show the running channels
correctly.
APAR IC73800 - A queue manager with the MaxUnthreadeAgents parameter defined
in the QMPROC.INI file with a value greater than 812
reports unexpected process termination FDCs and or ERROR 22
FDCs.
APAR IC74994 - Queue manager reports message sequence number error and
produces FDCs with probe CS094005, major error code
rrcE_CREATE_SYNC_FAILED following a queue manager
restart.
APAR IC75298 - In some complex cluster configuations with large numbers of
cluster members, large numbers of objects, or frequent changes
to cluster objects, the repository managers in a queue
manager are unable to distribute a complete set of object
metadata information, resulting in repeated FDCs from
rrmHPNSSGetMetaPacket, with probe RM527001, and cluster objects
not being visible in some CPUs in the queue manager reporting the
problems.
The fix for the problem adds a new configurable parameter to
allow the repository metadata buffers to be increased to handle
larger configurations, and changes the reporting of the metadata
errors to include information on the amount of storage requested
by the repository managers. The default size of the buffer is
512K, this is sufficient for most configurations. If the buffer
size is insufficient, the queue manager reports FDCs from
rrmHPNSSPutMetaPacket that indicate the present size of the
buffer and the size demanded.
The buffer size is configured using an environment variable,
AMQ_CLUSTER_METABUFFER_KILOBYTES. To change the value, the
environment variable should be specified in the
"RepositoryManager" stanza of the qmproc.ini file of the queue
manager using the following syntax:
Env0n=AMQ_CLUSTER_METABUFFER_KILOBYTES=x
Where "n" is the number of the environment variable, and "x" is
the required new size of the buffer in kilobytes. In a default
configuration, environment variables are not present in the
RepositoryManager stanza, hence "n" will be 1. If the
configuration has existing environment variables specified in
this stanza, the value of "n" selected should be the next
available value.
APAR IC75356 - Partial repository queue managers in complex configurations
where applications connected to the partial repository queue
manager attempt to open large numbers of nonexistant objects
can suffer from a build up of subscription objects on the
SYSTEM.CLUSTER.REPOSITORY.QUEUE such that the TMF lock limits
for the volume containing the queue file are breached when the
queue is reconciled. This produces 2024 errors during operations
on the SYSTEM.CLUSTER.REPOSITORY.QUEUE The following error
message appears in the QMgr error logs.
EXPLANATION: The attempt to get messages from queue
'SYSTEM.CLUSTER.REPOSITORY.QUEUE' on queue manager 'xxx' failed
APAR IY57123 - Attempts to put to a clustered queue via a queue manager
alias when the queue has been opened using BIND_AS_QDEF
fail. Following this fix, queue name resolution functions
as described in the Application Programming Guide.
APAR IY78473 - Cluster workload management is not invoked if a queue is
resolved using clustered queue manager aliases where there
are multiple instances of the alias in the cluster.
APAR IY86606 - Cluster subscriptions are created for non-clustered queues
when MQOPEN is called with non-clustered ReplyToQ or
ReplyToQMgr. This can result in a build up of subscriptions
in partial repositories in the cluster. The error was
introduced by IY8473.
APAR IZ14977 - Queue manager cluster membership missing when namelists are
used to add and remove queue managers from clusters. This can
result in the queue manager not acting as a repository for
one or more clusters, or other queue managers in the cluster
not recognizing that a given queue manager is a repository
for the cluster.
APAR IZ20546 - The repository manager process (amqrrmfa) consumes high CPU
resources on an hourly basis for several minutes, and
applications are unable to issue messaging MQ API calls
during this period. The problem is observed only in
configurations where clustered queue manager aliases are
used and they resolve to more than 50 destinations.
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.8:
1228 - COBOL Binding library mqmcb does not contain VPROC information
1378 - FDC's cut as a result of errors from PATHWAY SPI operations do
not report the associated Guardian error code
1566 - WMQ service installation tool svcmqm does not correctly detect
and report that files it is attempting to modify are in use,
as is the case where WMQ applications are still runnning.
1606 - Certain FDCs containing comment text have the comment text
truncated
1710 - In some cases on heavily loaded systems, the EC re-allocates
an MCA that has already been told to terminate. This results in
FDC's cut with probe EC134000, from eclDeallocMCA.
2534 - WMQ service installation tool svcmqm does not log some aspects
of its progress, making diagnosing some installation problems
difficult
2856 - WMQ service installation tool should not attempt SECURE
operations in installations where SAFEGUARD is enabled
2943 - crtmqm does not correctly handle validation of the command line
if a specified CPU number is invalid
3037 - If the var/mqm/errors directory contains non-mqm user files
the WMQ service data collection tool, sdcp, does not capture the
MQM group FDC and ZZSA files.
3128 - WMQ Service data collection tool, sdcp, does not capture VPROC
of the AF_UNIX R2 socket process
The following changes were released in FixPack V5.3.1.7:
APAR IC65774 - Under given conditions, the response time measured for MQGET
operations when using SSL activated channels with multi-threaded
MCA agents are found to be longer than that measured when using
SSL activated channels with unthreaded MCA agents. This problem
is seen with both distributed and cluster queue managers.
Moreover, this problem does not occur prior to WMQ V5.3.1.5
release. A DELAY was introduced as part of an APAR(IC57744) fix
in WMQ V5.3.1.5 release for mutli-threaded SSL channels which
caused this difference in measured response time. The problem is
now rectified in this release.
APAR IC67032 - Improvement to LQMA FDC during MQCONN processing. At times, when
application dies during MQCONN processing, LQMA's generate FDCs
for this rather unusual event to let the user take any
corrective action and/or find the root cause. The problem can
occur with a standard bound application in a very narrow timing
window when the application connects with the LQMA agent but
dies before the LQMA gets a chance to read the incoming message
from the application. When LQMA detects this situation, it
cleans up and generates an FDC. But unfortunately, the FDC does
not contain the application information. To let the user
identify the mis-behaving application and to possibly take
corrective action, LQMA FDC will be updated to include
application information. The updated FDC will contain the
following information :
Comment1 :- Application died during MQCONN Processing.
Comment2 :- Application: <Application PID>
APAR IC67057 - Unused LQMA agents(processes or threads) are found during MQCONN
processing. During MQCONN processing if a standard bound
application fails to connect successfully to an allocated LQMA
process or thread (depending on your configuration), then that
LQMA process or thread remains in a hanging state for ever and
do not get re-used by Execution Controller for any further
MQCONN processing. If an LQMA agent process ever goes into this
"limbo" state, ecasvc utility shows
"Allocated, Pending Registration" flag if it is unthreaded LQMA
or a positive number against "Conns Pending" flag if it is a
multi threaded LQMA. If this problem occurrs multiple times,
then depending on the user configuration, it might lead to LQMA
resource problems where Execution Controller will run out of
available LQMA agents to serve application MQCONN requests.
APAR IC65966 - runmqsc <queue manager name> causes FDC on a CPU due to missing
OSS shared memory files (shm.x.x) for that CPU. Due to a yet
unknown reason, the queue manager shared memory files for a
particular CPU on the system are deleted even when the queue
manager is in running state. This prevents any new MQ connection
requests from succeeding for the same queue manager on the
affected CPU. This patch contains changes that will better
protect WMQ shared memory files and will prevent accidental
deletion of files by WMQ programs. The changes in this patch
will also report any such incidence by producing FDC files. The
FDC file produced by this detection mechanism will contain the
following information :
Comment1 :- xcsIC_QUEUE_MANAGER_POOL being destroyed.
APAR IC67569 - When WMQ Queue Server detects an error due to invalid context
data during the completion of a TMF transaction started by the
Queue Server for a PUT or GET no-syncpoint persistent message
operation, it marks the message on the queue object as
accessible. This causes the Queue Server to FDC with
"Record not found" on any subsequent MQGET operation to retrieve
the same message. The particular message on the queue that has
this problem remains in this limbo state forever and can not be
retrieved. However other messages on the same queue that do not
have this problem can be retrieved without any problem using
their msgids. The Queue Server has been revised to correct this
behavior such that detection of inconsistent context data during
MQPUT/MQGET is logged in the form of an FDC but is otherwise
committed as a normal operation. This will resolve the problem
of MQGET failing on the retrieval of the message.
APAR IC68569 - Channel server FDCs during starting/stopping of channels. The
problem occurrs due to a defect in the product where the channel
server erroneously closes its open to Queue Server but assumes
that its open to Queue Server is valid. The open handle
to Queue Server after it is closed, gets reused by another open
and hence any subsequent communication by Channel Server to
Queue Server always fails. Typically, this problem is seen when
the Channel Server experiences transient socket errors with the
channel agent (MCA) and it wants to close the socket connection.
After closing the socket connection, Channel server sends a
message to Execute Controller process to de-allocate the MCA
with which it had socket error with. It is during this
communication between the Channel Server and the Execution
Controller when the Channel Server erroneously closes the open
to the Queue Server.
APAR IC69572 - Channel server abends due to illegal address reference during
adopt MCA processing. This problem happens if the Queue Manager
has enabled adopt MCA processing to a remote WMQ Queue Manager
that does not send a remote queue manager name during channel
initialization/negotiation. The remote Queue Manager field
remains NULL and during the Adopt MCA processing logic,
the channel server incorrectly references the NULL pointer and
abends.
APAR IC69932 - SNA WMQ listener fails to start the channel when HP SPR
T9055H07^AGN is present on the NonStop system. HP, in its SPR
SPR T9055H07^AGN, changed the behavior of sendmsg() API if '-1'
is used as a file descriptor to the API. This caused
incompatibility with WMQ SNA listener process. WMQ code has now
been revised to work with the updated sendmsg() behavior.
APAR IC69996 - WMQ Queue Server generates FDC with reply error 74. When an
application with a waiting syncpoint MQGET suddenly dies
before getting a reply, the Queue Server can cause FDC
sometimes. This happens in a narrow timing window when a message
becomes available on the queue and the Queue Server starts
processing the waiting MQGET request. If the application dies
after Queue Server starts processing the waiting MQGET request,
then Queue Server detects the inherited TMF transaction error
and replies back with error 2072 (MQRC_SYNCPOINT_NOT_AVAILABLE).
However in this case, Queue Server erroneously does not delete
its internal waiter record for MQGET. When the timer pops for
the waiter record, Queue Server attempts to reply back with
no message available but the call to Guardian REPLYX procedure
fails with error 74 as the reply to the same request has already
been made(with error MQRC_SYNCPOINT_NOT_AVAILABLE). This causes
the Queue Server to FDC.
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.7:
363 - WMQ Queue Server under certain conditions fails to handle non-persistent
and persistent segmented messages at the same time.
1317 - Support for parallel execution of multiple endmqm programs on the same
Queue Manager.
1404 - MQGET WAIT is not being rejected with error 2069 when there is an
existing MQGET set signal on the same queue handle.
1434 - svcmqm utility fails when install files are open but does not tell the
user which files are open.
1566 - svcmqm does not exit immediately when 'cp' command fails to copy
binaries during fixpack installation.
1679 - WMQ Queue Server generates FDC while failing to open the message
overflow file during retrieval of very large messages (equal to or
larger than 'Message Overflow Threshold' displayed with dspmqfls
utility).
1709 - instmqm changes related to the use of /opt directory for the creation of
backup archive file.
1789 - Port of distributed APAR IY66826. Cluster sender channel does not start
and Queue Manager cache status remain in STARTING state.
1780 - Port of distributed APAR IY85542. RESET CLUSTER command does not remove
deleted repository entry.
1993 - Enhancement to WMQ mqrc program to print messages related to errors
being returned on NonStop platform.
2018 - WMQ lqma agent process leaks catalog file open.
2181 - svcmqm does not output the fixpack that is being installed.
2282 - instmqm -b archives everything under /opt directory.
2290 - Enhancement to Execution Controller process to aid development dubugging
and troubleshooting.
2534 - svcmqm has no log of its progress.
2580 - Potential SEGV in internal function call during Pathway serverclass
starting.
2632 - Incorrect message output in FDC generated by WMQ Queue Server during
nsrReply function call.
2633 - WMQ Command Server memory leak found in CLEAR QL command.
2842 - WMQ Repman process priority was not set correctly when the
qmproc.ini 'AllProcesses' stanza Priority attribute is configured.
The following serviceability fixes were made to the SDCP tool:
1971 - sdcp data was not collected correctly when there is a default
Queue Manager defined.
2211 - sdcp testing of the permission of /tmp directory.
2289 - sdcp logging of progress to a file to aid in sdcp problem diagnosis.
2298 - sdcp performance improvement.
2531 - sdcp logging of scheduled CPU information.
2582 - sdcp capture of Pathway data for PATHMON, PATHWAY, TCP, PROGRAM
and TERM attributes.
The following APAR fixes were released in V5.3.1.6:
APAR IC60204 - A Memory leak occurs with repeated DISPLAY NAMELIST command. The
leak is observed only when the NAMELIST(s) has one or more NAMES
values defined. The problem is observed from within the runmqsc
DISPLAY NAMELIST command, or from the PCF/MQIA equivalent. Tools
that request NAMELIST data via the command server with PCF/MQIA
requests such as WMQ Explorer will cause the WMQ command server
memory to grow.
APAR IC61324 - An orphan MCA problem occurs when a connection request from WMQ
LISTENER or CHANNEL SERVER process to an MCA process fails. WMQ
Execution Controller (EC) process fails to recognize the
situation and does not re-use the MCA process for future agent
allocation requests. Over a period of time, this problem can
cause the queue manager to run out of MCA resources which may
lead to a situation where no new channel can be started. The
problem is observed during heavy load conditions where the WMQ
Execution Controller(EC) process hands over a selected MCA
process to LISTENER/CHANNEL SERVER before the MCA process
becomes ready to accept connection requests.
APAR IC61551 - The use of cluster administrative command
RESET CLUSTER ACTION(FORCEREMOVE), to forcibly remove a Queue
Manager out of the cluster can cause FDCs. The problem can cause
multiple FDCs and sometimes abend in WMQ REPMAN process in a
slave role. Once the command has been issued and the error has
occurred, the concerned Queue Manager must be restarted to
restore the clustering function to normal operation. The problem
occurs because WMQ REPMAN process in a master role does not
distribute the RESET CLUSTER ACTION(FORCEREMOVE) command to the
slave REPMAN processes correctly.
APAR IC61651 - When a NAMELIST object with one or more NAMES values is defined
for a Queue Manager and a dspmqfls command is issued to retrieve
the details of either all objects under the Queue Manager or
the specific NAMELIST object, FDCs are seen from the WMQ LQMA
process. The problem exists for both unthreaded and
multi-threaded LQMA process. The problem occurs because WMQ LQMA
process does not allocate sufficient memory for NAMES buffer
during dspmqfls processing.
APAR IC61660 - A PCF RESET QUEUE STATISTICS command causes the WMQ QUEUE SERVER
process to FDC. The problem occurs during a timing window when
there is an outstanding unit of work on the queue and a
RESET QUEUE STATISTICS command is processed and then the
outstanding unit of work is backed out. RESET QUEUE STATISTICS
causes certain internal counters inside WMQ QUEUE SERVER to be
reset to zero without taking into account the outstanding unit
of work. If the outstanding work is later backed out for any
reason, the already reset counters become negative which causes
an internal consistency check to fail within the QUEUE SERVER
process and the FDC is generated.
APAR IC61681 - WMQ QUEUE SERVER causes FDC during MQCLOSE processing on a
cluster alias queue. The problem occurs during MQCLOSE
processing of a cluster QALIAS object hosted by a different
QUEUE SERVER than that hosts the target local queue. MQOPEN
incorrectly failed to allocate an internal handle for the QALIAS
object if it was a cluster alias queue. This lead to the FDC
during MQCLOSE processing.
APAR IC61846 - When a START CHANNEL command is issued after WMQ trace is
enabled, the SSL channel logs a Queue Manager error message and
fails to start.
The problem exists only for a SSL channel, no such problem
is found for regular (non-SSL) channel. The problem occurs
because of an incorrect implementation of NULL terminated string
to store SSL Cipher data.
APAR IC61920 - WMQ on NonStop reports an extra EMS message when the generation
of an FFST is reported.
The intention behind the second EMS message was to report the
case when the open of the generated FFST file fails but the
check to see the status of the opened FFST files was missing
in the code.
APAR IC62341 - A Security error from the WMQ OAM server is not propagated
correctly. It was reported as MQRC_UNKNOWN_OBJECT_NAME instead
of MQRC_NOT_AUTHORIZED. When running the Java IVP (MQIVP) to NSK
with a non-mqm user specified as the MCAUSER in the SVRCONN
channel, and if the non-mqm group isn't given authorization,
MQIVP fails to connect to the Queue Manager
with an authorization failure (MQRC_NOT_AUTHORIZED) but it is
reported incorrectly as a MQRC_UNKNOWN_OBJECT_NAME error.
APAR IC62389 - A REFRESH CLUSTER command within a cluster with more than two
repositories causes the Repository Manager to fail. The problem
is not common in normal cluster operation but is likely if
extensive administrative changes are being made to the cluster
that includes a REFRESH CLUSTER command. Incorrect distribution
of channel status information across multiple copies of the
REPOSITORY MANAGER cache in different CPUs of the system lead to
the FDC.
APAR IC62391 - Sometimes cluster queues are not visible on CPUs hosting WMQ
REPOSITORY MANAGER process in a slave role. This is a cluster
queue visibility problem across the CPUs. Any attempt to open
the cluster queue on CPUs that have the visibility problem
results in error MQRC_UNKNOWN_OBJECT_NAME or
MQRC_CLUSTER_RESOLUTION_ERROR.
The problem does not occur on the CPU that hosts WMQ REPOSITORY
MANAGER process in a master role.
APAR IC62449 - The WMQ QUEUE SERVER does not log storage related problems to
the Queue Manager log. Some of NSK storage related errors like
error 43 and error 45 are useful errors for which corrective
action can be taken to restore normal operation.
The WMQ QUEUE SERVER now logs these errors to Queue Manager log.
APAR IC62480 - Port of IZ51686. Incorrect cache object linkage causes
unexpected failures (AMQ9456) based on coincidental event
sequences.
APAR IC62511 - Port of the following clustering-related APARs from other
platforms and versions:
IZ14399 - Queue managers successfully rejoin a cluster when
APAR IY99051 is applied but have mismatching sequence
numbers for the cluster queue manager object and its
associated clusters.
IZ21977 - MQRC_OBJECT_CHANGED(2041),
AMQ9511 SYSTEM.CLUSTER.TRANSMIT.QUEUE,AMQ9448,
Repository manager ends.
IZ37511 - Generation of an FDC by the Repository Manager causes
it to terminate.
IZ14977 - Missing cluster information when Namelists are used
to add and remove queue managers from multiple
clusters at once.
IZ36482 - Changes to CLUSRCVR shared using a Namelist not
published to all clusters.
IZ10757 - Repository Manager process terminates with error
RRCI_CLUS_NO_CLUSRCVR_DEFINED.
IZ41187 - MQRC_CLUSTER_PUT_INHIBITED was returned when an out-
dated object definition from the cluster repository
was referenced.
IZ34125 - MQ fails to construct and send an availability message
when REFRESH CLUSTER REPOS (YES) is issued on a queue
manager with more than 1 CLUSRCVR.
96181 - Object changed problems with the Repository manager.
IY97159 - Repository Manager process tries to access the cache
while restoring the cache, resulting in a hang.
IZ44552 - AMQ9430 message after REFRESH CLUSTER.
135969 - Refresh bit not set when demoting QM to partial repository.
APAR IC62850 - The md.Priority of a message was not being set to the queue
DEFPRTY when a no syncpoint MQPUT is performed using
MQPRI_PRIORITY_AS_Q_DEF while there is a waiting MQGET.
A syncpoint MQPUT with a waiting MQGET does not have this
problem.
APAR IC63081 - A WMQ application abends in the MQI library when attempting to
enqueue messages to a Distribution LIST with one queue entry.
Also in certain circumstances, WMQ applications may receive
incorrect status and reason code while using Distribution LISTS.
APAR IC63105 - A Memory leak occurs in the WMQ COMMAND SERVER with repeated
DISPLAY QSTATUS command. The leak is observed only when
TYPE HANDLE is used with the above command. The problem also
occurs from within runmqsc DISPLAY QSTATUS command, or from the
PCF/MQIA equivalent. Tools that request queue status data via
WMQ COMMAND SERVER with PCF/MQIA requests such as WMQ Explorer
will cause the WMQ COMMAND SERVER memory to grow.
APAR IC63271 - When an MQ application delays in replying to the HP system OPEN
message from the WMQ QUEUE SERVER and an MQ message arrives on
queue during this period, the MQ message did not get delivered
even after the reply to the HP system OPEN message is made.
APAR IC63757 - In a standard bound application, a memory leak occurs in WMQ
LQMA process during MQCONN/MQDISC processing. The problem occurs
with both unthreaded and multi-threaded LQMA process. If the
LQMA agents are configured to have high use count
(MaxAgentUse for unthreaded LQMA and MaxThreadedAgentUse for
multi-threaded LQMA) and WMQ Execution Controller process
re-uses the same LQMA process to satisfy a application MQCONN
request, then the heap memory of LQMA process grows even if
application calls MQDISC to disconnect from the
Queue Manager.
APAR IC64297 - WMQ Queue Manager becomes non-responsive because the
WMQ QUEUE SERVER does not clean up the internal queue manager
object opens. The QUEUE SERVER internal links for
MQOPEN of the Queue Manager object were not
being released during the MQCLOSE processing. This causes a
buildup of QUEUE SERVER memory as the application repeated
MQOPEN of the queue manager object repeatedly. When either
a MQDISC occurred or the application process ended,
the QUEUE SERVER cleans up its internal lists for the process.
This resulted in a perceived QUEUE SERVER loop and
non-responsive WMQ as there were
over 246,000 opens found of the Queue Manager object with a
high mark of 548,000 within the QUEUE SERVER when a dump of the
Queue Server was analyzed.
MQOPEN of other MQ objects (qlocal, qalias, qremote etc.) do not
have this issue.
APAR IC64373 - The COBOL copybook now includes missing definitions for MQGET
SET SIGNAL processing.
APAR IC64435 - An incorrect persistence attribute was being set for a
non-persistent message on XMIT queue. When MQPUT of a
non-persistent message is done using MQPER_PERSISTENCE_AS_Q_DEF
attribute to a remote queue while the channel is idle, the MD
MD data in transmit queue header contained
MQPER_PERSISTENCE_AS_Q_DEF value of (2).
APAR IC64630 - It takes longer for unthreaded SSL sender channel to end when
communication to the remote Queue Manager is lost. The problem
occurs because the unthreaded MCA process running the SSL channel
fails to timeout correctly causing the channel to end differently
than non-SSL channels. The problem is observed only with SSL
channels and no such problem is visible with regular SSL
channels.
The following fixes discovered during IBM's development and testing work are
also released with V5.3.1.6:
1096 - An MQGET BROWSE operation can return prematurely with no message
available as well as can cause a waited GET to hang indefinitely.
1513 - amqrrmit erroneously reports multiple master REPMAN processes.
1587 - Enhancements to Execution Controller log messages pertaining to
Threshold and MaxAgent capacity situations. For unthreaded agents,
the "max unthreaded agents reached" message will now be logged when the
MaxUnthreadedAgent is allocated to perform work. In previous releases,
the message was logged when the MaxUnthreadedAgent was added to the idle
pool, which actually left one agent still available for use.
For threaded agents, messages have been enhanced to display separate
Threshold/Maximum messages for agents and threads. In previous releases,
"further connections refused" was displayed when the MaxThreadedAgent
was started, which actually left MaximumThreads connections still
available for use. In this release "further connections refused" will be
displayed when the MaximumThread is allocated for use.
Also in this release, messages will be logged when the number of agents
or threads, after having exceeded the threshold or reached maximum,
fall below the Maximum or threshold limit.
1669 - ZCG_FFST does not report the error code for a TMF error.
1675 - The Execution Controller now provides an API to mark MCAs that are no
longer used.
1670 - Correction to missing component data in a FDC generated by the
Queue Manager server(MQS-QMGRSVR00) process(amqqmsvr).
1678 - Queue server abends due to uninitialized FFST inserts.
1706 - dmpmqaut -m <qmgr> sometimes only reports the first QALIAS object and
FDCs in kpiEnumerateObjectAuthority.
1710 - The Execution Controller sometimes allocates an MCA that it has
previously asked to end
1712 - Cluster queue manager STATUS and queue visibility problems in slave
REPMAN.
1734 - Fixes/Updates to MQ tracing mechanism.
1735 - Subscription id distribution.
1751 - Fixes/Updates to MQ tracing mechanism.
1807 - Changes to improve service and debug capability of Execution Controller
started processes.
1873 - "*" subscriptions are generated incorrectly causing them to be ignored.
1989 - Fix for channel server hanging problem while opening REPMAN process.
2003 - LQMA now FDCs when a invalid message is received.
The following serviceability fixes were made to the SDCP tool:
1659 - sdcp is not capturing the PSTATE of backup processes.
1676 - sdcp takes too long.
1690 - sdcp MQ utilities are not using the correct Queue Manager name if
the name is mangled because of non-alphabetic characters.
1702 - sdcp doesn't collect all relevant Saveabend files.
1705 - sdcp workaround to avoid APAR IC61651.
1971 - sdcp now gives correct output for default Queue Manager.
The following APAR fixes were released in V5.3.1.5:
APAR IC55607 - FDC files can fail to be written, or written to the wrong file
FDCs can be suppressed or overwritten when application processes
raise FDCs under User IDs that are not members of the MQM group.
In addition, because FDC files are named with the CPU and PIN of
the generating process, and PIN is reused frequently on
HP NonStop Server, FDCs from different processes can be appended
to the same file.
The format of the file name for FDCs is:
AMQcccpppp.s.FDC
where ccc is the CPU number
pppp is the PIN
s is the sequence number
In V5.3.1.4 and earlier releases, the sequence number was
always set to 0. This fix introduces the use of the sequence
number field to ensure that FDCs from different processes are
always written to different files, and that FDCs can always be
written. FDC files are created with file
permissions "rw-r----" to prevent unauthorized access to the
FDC data.
APAR IC57435 - Attempts to end a queue manager with either -t or -p following
a cpu failure in some cases did not work as a result of
damage to the WMQ OSS shared memory files. The shared memory
management code was revised to tolerate OSS shmem/shm files
containing invalid data. Invalid data in these files is now
ignored and memory segment creation will continue normally.
APAR IC58165 - Triggered channels sometimes do not trigger when they should
Some attributes of a local queue that determine if trigger
messages get generated are not kept up to date for long-running
applications. The most critical attribute is the GET attribute
that controls whether MQGET operations are enabled for a queue
or not. If the application opened the triggered queue while
the queue was GET(DISABLED), and the queue is subsequently
modified to be GET(ENABLED), triggering will not occur when it
should.
APAR IC58377 - Trace data is not written when PIDs are reused for processes
running under different User IDs.
Trace files are named according to the CPU and PIN of the process
that is being traced. On HP NonStop Server, since PINs are
rapidly reused, it is likely that a process attempting to write
trace data will encounter an existing file written with the same
CPU and PIN. The traced process will be unable to write data if
the original file was written (and therefore owned) by a
different User ID.
This fix introduces a sequence number into the trace file names
to prevent trace file name collisions.
The format of trace file names will change from:
AMQccppppp.TRC to AMQccppppp.s.TRC
where s is a sequence number that will usually be 0.
Trace files are now created with file permissions "rw-r----"
to prevent unauthorized access to the trace data.
APAR IC58717 - The Queue Server backup process generates FDCs showing ProbeId
QS123006 from qslHandleChpPBC when attempting to locate a browse
cursor message, with the comment text of
"Error locating Last Message in Browse Cursor checkpoint in
Backup" or "Error locating Last Logical Message in Browse Cursor
checkpoint in Backup". The problem appears only when running a
number of parallel browse / get applications for the same queue
object.
APAR IC58792 - strmqm fails to delete orphaned temporary dynamic queues if the
associated touch file is missing. This results in these queues
remaining in the object catalog indefinitely, and FDC files
being generated each time the queue manager is started,
reflecting the fact that the queue could not be deleted. The
housekeeping function was modified to always silently remove
temporary dynamic queue objects from the catalog, whether or
not they are damaged. FDC files are no longer generated.
APAR IC58859 - wmqtrig script does not pass TMC with ENVRDATA correctly.
If ENVRDATA is part of the PROCESS definiton used by
runmqtrm to trigger applications the TMC is not delivered to
the application correctly. The problem does not occur with
blank ENVRDATA. Additionally, ENVRDATA or USERDATA attributes
that contain volume names ($DATA for example) are not processed
correctly by the wmqtrig script.
APAR IC58891 - Sender channels that were running in a CPU that failed are not
restarted in some circumstances. Sender channels that are not
restarted report "AMQ9604: Channel <...> terminated
unexpectedly" in the queue manager error log, and the channel
server create FDCs with ProbeID RM487001, Component
"rriChannelTerminate".
APAR IC58976 - A server channel without a specified CONNAME enters a STOPPED
state when the MCA process running the channel is forcibly
stopped or ends following a CPU failure. The channel state
should be set to INACTIVE following this type of event. To
recover the situation the channel has to be manually restarted
or stopped using MODE(INACTIVE).
APAR IC59024 - The copyright data in the COBOL COPYBOOK CMQGMOL file
is incorrect.
APAR IC59126 - Context data is missing in COA message.
When an MQPUT application sends a message with COA report
option, the generated replied COA message does not contain
context data eg. PutDate, PutTime, etc.
APAR IC59364 - Queue Server primary incorrectly commits an WMQ message in
certain cases where the backup process has failed to process
an internal checkpoint message. This causes an inconsistency
between the primary and backup processes when an MQGET is
attempted on this message, resulting in FDCs with the comment
text "Invalid Message Header context in Backup for Get" from
Component "qslHandleGetCkp". The queue object is no longer
accessible via MQGETs, but can be recovered by stopping the
backup process.
APAR IC59388 - V5.3 OAM Implementation contains migration logic which may be
triggered erroneously in some circumtances, removing authority
records from the SYSTEM.AUTH.DATA.QUEUE. This change removes the
migration logic, since there are no previous versions of the
OAM which require migration.
APAR IC59395 - Threaded LQMA actual usage is one larger than the configured
maximum use count in the qmproc.ini file. Unthreaded LQMAs
and MCAs (both threaded and unthreaded) do not suffer from this
problem.
APAR IC59428 - In some circumstances where connecting applications terminate
unexpectedly during the MQCONN processing, either by external
forcible termination, or as a consequence of other failures that
result in termination, the resulting error can cause the LQMA
process handling the application to terminate. This will
cause collateral disconnections of all other applications using
the same LQMA, with the application experiencing either a 2009
(connection broken) or 2295 (unexpected) error. The problem
window occurs only during one section of the connect protocol
and has been observed only on very busy systems with repeated
multiple forced terminations of applications.
APAR IC59742 - qmproc.ini file will fail validation if configured with both
MinIdleAgents=0 and MaxIdleAgents=0.
APAR IC59743 - Queue Manager server expiration report generation is not fully
configurable. The frequency with which the queue manager server
generates expiration reports is configurable but the number of
reports generated is not. This change introduces a new
environment variable (MQQMSMAXMSGSEXPIRE), to allow
configuration of the number of expiration reports generated
at any one time. The parameter can be added to the WMQ
Pathway MQS-QMGRSVR00 serverclass:
ALTER MQS-QMGRSVR00, ENV MQQMSMAXMSGSEXPIRE=<1-99999>
If this value is not specified in the queue manager
serverclass configuration, the value defaults to 100.
APAR IC59802 - Memory leak occurs with repeated DIS CHSTATUS SAVED command.
A memory leak exists in the Channel Saved Status query. This
leak is observed within either the runqmsc DISPLAY CHSTATUS
SAVED command, or the PCF/MQIA equivalent. Tools that request
saved channel status data via the Command Server with PCF / MQIA
requests such as WMQ Explorer will cause the Command Server
memory to grow.
APAR IC60114 - WMQ processes or user application processes generate FDCs
referring to "shmget" following forcible termination of the
process or failure of the CPU running it. This is a result of
the Guardian C-files (Cnxxxxxx) for a CPU becoming corrupt
during an update operation, rendering the file and associated
shared memory segment unusable. C-file update operations are
now performed atomically to prevent this problem.
APAR IC60135 - Improve servicability of the "endmqm -i" command to prevent the
command from waiting indefinitely for the queue manager to end.
Following this change after a specified number of seconds, the
command will complete with the message "Queue Manager Running"
and return to the command line with exit status 5.
APAR IC60175 - Description is not available (security/integrity exposure)
APAR IC60361 - Memory leak occurs in SVRCONN channel MCAs which repeatedly open
local queue objects.
APAR IC60455 - WMQ Broker restart may not work correctly.
If the WMQ Broker is restarted using strmqbrk/endmqbrk,
subsequent attempts to restart the broker may fail, and 2033
errors my arise when running the test broker samples and
recycling the broker processes.
APAR IC60119 - System Administration manual incorrectly states the default
value of the TCP/IP Keepalive is "ON"
The following fixes discovered during IBM's development and testing work were
released with V5.3.1.5:
1403 Erroneous SVRCONN channel ended message.
SVRConn channels should not generate "Channel Ended" messages in
the error log, but in some circumstances, threaded svrconn
channels do generate these messages.
1451 Internal changes relating to trace and FDC files sequence numbers
1453 Problem with MQCONN after restart of broker
1516 strmqm fails with invalid ExecutablePath attribute (qmproc.ini)
1560 Port of V51 MQSeries for Compaq NonStop Kernel APAR IC57981.
Backup Queue Server runs out of memory processing non-persistent
messages in 27K range.
1564 runmqlsr abends in nssclose after a previous 'socket' calls fails
1570 Added Agent type to EC logged threshold and max agent messages.
1576 Change ECA interface to V4
1577 Queue Server message expiration deletion phase log message
1583 Blank channel status entries can get created triggering
channels when AdoptMCA is enabled.
Under certain timing situations, when triggered channels are
used and AdoptMCA is enabled for the queue manager, blank
channel status entries can be created with the JOBNAME
referencing the Channel Initiator (runmqchi), for example:
AMQ8417: Display Channel Status details.
CHANNEL() XMITQ()
CONNAME() CURRENT
CHLTYPE() STATUS(BINDING)
MSGS() BATCHES()
JOBNAME(5,333 $MQCHI OSS(318898190)) RQMNAME()
This problem does not cause any immediate functional problem,
however the blank entries consume channel status table entries
and therefore could prevent legitimate channel starts in the
event that the status table becomes full.
1594 C++ unthreaded libraries use threaded semaphores
1596 Improved cs error reporting
1597 EC started processes sometimes not started in intended CPU
1598 NSS Incorrect component identifiers used in some parts of zig
1608 Queue status errors on failure of a no syncpoint persist message
put or get
1601 Tracing details to the EC to augment the Entry and Exit trace
calls
1611 LQMA Queue manager attribute corruption
1613 Enhanced LEC Failure Handling
1615 The EC may allow the os to choose in which cpu a MCA will start
1616 Channel server comp traps have potential performance impact.
1621 MQCONN does not report valid reason code when agent pool is full
1622 After a channel is started dis chs displays "binding" in some
circumstances when it should display "running"
1623 Incorrect message when MCA allocation fails
1626 Addition of Service information collection tool (SDCP)
New platform support was released in V5.3.1.4:
Fixpack V5.3.1.4 introduced support for the HP Integrity NonStop BladeSystem
platform, NB50000c. Use the H-Series (Integrity) package of WebSphere MQ
for execution on the BladeSystem. Please refer to the Hardware and Software
Requirements section for details about the levels of the J-Series software
required.
The following APAR fixes were released in V5.3.1.4:
APAR IC57020 - runmqtrm does not function correctly and produces errors in some
cases.
When a triggered application is a guardian script file
(ie filecode 101). runmqtrm produces an "illegal program
file format" error. Triggering also does not work correctly
for COBOL or TAL applications.
APAR IC57231 - The execution controller starts repository processes at the
same priority as itself in some cases, and does not take
account of the values set in the qmproc.ini file.
APAR IC57420 - Repository manager restart following failure causes cluster
cache corruption in some circumstances.
If a repository manager abends while a queue manager is under a
heavy load of cluster-intensive operations, in some
circumstances the repository manager that is restarted can
damage the cluster cache in the CPU in which it
is running. This can prevent further cluster operations in that
CPU and cause WMQ processes to loop indefinitely. This release
changes the repository startup to prevent this from happening.
APAR IC57432 - OSS applications that attempt to perform MQI operations from
forked processes encounter errors.
If an oss WMQ application forks a child process, that child
process will encounter errors if it attempts to perform MQI
operations. Some operations may succeed, but will result
in the generation of FDC files.
APAR IC57488 - MQMC channel menu display display error after channel is
deleted.
If a channel is deleted while the channel menu in MQMC
is displayed, refreshing the channel menu produces the
error: "Unknown error received from server. Error number
returned is 1" and will not correctly display the channel
list without restarting MQMC.
APAR IC57501 - unthreaded sender channels to remote destinations with
significant network latency may fail to start with timeout
errors.
APAR IC57524 - Applications launched locally from remote nodes cannot access
some of the queue manager shared memory files due to default
security on those files.
APAR IC57627 - Handling of TMF outages to improve operational predictability.
If TMF disables the ability to begin new transactions
(BEGINTRANS DISABLED), WMQ does not always react in a
predictable or easily diagnosed manner, and applications can
suffer a variety of different symptoms. If TMF is stopped
abruptly (STOP TMF, ABRUPT) queue managers can become unstable
and require significant manual intervention to stop and restart.
Refer to item 18 in "Known Limitations, Problems and
Workarounds" later in this README for more information.
APAR IC57712 - altmqfls --qsize with more than 100 messages on queue fails.
When a altmqfls --qsize is performed with more the 100 MQ
messages in the queue the processing fails.
APAR IC57719 - FDCs from MQOPEN when an error exists in alias queue manager
resolution path. If a queue resolution path includes a queue
manager alias, and the target of the alias does not exist,
this will produce an FDC, rather than just failing the
MQOPEN as would be expected.
APAR IC57744 - CPU goes busy when stopping a threaded SSL receiver channel
using MODE(TERMINATE).
If a stop channel mode(terminate) is used to stop an SSL
receiver channel that is running in a threaded MCA, the CPU
where the MCA is running in begin using large amounts of CPU
time (95% range). This is due to a problem in the threads
library.
APAR IC57876 - Very infrequently, messages put via threaded LQMAs can in some
circumstances contain erroneous CCSID information. This has
been observed to cause conversion errors if the message is
destined for a channel that has the CONVERT(YES) set.
Unthreaded LQMAs do not suffer from this problem.
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.4:
993 - Due to the way that default file security was used, file security for
certain shared memory files used by the queue manager (SZ***) may
inadvertently change in a way that prevents applications not in
the mqm group from issuing MQCONN. File permissions were rationalised
in this release to reflect those used for other shared memory files.
1458 - Resolve Channel command generates FFSTs.
When resolving In-Doubt channels, FFSTs were generated by the Channel
Server and the MCA. Although the channels were successfully resolved,
the In-Doubt status in a DIS CHS query was not correctly updated.
When resolving In-Doubt channels using the COMMIT option the following
error message was displayed "AMQ8101: WebSphere MQ error (7E0) has
occurred."
1493 - The validation of the qmproc.ini file does not report the error case
where multiple ChannelsNameMatch entries are specified ChlRule1.
1498 - Instmqm does not support installation of the product on
Integrity NonStop BladeSystem platforms.
1507 - Some Execution controller messages were missing "Action" descriptions
when reported in the error log.
1517 - In the qmproc.ini file, the AppRule4-SubvolMatch argument was not
working
1522 - Communications Component Function ids and probes are incorrect. This
resulted in misleading or missing information in trace files generated
for support purposes.
1546 - MQBACK operation incorrectly reports error during broker operations
1549 - Channel Server doesn't shutdown after takeover.
If the Primary Channel Server process is prematurely ended, for
example by a CPU crash, the Backup Channel Server process becomes the
new Primary process. Subsequent attempts to use endmqm will hang
because the new Primary Channel Server process will not end.
The following documentation APARs are addressed by the V5.3.1.4 readme:
APAR IC55404 - REFRESH QMGR PCF command is not documented in the Programmable
command formats manual.
Also - please check the "Limitations" and "Known Problems and Workarounds"
sections later on in this readme for updates.
The following APAR fixes were released in V5.3.1.3:
APAR IC54305 - The HP TNS (non-native) C compiler generates Warning 86 when
compiling MQI applications
APAR IC55501 - The altmqfls command does not return the correct completion
status; it always returns success
APAR IC55719 - Non-native MQINQ binding does not deal with some null pointer
parameters correctly
APAR IC55977 - Channel retry does not obey SHORTTMR interval accurately enough
APAR IC55990 - Trigger data changes not being acted upon if they were made
while the queue was open, leading to incorrect triggering
behavior
APAR IC56277 - Command Server can loop with INQUIRE QS command with a single
parameter
APAR IC56278 - A remote RUNMQSC DIS QS command always times out
APAR IC56309 - MCAs do not disconnect from some shared memory when ending,
which causes a slow memory leak, and under some conditions an
abend
APAR IC56458 - Channel Server loops after installing V5.3.1.2 due to corrupt
data on SYSTEM.CHANNEL.SYNCQ
APAR IC56493 - Cannot use "qualified" hometerm device names with V5.3.1.2
APAR IC56503 - Channel Server and MCA can deadlock after repeated STOP CHANNEL
MODE(FORCE) or MODE(TERMINATE) commands
APAR IC56536 - Unthreaded responder channels don't de-register from the EC when
an error occurs during or before channel negotiation. For
example, bad initial data will cause this. Unthreaded MCAs
build up and eventually reach the maximum which prevents further
channel starts
APAR IC56681 - C++ unthreaded Tandem-Float SRLs have undefined references
APAR IC56834 - endmqm -p can sometimes leave MCA processes running
The following fixes discovered during IBM's development and testing work were
released with V5.3.1.3:
663 - Guardian command line utility return status is not the same as the OSS
utilities return status
1402 - Add additional tracing when testing for inconsistencies in processing
a channel start in the Channel Server
1416 - Ensure that the Channel Server can support the maximum BATCHSZ of 2000
1446 - Pub / Sub command line utilities do not behave well if no broker has run
since the queue manager was created
1470 - EC abends attempting to start a non-executable REPMAN
1474 - Pub / Sub broker process handling corrections for the EC
1476 - The EC checkpoints the number of threads running in agents incorrectly
1477 - Enhancement to ecasvc utility: the creation date/time of LQMAs, MCAs,
and REPMEN are now displayed
1487 - Enhancement to ecasvc utility: changed the display of Agent attributes
to use the "real" qmproc.ini attribute names. Added a new option,
that displays information about all connected applications
1494 - A small memory leak occurs for the delete channel operation
1508 - Multiple qmproc.ini environment variables don't get propagated to
agents or repmen
1509 - The EC failed to stop an MCA that was hung when a preemptive shutdown
was initiated
The following documentation APARs were addressed by the V5.3.1.3 readme:
APAR IC55380 - Transport provider supplied during install is not propagated to
Pathway configuration by crtmqm. Please see the documentation
update below made for Page 17 of the "Quick Beginnings" book.
The following APAR fixes were released in V5.3.1.2:
APAR IC52123 - LQMA abend handling rollback of a TMF transaction in MQSET
APAR IC52963 - The PATHMON process is not using configured home terminal for
MQV5.3 on HP Nonstop Server
APAR IC53205 - FDC from Pathway runmqlsr when STOP MQS-TCPLIS00
APAR IC53891 - There is a memory leak in the Channel Server when processing
the DIS CHS command
APAR IC53996 - C++ PIC Guardian DLLs missing.
APAR IC54027 - MQRC_CONTEXT_HANDLE_ERROR RC2097 when loading messages using
MO71
APAR IC54133 - Multi-threaded LQMA should not try to execute Unthreaded
functions if qmproc.ini LQMA stanza sets MaximumThreads=1
APAR IC54195 - runmqtrm data for Trigger of Guardian application not
reinitialized
APAR IC54266 - MinThreadedAgents greater than PreferedThreadedAgents causes
MQRC 2009 error
APAR IC54488 - MCA's abend after MQCONN/MQDISC 64 times.
APAR IC54512 - OSS runmqsc loops if Guardian runmqsc is TACL stopped
APAR IC54517 - upgmqm does not handle CPUs attribute for PROCESS specifiction
in a SERVERCLASS
APAR IC54583 - SSL channel agent can loop if an SSL write results in a socket
I/O error
APAR IC54594 - EC abends with non-MQM user running application from non-MQM
directory
APAR IC54657 - Channel stuck in BINDING state following failed channel start
due to unsupported CCSID.
APAR IC54666 - Queue Server deadlock in presence of system aborted
transactions.
APAR IC54798 - upgmqm fails with Pathway error on 3 or more status servers that
require migration from V5.1 queue manager.
APAR IC54841 - When a temporary dynamic queue is open during "endmqm -i"
processing an FDC is generated
APAR IC55008 - Added processing that will cause Channel Sync data to be
Hardened at Batch End
APAR IC55073 - altmqfls --qsoptions NONE is not working as specified
APAR IC55176 - Abend in MQCONN from app that is not authorized to connect (2035)
or with invalid Guardian Subvolume file permissions
APAR IC55500 - QS Deadlock with Subtype 30 application using MQGMO_SET_SIGNAL
APAR IC55726 - Channel stuck in BINDING state following failed channel start
due to older FAP level
APAR IC55865 - Abend on file-system error writing to EMS collector
The following fixes discovered during IBM's development and testing work were
also released with V5.3.1.2:
1122 - Invalid/incomplete FFST generated during MQCONN when Guardian
subvolume cannot be written to.
1392 - Add support for Danish CCSID 65024
1397 - Command Server fails to start and EC reports failure to initialize
a CPU - error 12 purging shared memory files.
1409 - Guardian WMQ command fails when invoked using Guardian system() API
1413 - MCA looping after SSL socket operation fails
1419 - altmqfls --volume attempted using a open object causes FDCs
1439 - On non-retryable channel, runmqsc abends while executing RESOLVE CHANNEL
command
The following documentation APARs were addressed by V5.3.1.2:
APAR IC53996 - C++ PIC Guardian DLLs missing.
originally released in V5.3.1.1 Patch 1:
APAR IC53891 - There is a memory leak in the Channel Server when
processing the DIS CHS command
originally released in V5.3.1.1 Patch 2:
APAR IC54583 - SSL channel agent loops
originally released in V5.3.1.1 Patch 3:
APAR IC54666 - Queue Server deadlock in presence of system aborted
transactions.
originally released in V5.3.1.1 Patch 4:
APAR IC54512 - OSS runmqsc loops if Guardian runmqsc is TACL stopped
The following APAR fixes were released in V5.3.1.1:
APAR IC52737 - When in SSL server mode and the sender is on zOS a list of CAs
that the server will accept must be sent to the zOS sender
during the SSL handshake
APAR IC52789 - upgmqm support for upgrading V5.1 queue managers that do not use
OAM (created with MQSNOAUT defined). Also add diagnostics as to
reasons and preventative actions for failure to create a PATHMON
process
APAR IC52919 - Problems in synchronization of starting a queue manager when
multiple Queue Servers are defined
APAR IC52942 - Trigger Monitor holds Dead Letter Queue open all the time
APAR IC53240 - Correct sample API exit to build for PIC and SRL/Static
APAR IC53243 - Start of many applications simultaneously causes LQMA FDC
APAR IC53248 - Kernel not informing repository cache manager of updates to
cluster object attributes
APAR IC53250 - Flood of FDCs when trace is enabled before qmgr start
APAR IC53254 - Browse cursor mis-management left locked message on queue.
In addition, browse cursor management was not correct in the
event that a syncpoint MQGET rolls back
APAR IC53288 - Cluster Sender channel is not ending until the HBINT expired
APAR IC53383 - upgmqm was losing the MCAUSER attribute on channels
APAR IC53492 - TNS applications fail in MQPUT with more than 106920 bytes of
data
APAR IC53524 - SVRCONN channels are not ending after STOP CHANNEL if client
application is in a waited MQGET
APAR IC53552 - OAM uses getgrent() unnecessarily, causing slow queue manager
startup
APAR IC53652 - Guardian administration commands don't work with VHS or other
processes as standard input or output streams
APAR IC53728 - ECONNRESET error when Primary TCP/IP process switched should
not cause listener to end
APAR IC53835 - Assert in xtrInitialize trying to access trace shared memory
The following documentation APARs were addressed by V5.3.1.1:
APAR IC51425 - Improve documentation of crtmqm options
APAR IC52602 - Document ClientIdle option
APAR IC52886 - Document RDF setup ALLSYMLINKS
APAR IC53341 - Document OpenTMF RMOPENPERCPU / BRANCHESPERRM calculation
The following fixes discovered during IBM's development and testing
work were also released with V5.3.1.1:
634 - Correct function of altmqfls option to reset measure counter
822 - Message segmentation with attempted rollback operation failed
862 - PCF command for Start Listener fails
903 - Channel status update problems during shutdown after Repman has ended
922 - Channel status incorrect when attempting to start a channel and the
process management rules prevent the MCA thread or process from starting
929 - Incorrect response records when putting to distribution list
1012 - Two of the sample cobol programs give compilation error
1059 - C-language samples use _TANDEM_SOURCE rather than __TANDEM
1064 - errors checkpointing large syncpoint MQPUT and MQGET operations
when transactions abort
1069 - Not able to delete CLNTCONN channels
1108 - Error logged when MCA exits because maximum reuse count is reached
1152 - strmqm -c option gives unexpected error if executed after strmqm
1176 - Sample cluster workload exit not functioning correctly
1177 - QS backup abend on takeover with local TMF transactions
1180 - Segmentation of messages on transmission queues by the queue manager
was incorrect.
1182 - Replace fault tolerant process pair FDCs with log messages for better
operator information when a takeover occurs
1185 - Opens left in all three NonStop Server processes after divering CPUs
1208 - Trace info is incorrect for zslHPNSS functions. FFSTs show incorrect
component and incorrect stack trace info
1210 - FFSTs generated by criAccessStatusEntry when starting channel with
same name from another queue manager
1213 - Pathway listener generates FDCs on open of standard files
1229 - Permanent dynamic queues being marked as logically deleted on last close
1240 - Channel Server needs to update status for unexpected thread close
1244 - Speed up instmqm
1246 - implement workaround for the regression in the OSS cp command introduced
in G06.29/H06.06 where a Format 2 file is created by default when
copying to Guardian
1247 - Fixes to SSL CRL processing, added CRL reason to message amq9633
1253 - SSL samples required updating to reflect enhanced certificate file
organization - cert.pem and trust.pem
1254 - Fix an MQDISC internal housekeeping problem
1256 - MCA does not exit after re-use count if an error occurs during early
initialization
1260 - Speed up strmqm when performed on very busy systems with large number
of CPUs by minimizing calls to HP FILE_GETOPENINFO_ API
1264 - Correct the handling of the option to make Message Overflow files
audited in QS
1266 - Improve diagnostic information of FFST text for semaphore problem
1271 - After sequence of 2 CPU downs, EC, QS and CS still have openers
1272 - Improve protection in svcmqm when files in the installation are open
1273 - Memory leak in the command server caused by unreleased object lists
1277 - Don't FFST if initialization fails because the mqs.ini file doesn't
exist
1281 - LQMA thread doesn't end when CPU containing application goes down
1288 - Channels not retrying after CPU failure that also causes takeover of CS
1290 - MQDISC when connection broken doesn't tidy up transaction
1291 - Correct the syncpoint usage when amqsblst is used as a server. Enhance
amqsblst for fault tolerant behavior - makes amqsblst attempt to
reconnect and reopen objects on 2009 errors so it can be used during
fault tolerant testing
1294 - Application process management rules don't always work correctly
1297 - Correct file permission of trace directory and files - changed
permission of trace directory to 777
1301 - No queue manager start or stop events generated
1302 - instmqm function get_Guardian_version should look for string
"Software release ID"
1306 - instmqm validation fails when Java is not installed - issue a warning
if the java directory doesn't exist and continue the installation
1310 - OSS Serverclasses not restarting in Pathway if they end prematurely
1313 - EC process management can exceed maximum number of threads for LQMA
1317 - REFRESH CLUSTER command with REPOS(yes) fails
1319 - MQPUT and MQPUT1 modifying PMO.Options when creating a new MsgId
1324 - MQPUT returned MQRC_BACKED_OUT when putting message that required
segmentation to local queue
1325 - Trace state doesn't change in servers unless process is restarted
1340 - QS error handling MQPUT checkpoint. Also can lead to zombie messages on
queue requiring queue manager restart to clear
1341 - MQGET not searching correctly in LOGICAL_ORDER for mid-group messages
1346 - EC initial memory use too high. Initial allocation was approximately
18 megabytes
1351 - Upgrade logging format to V6.x style
1353 - MQGET of 210kbyte NPM from queue with checkpointing disabled caused
message data corruption at offset 106,906
1355 - xcsExecProgram sets current working directory to /tmp - changed to
installation errors directory
1357 - instmqm fails to create an OSS symbolic link after a cancelled install
1362 - MsgFlags showing segmentation status should still be returned in
MQGET even if applications specifies MQGMO_COMPLETE_MSG
1364 - endmqlsr sometimes hangs
1366 - Correct trace, FDC and mqver versioning information for V5.3.1.1
All fixes that were previously released in V5.3.0 and V5.3.1 are also included
in this release. For information on fixes prior to V5.3.1.1, please refer to
the readme for V5.3.1.3 or earlier.
Backward compatibility
----------------------
IBM WebSphere MQ V5.3.1 for HP NonStop Server is interoperable over channels
with IBM MQSeries(TM) V5.1 for Compaq NSK, as well as any other current or
earlier version of IBM MQSeries or IBM WebSphere MQ on any other platform.
Product compatibility
---------------------
IBM WebSphere MQ V5.3.1 for HP NonStop Server is not compatible
with IBM WebSphere MQ Integrator Broker for HP NonStop Server.
For other compatibility considerations, review the list of suitable
products in the WebSphere MQ for HP NonStop Server Quick Beginnings book.
IBM WebSphere MQ V5.3.1 for HP NonStop Server is compatible with any
currently supported level of IBM WebSphere MQ Client. IBM WebSphere MQ V5.3.1
for HP NonStop Server does not support connections from WebSphere MQ
Extended Transactional Client.
INSTALLATION, MIGRATION, UPGRADE AND CONFIGURATION INFORMATION
==============================================================
Hardware and Software Requirements
----------------------------------
The list of mandatory HP operating system and SPR levels has changed
since the V5.3.1.1 release. Please read the following information carefully,
and if you have any questions, please contact IBM.
For the HP Integrity NonStop Server H-Series systems, the following system
software versions are the minimum mandatory level for V5.3.1.17:
- H06.23.01 or later
- SPR T8306H01^ABJ or later
- SPR T8994H01^AAM or later
- SPR T8397H01^ABD or later
- SPR T1248H06^AAX or later
For the HP Integrity NonStop BladeSystem J-Series systems, the following
system software versions are the minimum mandatory level for V5.3.1.17:
- J06.14.00 or later
Note that V5.3.1.17 is not supported on G-Series systems
Recommended SPRs
-----------------
It has become increasingly complicated to document fixes made by HP
for some of their products, as the products themselves often have multiple
threads (H01, H02, G01, G06 etc..) that can be used on multiple OS levels.
To make it more convenient for our customers to determine whether they already
have a recommended fix installed, or to find the appropriate fix in Scout on
the NonStop eServices Portal, we are now referencing particular problems by
their HP solution number.
If you wish to determine whether your particular level of an SPR contains
the solution, review the document included when you downloaded the product
from Scout or review the softdocs in Scout for the solution number for that
product.
We have added more information about the specific problems reported,
what the symptoms are, workarounds to these problems if relevant,
and the likelihood of it happening.
Please note:
Where versions are inside parentheses beside an HP solution #,
those versions only are affected by that particular solution.
Product ID: T0845 - Pathway Domain Mgmt Interface/TS/MP 2.5
Problem: PATHCTL file can be corrupted if a pathway serverclass abends.
Symptom: Pathway unusable
(Reported as possible by HP but not independently confirmed by IBM).
HP Solution: 10-120322-2183
Likelihood: Possible
Workaround: None
Recovery: May be possible to re-configure the WMQ serverclasses in some circumstances.
Product ID: T1248 - pthreads
Problem: Threaded server application, MCA - amqrmppa_r, causes 100% CPU
utilization
HP Solution: 10-080818-5258 (H07, H06, G07)
Symptom: CPU 100% busy while processing SSL channels. MCA process consumes
all available CPU. May be communication errors on channels
Likelihood: Certain when attempting to Stop SSL Channels using Mode Terminate
if priority of MCA process is higher than Channel Servers
Workaround: For SSL channels use unthreaded MCAs or upgrade to WMQv5.3.1.4
Recovery: None, CPUs will go back to normal after about 5 mins
Problem: Assert in spt_defaultcallback for threaded MCAs, amqrmppa_r
HP Solution: 10-080519-3266 (H06, G07)
Symptom: FDCs from MCAs, plus MCAs abend (qmgr log message), channels fail
and restart. Error 28 seen in FFST from MCA process on WRITEREADX
call
Likelihood: Rare.
Workaround: Use unthreaded MCAs
Recovery: None, MCAs will abend but MCAs and associated channels will
restart
Product ID: T6533 - STDSEC-STANDARD SECURITY PROD
Problem: The GROUP_GETINFO_ Guardian procedure call used by dspmqusr returns error 590 with
group ID greater than 65535.
Symptom: dspmqusr abends with AMQ7047: An unexpected error was encountered by a command.
FFST generated reports error 590.
HP Solution: SOLN 10-091111-6280
Likelihood: Definite if a user is created in group with a group ID greater than 65535,
eg. SECURITY-ENCRYPTION-ADMIN and that user id added as a principal.
Workaround: Members of a group with group ID greater than 65535 cannot be added as WMQ principal
Recovery: None
Product ID: T8306 - OSS Sockets Version: H04, H02, G12, G10
Problem: OSS socket APIs fail with ENOMEM (4012) error.
HP Solution: 10-081205-7769. (H04, G12)
Symptoms: Channels fail to start, Error log and FFSTs indicate error 4012.
Likelihood: Rare
Workaround: None
Recovery: Reload CPU.
Problem: CPU halt %3031 and CPU Freeze %061043 after CPU down testing.
Symptoms: All processes in the CPU will end,,, backup NonStop processes
will take over. Error log will indicate backup servers have taken
over.
HP Solution: 10-080827-5452. (H04, H02, G12, G10)
Likelihood: Rare
Workaround: None
Recovery: Reload CPU
Product ID: T8397- OSS Socket Transport Agent
Problem CPU Halt %3031 or CPU Freeze %061043
Symptoms All processes in the CPU will end, backup NonStop processes will
take over. Error log will indicate backup servers have taken over.
HP Solution: 10-080827-5452 (H02, H01, G11)
Likelihood: Rare
Workaround: Reload CPU
Problem: OSS socket APIs fail with ENOMEM (4012) error.
Symptom: Channels fail to start. Error log and FFSTs indicate error
4012.
HP Solution: 10-081205-7769 (H02, G11)
Likelihood: Rare
Workaround: None
Recovery: Reload CPU
Product ID: T8607 - TMF
Problem: Multiple issues involving lost signals with OpenTMF
Symptom: Channel Server indicates Sequence number mismatches. Channel
server generates FDCs that report file system error 723.
HP Solution: 10-081027-6812, Hotstuff HS02990. (H01)
Likelihood: Rare, but may occur if audit trail is 90% full, or operator stops
TMF.
Workaround: Monitor audit trail size
Recovery: Stop primary channel server process.
Symptom: Queue Manager completely freezes up, Log messages every 10 seconds
for 50 attempts.
HP Solution: 10-081027-6812, Hotstuff HS02990. (H01)
Likeihood: Very likely if STOP TMF, ABRUPT command is issued while Queue
Managers are running.
Workaround: Do not issue STOP TMF, ABRUPT command while queue managers are
running until the SPR has been installed.
Recovery: Restart Queue Manager.
Product ID: T8620 - OSS file system Version: G13,H03, H04
Problem: lseek() fails with errno 4602.
Symptoms: FFSTs generated in xcsDisplayMessage component
HP Solution: SOLN 10-071012-8159 (G13,H03, H04)
Likelihood: Likely
Workaround: Turn off OSS Caching in all disks
Recovery: None needed, problem is benign.
Symptoms: Queue Manager slowdown along with (sometimes) lost log messages
in busy queue managers. System suffered major OSS lockups and CPU
halts
HP Solution: SOLN 10-071012-8159 (G13,H03, H04)
Likelihood: Rarely
Workaround: None
Recovery: Stop and restart all Queue Managers and Listeners. Reload CPUs.
If you use SNA channels with V5.3.1, we recommend the latest levels of the
HP SNAX or ACI Communication Services for Advanced Networking (ICE) be used
for the SNA transport. The following versions were verified by IBM with this
release of WMQ:
ACI Communication Services for Advanced Networking (ICE)
- v4r1 on both HP Integrity NonStop Server and S-Series systems
HP SNAX
- T9096H01 on HP Integrity NonStop Server (H-Series) systems
If you use the WebSphere MQ V5.3 classes for Java and JMS for HP NonStop Server
you will need to install HP NonStop Server for Java Version 1.4.2 or later.
The Java.pdf supplemental document in the <install_path>/opt/mqm/READMES/en_US
directory has been updated in this release. Java/JMS users should review the
updated document.
Upgrading to V5.3.1.17
---------------------
For systems running H Series operating systems, you may upgrade
any prior service level of WebSphere MQ V5.3.1.x for HP NonStop Server to
V5.3.1.17 level using this release. For NonStop BladeSystem running J series
operating systems, you may upgrade from version V5.3.1.4 only, since V5.3.1.4
is the earliest supported version on J series. If you need to perform a full
installation on a J series system from the original installation media, see the
section later in this readme file for instructions.
The installation tool, svcmqm, is used to upgrade existing installations
to this level. Additionally, the placed files for any prior level of V5.3.1
can be overlaid with the new files from V5.3.1.17 and then instmqm can be used
to create new installations at the updated V5.3.1.16 level.
You must end all queue managers and applications in an installation if you
want to upgrade that installation to V5.3.1.17.
You do not need to re-create any queue managers to upgrade to V5.3.1.17.
Existing queue managers (at any V5.3.1.x service level) will work with
V5.3.1.17 once an installation has been properly upgraded.
If you use SSL channels, and are upgrading from WMQ V5.3.1, you must perform
a small reconfiguration of the Certificate store before running any SSL
channels after you have upgraded. The steps that are required are described
below in the Post-Installation section. If you do not perform this
reconfiguration, SSL channels in the upgraded V5.3.1.17 installation will
fail with the log messages similar to the following:
For sender channels:
09/29/07 08:52:43 Process(0,483 $Z8206) User(MQM.ABAKASH) Program(amqrmppa)
AMQ9621: Error on call to SSL function ignored on channel
'ALICE_BOB_SDRC_0000'.
EXPLANATION:
An error indicating a software problem was returned from a function which is
used to provide SSL support. The error code returned was '0xB084002'. The error
was reported by openssl module: SSL_CTX_load_verify_locations, with reason:
system lib. The channel is 'ALICE_BOB_SDRC_0000'; in some cases its name cannot
be determined and so is shown as '????'. This error occurred during channel
shutdown and may not be sufficiently serious as to interrupt future channel
operation; Check the condition of the channel.
ACTION:
If it is determined that Channel operation has been impacted, collect the items
listed in the 'Problem determination' section of the System Administration
manual and contact your IBM support center.
---- amqccisx.c : 1411 ------------------------------------------------------
09/29/07 08:52:44 Process(0,483 $Z8206) User(MQM.ABAKASH) Program(amqrmppa)
AMQ9001: Channel 'ALICE_BOB_SDRC_0000' ended normally.
EXPLANATION:
Channel 'ALICE_BOB_SDRC_0000' ended normally.
ACTION:
None.
For client or receiver channels:
-------------------------------------------------------------------------------
09/29/07 08:05:28 Process(1,802 3 $X0545) User(MQM.HEMA) Program(amqrmppa_r)
AMQ9620: Internal error on call to SSL function on channel '????'.
EXPLANATION:
An error indicating a software problem was returned from an function which is
used to provide SSL support. The error code returned was '0x0'. The error was
reported by openssl module: SSL_load_client_CA_file, with reason: CAlist not
found. The channel is '????'; in some cases its name cannot be determined and
so is shown as '????'. The channel did not start.
ACTION:
Collect the items listed in the 'Problem determination' section of the System
Administration manual and contact your IBM support center.
---- amqccisx.c : 1347 ------------------------------------------------------
09/29/07 08:05:28 Process(1,802 3 $X0545) User(MQM.HEMA) Program(amqrmppa_r)
AMQ9228: The TCP/IP responder program could not be started.
EXPLANATION:
An attempt was made to start an instance of the responder program, but the
program was rejected.
ACTION:
The failure could be because either the subsystem has not been started (in this
case you should start the subsystem), or there are too many programs waiting
(in this case you should try to start the responder program later). The reason
code was 0.
WMQ Application re-compile considerations :
You do not need to re-compile any applications to upgrade to V5.3.1.17.
WMQ Application linkage considerations :
a) If upgrading from V5.3.1.5 or later releases (including patch releases) :
Existing applications will continue to work with V5.3.1.17 release. However,
IBM strongly recommends that if you are upgrading from a release prior to
5.3.1.7, you review the impact of APARs IC67057 and IC68569 that were fixed in
the V5.3.1.7 release. Please also note internal defect 4158 that is fixed in
V5.3.1.11. The IC67057, IC68569 and internal defect 4158 fixes will not be
effective in non-native applications unless the applications are relinked
using the HP BIND utility.
b) If upgrading from V5.3.1.4 or earlier releases (including patch releases) :
You MUST use the HP BIND utility to relink any non-native applications prior
to using them with V5.3.1.16. If an application is not re-bound with the
V5.3.1.17 WMQ product, MQCONN API calls will fail with a MQRC 2059 and the
WMQ EC process will output an FDC when the MQI incompatibility is detected,
as follows:
Probe Id :- EC075003
Component :- ecaIsECup
Comment1 :- Application MQ API not compatible, relink application
Comment2 :- <process cpu,pid process name>
Comment3 :- <application executable name>
Installation from Electronic Software Download H or J Series based systems
-----------------------------------------------------------------------------
These instructions apply to installing WebSphere MQ for HP NonStop Server,
Version 5.3.1.17, from the package downloaded from IBM. Please note the
additional restrictions for upgrading J Series systems to this version.
Use svcmqm to update an existing installation from the V5.3.1.17 placed files.
1. Unzip the fixpack distribution package - wmq53.1.17.tar.zip.
The fixpack distribution package contains the following files:
readme_wmq_5.3.1.17 - this README
wmq53.1.17_H07.tar.Z - H-Series H06 Package
2. Identify the correct fixpack package to install:
For H-Series (H06) or J-Series systems (J06) use: wmq53.1.16_H06.tar.Z
3. Upload the compressed fixpack archive to the OSS file system in binary mode.
You may wish to store the compressed archive and the expanded contents
in a location where you archive software distributions from IBM.
If you do this, you should store the compressed archive in a directory
that identifies the version of the software it contains,
for example, "V53117".
mkdir -p /usr/ibm/wmq/V53117
upload (in binary mode) the correct compressed tarfile to this directory
4. Extract the fixpack compressed tarfile using commands similar to:
cd /usr/ibm/wmq/V53117
uncompress wmq53.1.17_H06.tar.Z
tar xvof wmq53.1.17_H06.tar
5. Locate your WMQ V5.3.1.x installation(s). The service installation procedure
requires the full OSS path names of the opt/mqm and var/mqm directories for
each WMQ installation to which the fixpack will be installed.
6. Logon to OSS using the WMQ installation owner's userid
7. End all Queue Managers defined in the WMQ Installation.
endmqm <qmgr name>
Ensure all Queue Managers defined in the WMQ installation
are ended.
dspmq
Ensure that the WMQ installation is at a suitable V5.3.1 level.
mqver -V
See later notes concerning version requirements for NonStop BladeSystem
installation.
8. End any non-Pathway listeners for Queue Managers defined in the
WQM installation:
endmqlsr -m <qmgr name>
9. Verify that no files in the Guardian subvolumes of the installation to
be updated are open. The installation cannot proceed safely unless all
files in these subvolumes are closed. Use the TACL command 'FUP LISTOPENS'
for the files in all three subvolumes - an absence of output indicates
that no files are open. If files are shown to be open, use the output
from the command to identify processes that are holding files open.
10. Backup your WMQ Installation; the fixpack cannot be uninstalled.
instmqm -b can be used to back up an installation. Please refer
to the readme file included with release WMQ V5.3.1.
11. Install the fixpack by running the supplied service tool (svcmqm).
Svcmqm requires the location of the OSS var tree as well
as the OSS opt tree. These locations can be supplied automatically by
running svcmqm in an OSS shell where the environment variables for the
WMQ installation being updated have been established (typically by
sourcing "wmqprofile"). If this is the case, svcmqm does not require the
-i and -v parameters.
For example:
cd /usr/ibm/wmq/V53117
opt/mqm/bin/svcmqm -s /usr/ibm/wmq/V53116/opt/mqm
If the environment variables for the WMQ installation are not established in
the environment of svcmqm or if you want to update a WMQ installation other
than the one that your current WMQ environment variable points to, then
the locations of the OSS opt and var trees must be supplied explicitly using
the svcmqm command line parameters -i and -v.
For example:
cd /usr/ibm/wmq/V53117
opt/mqm/bin/svcmqm -s /usr/ibm/wmq/V53117/opt/mqm
-i /wmq1/opt/mqm
-v /wmq1/var/mqm
svcmqm will prompt to confirm the location of the OSS opt tree for the
installation to be updated.
Type "yes" to proceed.
Svcmqm will then update the installation. The current WMQCSTM file for
the installation will be renamed to BWMQCSTM as a backup copy, before it
is regenerated. Note that any changes to the WMQCSTM file you have made
will not be copied to the new WMQCSTM file, however they will be preserved
in the backup copy made before the WMQCSTM file was regenerated.
12. Repeat Steps 5-11 for any other WMQ installations that you want to update
with this fixpack.
13. You can install this fixpack in the WMQ placed installation files so that
any future WMQ product installations will include the fixpack updates.
To do this, locate your WMQ placed installation filetree containing the
opt directory, make this your current working directory (use 'cd') and
then unpack the contents of the tar archive for this fixpack over the placed
file tree. For example, if the placed files are located in the default
location /usr/ibm/wmq/V531, for a H-Series system:
cd /usr/ibm/wmq/V531
tar xvof /usr/ibm/wmq/V5319/wmq53.1.17_H06.tar
Initial Installation on a NonStop BladeSystem
---------------------------------------------
These instructions apply to installing WebSphere MQ for HP NonStop Server on
a NonStop BladeSystem using the original installation media, in conjunction
with the 5.3.1.17 package downloaded from IBM. NonStop BladeSystem platforms
are not supported prior to V5.3.1.4 and a "from scratch" installation requires
either V5.3.1.4 or later files to be overlaid on a set of placed files
from the base product media prior to performing the installation.
You do NOT need to perform these steps if you have already installed V5.3.1.4
on your NonStop BladeSystem. In this case, follow the standard installation
steps earlier in this readme file.
1. Place the files for the Refresh Pack 1 (5.3.1.0) version of WebSphere MQ
for HP NonStop Server on the target system. Refer to the "File Placement"
section in Chapter 3 of the "WebSphere MQ for NonStop Server Quick
Beginnings" guide. Pages 11-13 describes how to place the files.
Do not attempt to install the placed files using the instmqm script
that was provided with V5.3.1.0 at this time. The V5.3.1.0 version of
instmqm does not support installation on NonStop BladeSystem.
2. Unzip the 5.3.1.17 fixpack distribution package - wmq53.1.17.tar.zip.
The fixpack distribution package contains the following files:
readme_wmq_5.3.1.17 This README
wmq53.1.17_H06.tar.Z H-Series H06 Package
3. This installation requires the wmq53.1.17_H06.tar.Z package.
Locate the WMQ placed installation filetree containing the opt directory
prepared in step 1 above and upload the wmq53.1.17_H06.tar.Z fixpack
archive to this location in binary mode.
4. Extract the fixpack compressed tarfile using commands similar to:
cd /usr/ibm/wmq/V53117
uncompress wmq53.1.17_H06.tar.Z
5. Unpack the contents of the extracted tar archive for this FixPack over the
placed file tree. For example, if the placed files are located in the default
location /usr/ibm/wmq:
cd /usr/ibm/wmq
tar xvof /usr/ibm/wmq/V53117/wmq53.1.17_H06.tar
6. Use the extracted instmqm script in this FixPack to install the product
using the updated installation file tree and the instructions in Chapter 3
of "WebSphere MQ for NonStop Server Quick Beginnings" guide, pages 13-29.
Before beginning, review the list of changes to Chapter 3 detailed in the
"Documentation Updates" section at the end of this README file. Note also
that the list of installed files displayed will differ from those shown in the
examples in the manual.
Post-Installation
-----------------
If upgrading from WMQ V5.3.1, read the following post-installation instructions:
Non-Native TNS Applications:
Re-BIND any non-native (TNS) applications. See "Upgrading to V5.3.1.17" above
for more information.
Re-binding non-native (TNS) is REQUIRED if upgrading from V5.3.1.7 or earlier
releases but is RECOMMENDED if upgrading from V5.3.1.8 or later fixpack for
incorporation of internal defect 4158.
If you use SSL channels and have not already installed V5.3.1.1:
Edit the SSL certificate store, cert.pem and move all the CA certificates
to a new file, trust.pem, stored in the same directory as cert.pem. The
only items that should remain in cert.pem are the queue manager's personal
certificate and the queue manager's private key. These two items should
be located at the start of the cert.pem file. All other certificates
(intermediate and root CAs) must be moved to trust.pem. The trust.pem file
must be in the same directory as cert.pem, as configured in the queue
manager's SSLKEYR attribute.
Update the copy of the entropy daemon program that you run for SSL channels
on the system with the new version (...opt/mqm/ssl/amqjkdm0).
Enable new support for Danish CCSID 65024:
Customers who wish to enable the new support for Danish CCSID 65024
should do the following to install the revised ccsid.tbl file:
Issue the following commands on OSS:
1. Logon to OSS using the WMQ installation owner's userid
2. End all Queue Managers defined in the WMQ Installation.
endmqm <qmgr name>
3. Source in the installation's wmqprofile
. $MQNSKVARPATH/wmqprofile
4. cp -p $MQNSKOPTPATH/samp/ccsid.tbl $MQNSKVARPATH/conv/table/
5. Start queue managers
Guardian C++ DLLs:
Ensure that the WMQ Guardian C++ DLLs are 'executable' by using "FUP ALTER"
to set their FILECODE to 800 (for H-Series or J-Series). Use commands similar
to the following:
1. Logon to TACL using the WMQ installation owner's userid
2. OBEY your WMQ Installation's WMQCSTM file
3. VOLUME [#param MQNSKOPTPATH^LIB^G]
4. FUP ALTER IMQI2,CODE 800
5. FUP ALTER IMQI2T,CODE 800
6. FUP ALTER IMQI3,CODE 800
7. FUP ALTER IMQI3T,CODE 800
8. Logoff
Guardian Subvolume File Permissions
The WMQ Guardian Installation Subvolume and all WMQ Guardian Queue
Manager Subvolumes must accessible to both MQM group members
and to users that run WMQ application programs.
Ensure that:
All members of the MQM security group have read, write, execute
and purge permission to these subvolumes.
All users that run WMQ application programs, have read, write
and execute permission to these subvolumes
Restart Queue Managers:
Restart the queue managers for the installation you have
updated with this fixpack.
UNINSTALLATION INFORMATION
==========================
This fixpack cannot be automatically uninstalled if a problem occurs
during the update of an installation using svcmqm.
You should use the instmqm -b option to create a backup of an
installation before applying the service. If a problem occurs
or you need to reverse the upgrade at a later date, use the
instmqm -x option to restore a backup of the installation at the
prior service level.
KNOWN LIMITATIONS, PROBLEMS AND WORKAROUNDS
===========================================
This section details known limitation, problems, and workarounds for
WebSphere MQ for HP NonStop Server, Version 5.3.1.17.
Limitations
-----------
1. The current implementation of Publish / Subscribe is restricted to run
within a single CPU. The control program and all "worker" programs run in
the CPU that was used to run the 'strmqbrk' command.
The Publish/Subscribe broker does not automatically recover in the event
of CPU failures.
2. The current memory management implementation in the Queue Server limits
the total amount of non-persistent message data that can be stored on all
the queues hosted by a single Queue Server to less than 1Gb. The limit of
non-persistent message data on a single queue can not exceed approximately
1Gb therefore, even if a single Queue Server is dedicated to that queue.
3. The number of threads in threaded agent processes (LQMAs or MCAs) or in MQI
applications, is limited to a maximum of 1000 by the limit on open depth of
the HP TMF T-file.
4. API exits are not supported for non-native (TNS) applications. Any other
exit code for non-native applications must be statically bound with the
TNS application.
5. Cluster workload exits are only supported in "trusted" mode. This means
that a separate copy of each exit will run in each CPU and exit code in
one CPU cannot communicate with exit code in another CPU using the normal
methods provided for these exits.
6. Upgmqm will not migrate the following data from a V5.1 queue manager:
messages stored in Message Overflow files (typically persistent messages
over 200,000 bytes in size) will not be migrated. If the option to
migrate message data was selected, the upgrade will fail. if the option
to migrate message data was not selected, the upgrade will not be
affected by the presence of message overflow files.
clustering configuration data - all cluster related attributes of objects
will be reset to default values in the new V5.3 queue manager.
SNA channel configuration - channels will be migrated, but several of the
attributes values will need to be changed manually after the upgrade.
channel exit data - attributes in channels that relate to channel exit
configuration will be reset to default values in the new V5.3 queue
manager.
In all cases where upgmqm cannot migrate data completely, a warning message
is generated on the terminal output as well as in the log file. These can
be reviewed carefully after the upgrade completes for further actions that
may be necessary.
7. Java and JMS limitations
The Java and JMS Classes do not support client connections. WebSphere MQ
for HP NonStop Server does not support XA transaction management, so the
JMS XA classes are not available. For more detail, please refer to the
Java and JMS documentation supplement, Java.pdf.
8. Control commands in Guardian (TACL) environment do not obey the RUN option
"NAME" as expected
A Guardian control command starts an OSS process to run the actual
control command - and waits for it to complete. When the NAME option is
used, the Guardian control command process uses the requested name, but
the OSS process cannot and is instead named by NonStop OS.
If the Guardian control command is prematurely stopped by the operator
(using the TACL STOP command for example) the OSS process running the
actual control command may continue to run. The OSS process may need to be
stopped separately and in addition to the Guardian process.
9. Trace doesn't end automatically after a queue manager restart
(APAR IC53352) and trace changes do not take effect immediately
If trace is active and a queue manager is restarted, the trace settings
should be reset to not trace any more. Instead, the queue manager
continues tracing using the same options as before it was restarted.
The workaround is to disable trace using endmqtrc before ending, or while
the queue manager is ended.
Also, changes to trace settings do not always take effect immediately
after the command is issued. For example, it could be several MQI calls
later that the change takes effect. The maximum delay between making a
trace settings change and the change taking effect would be until the end
of the queue manager connection, or the ending of a channel.
10. Some EMS events generated to default collector despite an alternate
collector being configured (APAR IC53005)
An EMS event message "FFST record created" is generated using the OSS
syslog() facility whenever an FDC is raised by a queue manager. This
EMS event cannot be disabled and goes to the default collector $0. For
OSS processes, an alternate collector process can be specified by
including an environment variable in the context of these processes
as in the following example:
export EMS_COLLECTOR=\$ALT
Guardian processes always use the default collector because HP do not
provide the ability to modify the collector in the Guardian environment.
HP is investigating if a change is possible. No fix for this problem
has yet been identified.
11. The use of SMF (virtual) disks with WMQ is not supported on Release
Version Updates prior to H06.26 and J06.15 because of restrictions
imposed by the OSS file system. For more details, please refer to the
HP NonStop Storage Management Foundation User's Guide Page 2-12.
12. The maximum channel batch size that can be configured (BATCHSZ attribute)
is 2000. If you need to run channels with batch sizes greater than 680
you must increase the maximum message length attribute of the
SYSTEM.CHANNEL.SYNCQ to 60000.
e.g. from RUNMQSC - ALTER QL (SYSTEM.CHANNEL.SYNCQ) MAXMSGL (60000)
13. The SYSTEM.CHANNEL.SYNCQ is a critical system queue for operation of
the queue manager and should not be altered to disable MQGET or MQPUT
operations, or to reduce the maximum message size or maximum depth
attributes from their defaults.
14. Currently, the cluster transmission queue (SYSTEM.CLUSTER.TRANSMIT.QUEUE)
cannot be moved to an alternative Queue Server because it is constantly
open by several internal components. The following procedure (which
requires a "quiet" queue manager and a queue manager restart) can be
used to achieve this reconfiguration. Do not use this procedure on a
queue manager that is running in production. Read and understand the
procedure carefully first since it includes actions that cause
internal errors to be generated in the queue manager.
1/ Rename the OSS repository executable (opt/mqm/bin directory)
mv amqrrmfa amqrrmfax
2/ From OSS enter 'ps -f | grep amqrrmfa | grep X '
where X is the Queue Manager name.
3/ kill -9 those processes returned from step 2
At this point the EC will start continuously generating FDCs and
log messages as it attempts to, and fails, to restart the repository
servers that were stopped. Perform the remaining steps in this
procedure without delay to avoid problems with excessive logging
such as disk full conditions.
4. Verify the processes are stopped
From OSS enter 'ps -f | grep amqrrmfa | grep X '
Where X is the Queue Manager name
5. Issue the altmqfls --server command to move the cluster transmission
queue to an alternate queue server
6. Issue dspmqfls to verify the alternate server assignment
7. Rename the OSS repository executable back to the expected name
mv amqrrmfax amqrrmfa
8. End the queue manager using preemptive shutdown. The EC will generate
FDCs ending because of the earlier attempts to start a repository
manager while the executable was renamed. There will be FDCs and
EC Primary process failover
Component :- xcsExecProgram
Probe Description :- AMQ6025: Program not found
Comment1 :- No such file or directory
Comment2 :- /opt/mqm/bin/amqrrmfa
...
AMQ8846: MQ NonStop Server takeover initiated
AMQ8813: EC has started takeover processing
AMQ8814: EC has completed takeover processing
...
The EC may have to be manually TACL stopped if quiesce or immediate
end is used thus the need for the preemptive shutdown
endmqm -p <qmgr name>
9. Rename the repository manager executable to the original name
mv amqrrmfax amqrrmfa
10. Restart the queue manager
strmqm <qmgr name>
15. In Guardian/TACL environments, support for some WMQ command-line programs
has been deprecated for WMQ Fixpack 5.3.1.3 and later.
The affected command-line programs are:
amqoamd
amqrdbgm
amqrfdm
crtmqcvx
endmqlsr
runmqchi
runmqchl
runmqdlq
runmqlsr
runmqtrm
These programs will continue to function for now, however their use in
Guardian/TACL environments is discouraged. Support for these programs
in Guardian/TACL environments may be withdrawn completely in a future
WMQ 5.3 release/fixpack.
IBM recommends that customers use the OSS version of these programs instead.
Customers who want to route output from WMQ OSS tools to VHS or other Guardian
collectors should use the OSSTTY utility. OSSTTY is a standard utility
provided by OSS and is described in the HP publication "Open System Services
Management and Operations Guide".
Note: See Item 3. in "Known problems and workarounds" for a description of
restrictions when using the MQ Broker administration commands in the
Guardian/TACL environment.
16. Do not use WebSphere MQ with a $CMON process that alters attributes of WMQ
processes (for example the processor, priority, process name or program
file name) when they are started. This is not a supported environment since
there are components in WMQ that rely on these attributes being set as
specified by WMQ for their correct operation.
17. Support for forked processes
MQI Support from forked processes in OSS is subject to the following
restrictions:
1. If forking is used, MQI operations can be performed only from child
processes. Using MQI verbs from a parent process that forks child
processes is not supported and will result in unpredictable
behavior.
2. Use of the MQI from forked processes where the parent or child is
threaded is not supported.
18. TMF Outage handling
TMF outage handling was significantly improved with V5.3.1.4, however there
is still a limitation in V5.3.1.6 and later to be aware of:
1. If a STOP TMF, ABRUPT command is issued, TMF marks all open audited
files as corrupt and the queue manager cannot perform further
processing until this condition is rectified by restarting TMF.
In this state, the queue manager will freeze further operation and log
the condition in the queue manager log file every 10 seconds for a
maximum of 50 attempts. Whether or not TMF is restored within this
timeframe, the WMQ queue manager should be restarted to reduce the risk
of any undetected damage persisting.
19. Triggering HP NSS non-C Guardian applications
The MQ default Trigger Monitor process, runmqtrm, at present cannot
directly trigger the following application types:
Guardian TACL scripts or macro file
COBOL application
TAL application
An OSS script file (wmqtrig) provides indirect support for these
files types. To use this script, the PROCESS definition APPLTYPE should
be set to UNIX and the APPLICID should refer to the script as in
the following examples:
For a TACL script called "trigmacf":
APPLICID('/opt/mqm/bin/wmqtrig -c \$data06.fp4psamp.trigmacf')
APPLTYPE(UNIX)
For a COBOL or TAL application called "mqsecha":
APPLICID('/opt/mqm/bin/wmqtrig -p /G/data06/fp4psamp/mqsecha')
APPLTYPE(UNIX)
Notes:
1. TACL scripts use the wmqtrig script with a "-c" option.
The -c option should use the Guardian representation for file name
of the TACL script file, with the special character ($) escaped,
for example:
\$data06.fp4psamp.trigmacf
2. COBOL and TAL applications use the wmqtrig script with a "-p" option.
The -p option must use the OSS representation for the file name of
the application, for example:
/G/data06/fp4psamp/mqsecha
3. C applications can be triggered directly by specifying
APPLICID('$DATA06.FP4PSAMP.MQSECHA')
APPLTYPE(NSK)
To trigger a PIC application using the WMQ Pathway MQS-TRIGMON00
serverclass, a DEFINE is required:
=_RLD_LIB_PATH,CLASS SEARCH,SUBVOL0 <Guardian MQ binary Volume.Subvolume>
For example:
ALTER MQS-TRIGMON00,
DEFINE =_RLD_LIB_PATH,CLASS SEARCH,SUBVOL0 $DATA06.FP4PBIN
4. If the "-p" option is used, gtacl passes the complete MQTMC2 structure
text (which is 560 bytes) to the application being triggered, whereas if
the "-c" option is used, limitations in TACL will cause the triggered
application to receive 520 bytes only.
Applications intended to be triggered using -p option must handle the
complete 560 character startup character string.
This can cause problems, particularly with COBOL applications;
Since a COBOL GETSTARTUPTEXT call can process only 529 characters,
triggering a COBOL application with the -p option (560 character startup
string) can result in a memory overwrite and application abend.
In this case, the -c option should be used instead of the -p option."
20. Maximum numnber of LQMA processes
The maximum number of LQMA processes per queue manager is 1417. Attempts to
configure a MaxUnthreadedAgents value of 1418 or greater in the qmproc.ini file
will result in FDCs when the queue manager attempts to start the 1418th LQMA.
21. Limitation on ProcessNameRoot values in qmproc.ini file
The ProcessNameRoot value used in the qmproc.ini file for the MCA, LQMA and
RepositoryManager stanzas must be unique across all queue managers in all
installations on the system. If the values are not unique, two queue
managers attempting to create a new process name at the same time may attempt
to use the same sequence of names, resulting in heavy load on the OSS
nameserver and/or FDCs with probes EC062000 from eclStartMCA or EC065000 from
eclStartLQMA. This may result in the queue manager becoming unresponsive.
22. BIND/REPLACE Warning 9
The use of BIND/REPLACE when re-binding non-native (TNS) applications with
the latest WMQ TNS Library is supported, however, when using this command,
you may encounter many Bind 'warning 9' messages. These warnings are safe to
ignore, as the changes made in the MQMTNS library are completely contained
within that library and there is no external affect on an application that
would link with that library. Please refer to document
ID mmr_ns-4.0.6050052.2565545 in KBNS which has been updated by HP to include
this information.
23. PING CHANNEL uses the =TCPIP^PROCESS^NAME DEFINE as opposed to the
value set in the qmproc.ini file.
If the TCPIP^PROCESS^NAME DEFINE is invalid, attempts to issue a
PING CHANNEL request will fail with the following message -
":AMQ9212: A TCP/IP socket could not be allocated."
24. Using multiple cluster receiver channels for a queue manager can cause the
master and slave repository managers to get out of sync in some configurations,
this results in FFSTs from rrmHPNSSGetMetaPacket with Comment1 field
"Meta data mis-match Expected: metalen=xx".
The root cause of this problem has not been identified definitively, but it is
recommended that cluster configurations do not use multiple cluster receiver
channels for the same cluster on a given queue manager.
25. Format 2 Queue and Queue overflow files are not supported
If an attempt is made to use altmqfls to change extent size or max extents of
a Queue or Queue overflow file such that the new file size would require a
format 2 file, the attempt will fail and an FDC file containing the following
Probe Id :- XC066050
Component :- xgcDupPartFile
...
Major Errorcode :- xecF_E_UNEXPECTED_RC
Minor Errorcode :- krcE_UNEXPECTED_ERROR
Probe Description :- AMQ6118: An internal WMQ error has occurred
(20800893)
...
Arith1 :- 545261715 20800893
26. RUNMQSC DISPLAY QSTATUS(*) TYPE(HANDLE) ALL can return only 500 handles
The current implementation of DISPLAY QSTATUS can handle only 500 queue
handles. In prior releases, executing the command in a scenario where there
were more than 500 handles resulted in an error return and large numbers of
FFSTs (see internal defect 2230 earlier in this readme). These errors are the
result of a design limitation with the DISPLAY QSTATUS implementation.
The processing of the command has been modified to handle the scenario cleanly
and return only the first 500 handles retrieved, as supported by the design.
Known problems and workarounds
------------------------------
1. FDCs from Component xcsDisplayMessage reporting xecF_E_UNEXPECTED_SYSTEM_RC
On RVUs H06.06 and later:
These FDCs occur frequently on queue manager shutdown, and at times during
queue manager start, from processes that write to the queue manager log
files at these times, typically the cluster repository cache manager
(amqrrmfa) and the EC (MQECSVR). No functional problem is caused by these
FDCs, except that the queue manager log file misses some log messages
during queue manager shutdown. The FDCs report an unexpected return code
from the HP lseek() function. An example of an FDC demonstrating this
problem follows:
Probe Id :- XC022011
Component :- xcsDisplayMessage
Program Name :- $DATA06.RP1PBIN.MQECSVR
Major Errorcode :- xecF_E_UNEXPECTED_SYSTEM_RC
MQM Function Stack
nspPrimary
eclShutdownOK
xcsDisplayMessageForSubpool
xcsDisplayMessage
xcsFFST
6fffe660 000011FA ....
6fffe670 2F686F6D 652F726F 622F4D51 /home/test/MQ
6fffe680 352E332F 5250312F 50726F64 2F776D71 5.3/RP1/Prod/wmq
6fffe690 2F766172 2F6D716D 2F716D67 72732F51 /var/mqm/qmgrs/Q
6fffe6a0 4D312F65 72726F72 732F414D 51455252 M1/errors/AMQERR
6fffe6b0 30312E4C 4F47 01.LOG
This problem is fixed by the following HP SPRs
T8620ACL (OSSFS) for G06 HP OS
T8620ACM (OSSFS) for H06 HP OS
2. APAR IC54594 - EC abends with non-MQM user running application from non-MQM
directory.
Statically-bound TNS applications that are not relinked after installing
Fixpack 5.3.1.4 have additional considerations. For these applications,
qmproc.ini Application Rules 2 and 4 will not work if the application
is located in a non-MQM directory.
3. The Guardian control commands for the Publish / Subscribe broker
(strmqbrk, endmqbrk, dspmqbrk ... etc) will not work correctly unless they
are run in the same CPU as the broker is running in, or was last running in.
Please use the equivalent OSS commands instead of the Guardian versions, or
ensure that the Guardian Publish / Subscribe broker commands run in the same
CPU as the broker was or is running in.
4. Queue managers occasionally do not delete Temporary Dynamic Queues the
when the last application closes them. The cause of this is unknown at present.
The problem is rare and unlikely to cause significant impact on queue
manager operation unless the queues are present in very large numbers.
Queues orphaned in this way cannot be used by applications, and are removed
unconditionally as a part of the normal garbage collection activity during
a queue manager restart.
DOCUMENTATION UPDATES
=====================
Please note that several supplements to the documentation have been provided
with fixpacks since V5.3 was originally released. These supplements have
been released in Adobe Acrobat format and can found in the
opt/mqm/READMES/en_US directory of any installation as well as the original
software distribution tree (placed files). The following supplements have
been released to date (the name of the file describes the content):
Exits.pdf
Java.pdf
Pubsub.pdf
SNAChannels.pdf
SSLUpdate.pdf
Upgmqmreadme.pdf
Sdcp.pdf
Also please note that the current published versions of the cross-platform
("family") books contain references to the IBM MQSeries V5.1 for Compaq NSK
product which is the previous major version of WebSphere MQ for HP NonStop
Server. Consequently, these references may not be accurate with respect to
the functional support provided by V5.3.1.
Websphere MQ Programmable Command Formats and Administration Interface
(SC34-6060-03)
-----------------------------------------------------------------------
Chapter 3 - PCF Commands and Responses in Groups
Page 19: Add "Refresh Queue manager" as a supported command
Chapter 4 - Definitions of Programmable Command Formats
Page 173: Add the following new command:
Refresh Qmgr
The Refresh Qmgr (MQCMD_REFRESH_Q_MGR) command refreshes the
Execution Controller (EC) process management rules.
This PCF is supported only on WebSphere MQ V5.3 HP NonStop Server.
Required parameters:
None
Optional parameters:
None
Error codes
This command might return the following in the response format
header, in addition to the values shown on page 18.
Reason (MQLONG)
The value can be:
MQRCCF_PARM_COUNT_TOO_BIG
Parameter count too big.
WebSphere MQ for HP NonStop Server Quick Beginnings (GC34-6626-00)
------------------------------------------------------------------
Chapter 1 - Planning to install WebSphere MQ for HP NonStop Server
Page 1: the baseline release level for V5.3.1 on the HP Integrity NonStop
Server is now H06.05.01
Page 1: the typical approximate storage requirements are as follows:
+ OSS files placed before installation:
H-Series: 160Mb
+ For each installation:
H-Series: Guardian 220Mb, OSS 350Mb, Total 570Mb
+ For each queue manager:
H-Series: Guardian 9.5Mb, OSS 0.2Mb, Total 10Mb
Pages 2 & 3: please review the section "Hardware and Software
Requirements" in these release notes for the details of all other updated
requirements
Chapter 3 - Installing WebSphere MQ for HP NonStop Server
Page 12: Product Selection dialog. The names of the products have been
updated to "WebSphere MQ V5.3.1" and "WebSphere MQ V5.3.1 Integrity".
Page 14: instmqm now includes the function of creating an automatic
backup archive of a successful installation, as follows:
Instmqm has been enhanced to provide the ability to back-out an ungrade
installation, and the ability to archive and restore installations
individually. Before instmqm starts to make changes to a system, it will
automatically create an archive of the current installation (OSS opt tree
and Guardian installation subvolumes only) in the root directory
containing the opt tree in OSS. If a failure occurs during installation,
and instmqm has made changes, the user will be asked if they wish to
restore the installation to its original state using the archive created
before changes were made. At the end of a successful installation,
instmqm will now automatically create a backup archive of the new
installation.
Instmqm also supports two new command line options to support creating
and using backup archives independently from an installation:
-b create a backup archive of the installation
-x restore an installation from a backup archive
These options may not be combined with any other options. Both options
require the user to respond to questions at the terminal.
A backup archive file is an OSS pax file, created as follows:
+ the Guardian PAK utility is used to create a backup of the three
Guardian subvolumes for the installation in a file named "WMQSAVED"
+ the PAK backup file is copied to the OSS opt directory of the
installation that is being archived
+ the entire OSS opt tree of the installation (which now includes
WMQSAVED) is then archived by the OSS pax utility
Backup archive files are always created in the directory that holds the
OSS opt tree for the installation. Archive files created automatically
by instmqm are named "mqarchive-yymmdd-hhmmss" where "yymmdd" and
"hhmmss" are numeric strings of the date and time that the backup archive
was created - for example: "mqarchive-061005-143606".
Page 15: instmqm has new command line options as described in these
release notes for creating and restoring backup archives
Page 17: the SnaProviderName and TcpProviderName fields of the
QmgrDefaults stanza in the instmqm response file are used to populate
the proc.ini file to provide installation wide defaults for channels.
Please note that these fields do not get used for the default listener
configuration either on the command line (runmqlsr) or in the queue
manager's Pathway environment. Users must manually configure the
transport names for all listeners.
Page 28: in addition to the manual methods for cleaning up after a failed
installation, instmqm will offer the option to restore the previous
installation from a backup archive in the event of a failure while
upgrading a V5.3 installation to V5.3.1 level. These release notes
describe the additional function.
If an installation was initially created without SSL (selection of the
installation type "CORE" for instmqm), the following procedure can be
used to update the installation to include SSL components. In the
instructions below, <MQInstall> refers to the location of the
installation that needs to be updated and <PlacedInstall> means the
location of the complete set of placed files for the level of WMQ that
corresponds to the installation being updated. All queue managers
must be ended before attempting this procedure.
1. mkdir <MQInstall>/opt/mqm/ssl
2. chmod 775 <MQInstall>/opt/mqm/ssl
3. cp <PlacedInstall>/opt/mqm/ssl/* <MQInstall>/opt/mqm/ssl
4. chmod 775 <MQInstall>/opt/mqm/ssl/amq*"...
5. cp <MQInstall>/opt/mqm/ssl/openssl <MQInstall>/opt/mqm/bin
6. chmod 664 <MQInstall>/opt/mqm/ssl/openssl
7. chmod 774 <MQInstall>/opt/mqm/bin/openssl
8. cp <MQInstall>/opt/mqm/ssl/amqjkdm0 <MQInstall>/opt/mqm/bin
9. chmod 775 <MQInstall>/opt/mqm/bin/amqjkdm0
10. mv <MQInstall>/opt/mqm/lib/amqcctca
<MQInstall>/opt/mqm/lib/amqcctca_nossl
11. mv <MQInstall>/opt/mqm/lib/amqcctca_r
<MQInstall>/opt/mqm/lib/amqcctca_r_nossl
12. cp <MQInstall>/opt/mqm/ssl/amqccssl <MQInstall>/opt/mqm/lib/amqcctca
13. cp <MQInstall>/opt/mqm/ssl/amqccssl_r
<MQInstall>/opt/mqm/lib/amqcctca_r
14. chmod 775 <MQInstall>/opt/mqm/lib/amqcctca*
15. The <MQInstall>/var/mqm/qmgrs/<qmgr name> directory should have an
ssl directory which is where you will store the certificate related
files (cert.pem, trust.pem etc.)
16. The <MQInstall>/opt/mqm/samp/ssl should exist already with the ssl
samples
17. If the entropy daemon is not configured on the system this will need
to be performed. Refer to the WMQ V53 HP NonStop System
Administration Chapter 11 page 165-167
18. Install the certificates per the updated instructions, SSLupdate.pdf
found in <MQInstall>/opt/mqm/READMES/en_US
Chapter 5 - Creating a Version 5.3 queue manager from an existing Version 5.1
queue manager
Pages 37 & 38: this section is completely replaced by the documentation
supplement Upgmqmreadme.pdf supplied with this release.
Chapter 7 - Applying maintenance to WebSphere MQ for HP NonStop Server
Pages 44 & 45: the tool for applying maintenance is named "svcmqm" and
not "installCSDxx".
Page 44: in step 3 of "Transferring and preparing the PTF for
installation", the top level directory of the PTF is opt, and is not
named differently for each PTF. Therefore it is important to manually
create a directory specific to each PTF, download the PTF to that new
directory and then expand the archive within the new directory.
Page 44: in step 2 of "Running the installation script for a PTF", the
svcmqm tool has a different command line from that documented for
"installCSDxx". svcmqm takes three parameters:
svcmqm -i installationtree -v vartree -s servicepackage
where "installationtree" is the full path to the location of the opt/mqm
directory of the installation to be updated
"vartree" is the full path to the location of the var/mqm
directory of the installation to be updated
"servicepackage" is the full path to the location of the opt/mqm
directory of the maintenance to be installed
For example:
svcmqm -i /home/me/wmq/opt/mqm -v /home/me/wmq/var/mqm
-s /home/me/wmqfiles/opt/mqm
which will update the installation in /home/me/wmq/opt/mqm and
/home/me/wmq/var/mqm from the maintenance package in directory tree
/home/me/wmqfiles/opt/mqm.
If either or both the "-i installationtree" and "-v vartree" parameters
are omitted, svcmqm will use the current setting of the appropriate
environment variable - either WMQNSKOPTPATH or WMQNSKVARPATH.
WebSphere MQ for HP NonStop Server System Administration Guide (SC34-6625-00)
-----------------------------------------------------------------------------
Chapter 2 - An introduction to WebSphere MQ administration
Page 16: before running any control commands on OSS or NonStop OS it is
necessary to establish the environment variables for the session. When
an installation is created a file called wmqprofile is also created in
the var/mqm directory that will establish the environment for an OSS
shell. Likewise, a file is also created in the NonStop OS subvolume
containing the WMQ NonStop OS samples called WMQCSTM that can be used
to set up the appropriate environment variables for a NonStop OS TACL
session.
To establish the WMQ environment for an OSS shell session:
. wmqprofile
To establish the WMQ environment for a NonStop OS TACL session:
obey WMQCSTM
The same steps are required before running any applications in the
OSS or NonStop OS environment.
Chapter 4 - Administering local WebSphere MQ objects
Page 48: when creating a Process definition, the default value for
the APPLTYPE attribute is "NSK" (indicating a Guardian program)
Chapter 7 - WebSphere MQ for HP NonStop Server architecture
Page 80: the MQSC command to reload the process management rules is
REFRESH QMGR TYPE(NSPROC) and not RESET QMGR TYPE(NSPROC)
Chapter 8 - Managing scalability, performance, availability and data
integrity
Page 104: the last paragraph of the OpenTMF section should be reworded
as follows:
No special administrative actions are required for this use of TMF.
WebSphere MQ uses and manages it automatically. You must ensure that
the RMOPENPERCPU and BRANCHESPERRM configuration parameters of TMF are
set to appropriate values for your configuration. Please see Chapter 12
Transactional Support - Configuring TMF for WebSphere MQ for
information on how to calculate the correct values. The HP TMF Planning
and Configuration Guide describes the subject of resource managers and
heterogeneous transaction processing.
Chapter 9 - Configuring WebSphere MQ
Page 119: the CPUS section should state that the default can be
overridden using the crtmqm -nu parameter. See Chapter 18 - The control
commands for a description of how to use this parameter with crtmqm.
Page 120: the section describing the ARGLIST attribute of a TCP/IP
Listener should also mention the use of the optional -u parameter to
configure channels started by the listener as unthreaded processes.
The default is to run incoming channels as threads in an MCA process.
Page 130: the MQSC command to reload the process management rules is
REFRESH QMGR TYPE(NSPROC) and not RESET QMGR TYPE(NSPROC)
Page 133: Figure 23 remove :
OAM Manager stanza #
OamManager:
Page 136: the Exit properties section should state that the only
supported way of configuring and running a Cluster Workload (CLWL) Exit
for HP NonStop Server is in FAST mode. The CLWLMode setting in qm.ini
is required to be set to FAST, which is the default for WebSphere MQ
on this platform.
Page 139: the MQIBindType attribute of the Channels stanza is set by
crtmqm to FASTPATH. This should not be changed, except under the
direction of IBM Service.
Page 140: the AdoptNewMCA=FASTPATH option is always required for
this platform in order for the adoption of MCAs to be effective. The
"Attention!" box after the description of the FASTPATH option should
be ignored.
Page 140: add the following description of the ClientIdle attribute:
ClientIdle=seconds
ClientIdle specifies the number of seconds of inactivity to permit
between client application MQI calls before WebSphere MQ terminates
the client connection. The default is to not terminate client
connections however long they remain inactive. When a client connection
is terminated because of idle activity, the client application receives
a connection broken result (2009) on its next MQI call.
Chapter 11 - Working with the WebSphere MQ Secure Sockets Layer (SSL) support
A documentation supplement has been written to replace the sections on
Page 170 (Preparing the queue manager's SSL files) to Page 176 (Building
and verifying the sample configuration) because of changes to the files
that WebSphere MQ uses to hold certificates. The documentation supplement
is called SSLupdate.pdf, and can be found in the opt/mqm/READMES/en_US
directory of an installation.
Chapter 12 - Transactional Support
Page 185: The descriptions of the TMF attribute RMOPENPERCPU in the
Resource manager configuration section is modified as follows:
RMOPENPERCPU
Each WebSphere MQ thread or process that handles transactions has
an open of a Volatile Resource Manager in the CPU it runs in. In
addition, each application thread or process using the MQI also has
an open. The minimum requirement for this configuration parameter
is therefore the sum of:
+ all Queue Server processes in that CPU
+ all LQMA and MCA threads running in that CPU
+ all MQI application threads running in that CPU
+ 10 (to account for miscellaneous queue manager processes that
could be running in that CPU)
You should calculate the peak values of these numbers across all CPUs
and add a safety margin to arrive at the correct value for your system.
The HP default value of 128 for this parameter is often suitable for
small configurations, but unsuitable for medium or large ones.
Page 186: add the following paragraph to the Troubleshooting section
for Configuring TMF:
If the RMOPENPERCPU value is not configured to allow sufficient opens
of resource managers in a CPU, WebSphere MQ connections will fail with
an unexpected return code, and FDCs will be generated reporting an
error with the TMF_VOL_RM_OPEN_. The workaround is to distribute
applications and queue manager processes in the CPU that exceeds
the limit to other CPUs. The correct remedy is to schedule an outage
and modify the TMF configuration.
Page 186: add the following paragraph to the Troubleshooting section
for configuring TMF:
If TMF is stopped, or new transactions are disabled, and WMQ requires
an internal "unit of work" (TMF transaction) to perform an update to
a protected resource requested by an MQI call, that call will fail
and the reason code returned will be MQRC_UOW_NOT_AVAILABLE (2255).
Note that in some cases, updates to protected resources may be
required by MQI operations do not directly perform messaging
operations - for example, MQOPEN of a model queue that creates a
permanent dynamic queue. If MQI calls return MQRC_UOW_NOT_AVAILABLE,
check the status of the TMF subsystem to determine the likely cause.
Chapter 14 - Process Management
Page 197: the MQSC command to reload the process management rules is
REFRESH QMGR TYPE(NSPROC) and not RESET QMGR TYPE(NSPROC)
Page 200 and 204: the default value for the maximum number of unthreaded
agents is now 200. The default value for the maximum number of threaded
agents is now 20. the default value for the maximum use count for
threaded agents is now 100.
Page 203: add a new paragraph titled "Pathway":
Pathway
This stanza contains 3 attributes:-
- ProcessName
- DynamicProcessName
- Hometerm
ProcessName is the name of the Queue Managers pathmon process.
If the -np option was specified at queue Manager creation,
the value of ProcessName will be set to the value of that option
when the qmproc.ini file is created
If DynamicProcessName is set to Yes, the system will generate a name
for the pathmon process at the time the Queue Manager starts.
If the value is set to no, the value of the ProcessName attribute
will determine the pathmon process name for the queue manager
Hometerm specifies the value of the hometerm attribute for the Queue
Manager pathmon process
If the -nh opton was specified at queue manager creation, the
value of Hometerm will be set to the value of that option,
otherwise the default of $ZHOME will be used.
Page 204: the "valid attribute values" for the attribute "ExecutableName"
should be stated as "File name part only of the program to run for the
LQMA or MCA process".
Pages 203 - 205, Table 20: Process Management: Keyword definition Summary
There are a number of errors in the Process Management Keyword
definition table:
1. Environment variables:
ENVxx should be Envxx
2. Executable Name to Match:
ExecNameMatch should be ExeNameMatch
3. Fail if CPU unavailable:
FailOnCPUunavail should be FailOnCPUUnavail
4. Preferred number of Threaded Agents:
PreferedThreadedAgents should be PreferredThreadedAgents
Default values:
5. MaxThreadedAgents: change from 10 to 20
6. MaxUnthreadedAgents: change from 20 to 200
7. MaxThreadedAgentUse: change from 10 to 100
Pages 199 - 201, Table 16. Process management: agent attributes
The same default value changes are required:
1. Maximum number of unthreaded agents: 200
2. Maximum number of threaded agents: 20
3. Maximum reuse count for threaded agents: 100
Chapter 15 - Recovery and restart
Page 216: Configuring WebSphere MQ, NonStop RDF, and AutoSYNC to support
disaster recovery
To configure RDF to work with a existing WMQ V53 queue manager:
End the WMQ V53 queue manager.
Using the HP BACKUP or PAK utility specifying the AUDITED option
Backup the primary site Guardian WMQ queue manager subvolume.
Using the HP RESTORE or UNPAK utility specifying the AUDITED option
Restore the files on the backup site.
Ensure that on the backup system that the alternate key file
attribute (ALTKEY) for files amqcat and amqpdb of each queue
manager are set to the correct (backup system) node name
Page 217: the example of the altmqfls command to set the RDF
compatibility mode for large persistent messages is correct but too
simplistic. Please use care when using altmqfls to set the queue options
(--qsoptions parameter) and refer to the reference section for the
control commands for a complete description of using this option.
Page 217: the bullet point that describes the configuration of AutoSYNC
filesets is incorrect when it states that NO ALLSYMLINKS should be
specified. Replace sub-bullet item number 2 with the following text:
2. The entire queue manager OSS directory structure
var_installation_path/var/mqm/qmgrs/qmname.
You must specify the absolute path name of the queue manager's
directory. Specify the ALLSYMLINKS option for this fileset to
ensure that AutoSYNC correctly synchronizes the symbolic link
(G directory) in the queue manager's directory to the NonStop OS queue
manager's subvolume on the backup system.
Chapter 16 - Troubleshooting
Page 230: after the section "Is your application or system running
slowly?", insert the following new section:
Are your applications or WebSphere MQ processes unable to connect?
If connection failures are occurring:
is the User ID under which the application runs authorized to
use this queue manager?
are SAFEGUARD permissions preventing read access to the WebSphere
MQ installation files by the User ID running the application?
are the environment variables established for the application
process, so that the correct installation of WebSphere MQ is being
used?
if necessary, has the application been relinked or rebound with
any static MQI libraries that it uses?
is a resource problem preventing the queue manager from allowing
the connection? Review the troubleshooting section under TMF
Configuration on Page 185 and 186 for information about the
RMOPENSPERCPU TMF attribute.
Chapter 18 - The control commands
Page 93: The example of the --resetmeasure option is missing a mandatory
parameter having the value "YES" or "NO". The paragraph on page 93 describing
the --resetmeasure option should be replaced with the following
"The queue server can maintain the Measure counter only if it is included
in an active measurement. If it is not included in an active measurement,
and messages are put in the queue and removed from the queue, the value
of the counter will no longer represent the current depth of the queue.
If the counter is subsequently included in an active measurement, you can
cause the queue server to reset the Measure counter to the current depth
of the queue by using the --resetmeasure parameter on the altmqfls command,
as follows: altmqfls --qmgr QMGR --type QLOCAL --resetmeasure TEST.QUEUE YES"
Page 244: The mandatory YES|NO parameter is missing from the syntax diagram
Page 247: The mandatory YES|NO parameter is missing from the description
of the option
Page 243: the control commands for the Publish / Subscribe broker are
not referenced here. Refer to the WebSphere MQ V6.0 Publish/Subscribe
User Guide and the documentation supplement for Publish/Subscribe on
HP NonStop Server - Pubsub.pdf.
Page 255: if the OSS environment variable or Guardian PARAM MQPATHSEC
is defined and set to one of the standard NonStop OS security attributes
(A, N, C, G, O, or U) when crtmqm is run, the default PATHWAY SECURITY
attribute value of "G" will be overridden by the value of the environment
variable / PARAM. This can be used to restrict access to the queue
manager's Pathway environment. The current Pathway attributes can be
displayed in PATHCOM using the INFO PATHWAY command.
Page 255: the -nu parameter for setting the default CPUS attribute
in Pathway serverclasses does not accept all the values that Pathway
allows for this attribute. The only accepted values (and the result in
Pathway configuration) are of the form:
-nu value Pathway CPUS attribute
-------- ---------------------
-nu a CPUS (a:0)
-nu a:b CPUS (a:b)
More complex Pathway serverclass CPUS attributes settings must be
configured after the queue manager has been created, using the HP
PATHCOM utility.
Chapter 23 - API exits
Pages 373-375: please review the updates to this section in the
documentation supplement for API exits for HP NonStop Server called
Exits.pdf. This supplement has been extensively revised for V5.3.1.1
to clarify the requirements and process for creating and integrating
exits with WebSphere MQ.
Appendix B - Directory structure
Pages 430 and 431: there is a new G symbolic link to the Guardian
subvolume containing the product executables in
.../var/mqm/qmgrs/@SYSTEM
Page 431: the content of the ssl directory is revised with V5.3.1.1
as follows:
This directory contains up to four files used by the SSL support:
The queue manager certificate and private key store (cert.pem)
The trusted certificates store (trust.pem)
The pass phrase stash file for the queue managers certificate
and private key store (Stash.sth)
The certificate revocation list file (optional - crl.pem)
Appendix F - Environment variables
Page 446: there are several environment variables that are used by the
Guardian sample build scripts to locate the header files and the
libraries. Suitable settings for these are established in the
WMQCSTM file (in the Guardian samples subvolume). The environment
variables, and their meanings, are:
MQNSKOPTPATH^INC^G include file/header subvolume
MQNSKOPTPATH^BIN^G binaries subvolume
MQNSKOPTPATH^LIB^G binaries subvolume
MQNSKOPTPATH^SAMP^G samples subvolume
In addition, an HP environment variable is also required (and set in
WMQCSTM) that locates the OSS DLLs for dynamic loading from Guardian.
The environment variable is ^RLD^FIRST^LIB^PATH.
Page 468: add after the "Queue Server Tuning parameters" section
Queue Manager Server tuning parameters
MQQMSHKEEP If this ENV is set for the MQS-QMGRSVR00 serverclass, its value
specifies a numeric value in seconds to override the default housekeeping
interval of the queue manager server. The default interval is 60 seconds.
The housekeeping interval controls how frequently the queue manager
server generates expiration reports. The permitted range of values is 1-300.
Values outside this range will be ignored and the default value will be used.
MQQMSMAXMSGSEXPIRE If this ENV is set for the MQS-QMGRSVR00 serverclass,
its value specifies a numeric value to override the default maximum number
of expiration report messages that are generated during housekeeping
operations by a queue manager server. The default maximum number of expiration
messages generated is 100. The permitted range of values is 1-99999. Values
outside this range will be ignored and the default value will be used.
Appendix H - Building and running applications
Building C++ applications,
Table 47 - there is no multi-threaded library support in
Guardian so there should not be an entry for a
multi-threaded guardian library
Table 48 - the name of this table should be "Native non-PIC"
References to G/lib symbolic links have changed with WMQ 5.3.1 to lib/G
Note that the MQNSKVARPATH and MQNSKOPTPATH environment variables must
be established in the environment, before an application starts up.
They cannot be programmatically set once a program is running by using
putenv().
Page 461: Building COBOL applications
Add the following text:
"In both the OSS and Guardian environment, the CONSULT compiler
directive referencing the MQMCB import library must now be used along
with correct linker options. Refer to the BCOBSAMP TACL script described
in Appendix I for more information."
Appendix I - WebSphere MQ for NonStop Server Sample Programs
Pages 465-466: The section "TACL Macro file for building C Sample Programs"
is replaced by the following:
BCSAMP - Build a C-Language Sample.
This TACL script will compile and link a C-language sample into an
executable program. The script expects that the WMQ environment has
been established using WMQCSTM.
BCSAMP usage:
BCSAMP <type> <source>
<type> The type of executable program that should be built.
Valid values are:
pic A native PIC program
tns A non-native TNS program
<source> The filename of the source module to be compiled and linked
The <source> filename should end with a 'C'. The final program name is
the same as the source filename with the trailing 'C' removed.
Page 467: The section "TACL Macro files for building COBOL Sample
Programs" is replaced by the following:
BCOBSAMP - Build a COBOL Sample.
This TACL script will compile and link a COBOL sample into an executable
program. The script expects that the WMQ environment has been established
using WMQCSTM.
BCOBSAMP usage:
BCOBSAMP <type> <source>
<type> The type of executable program that should be built.
Valid values are:
pic A native PIC program
tns A non-native TNS program
<source> The filename of the source module to be compiled and linked
The <source> filename should end with an 'L'. The final program name is
the same as the <source> filename with the trailing 'L' removed.
Page 469: The section "TACL Macro files for building TAL sample programs"
is replaced by the following:
BTALSAMP - Build a TAL Sample.
This TACL script will compile and link a TAL sample into an executable
program. The script expects that the WMQ environment has been established
using WMQCSTM.
BTALSAMP usage:
BTALSAMP <source>
<source> The filename of the source module to be compiled and linked
The final program name is the same as the <source> filename with the
trailing character removed.
Appendix J - User exits
refer to the documentation supplement Exits.pdf for updated information
about configuring and building user exits. This supplement has been
extensively revised for V5.3.1.1 to clarify the requirements and process
for creating and integrating exits with WebSphere MQ.
The description of compile options for PIC unthreaded, threaded and
Guardian dlls in this document is incorrect. The option specified as
"-export all" should be "-export_all".
Appendix K - Setting up communications
Page 482: The TCP/IP keep alive function
By default, the TCP/IP keep alive function is not enabled. To enable
this feature, set the KeepAlive=Yes attribute in the TCP Stanza in the
qm.ini file for the queue manager.
If this attribute it set to "yes" the TCP/IP subsystem checks periodically
whether the remote end of a TCP/IP connection is still available. If it is
not available, the channel using the connection ends.
If TCP stanza KeepAlive attribute is not present or is set to "No", the
TCP/IP subsystem will not check for disconnection of the remote end.
Chapter 9 "Configuring WebSphere MQ" page 140 describes the TCP stanza
attributes.
APAR IC58859: wmqtrig script
The wmqtrig script processing the -c option, for triggering a TACL macro/script
file, will not normally propagate the TMC data to the macro/script file.
Some applications may need the TMC for processing. A switch used in conjunction
with the -c option, -5.1, has been added which will pass the TMC data to a TACL
macro/script file with the wmqtrig script. Define the APPLICID attribute with
the -5.1 switch, for example:
APPLICID(/wmq/opt/mqm/bin/wmqtrig -5.1 -c \$data06.test.trigmac)
SSLupdate.pdf page 7
-----------------------------------------------------------------------------
The SSLupdate.pdf document was first released with Fixpack 5.3.1.1
The SSL test scripts expect that a default TCP/IP process ($ZTC0) is
configured on the system to be used during the test. The configuration
will need modification if a non-default TCP/IP process does not exist
on the system or another TCP/IP process is used to communicate with the
partner system. The ALICE.sh and/or BOB.sh scripts that setup of the
listener (runmqlsr) will need modification to add the -g option to use a
non-default TCP/IP process.
CONTACTING IBM SOFTWARE SUPPORT
===============================
IBM Software Support provides assistance with product defects. You might
be able to solve you own problem without having to contact IBM Software
Support. The WebSphere MQ Support Web page
(http://www.ibm.com/software/integration/wmq/support/) contains
links to a variety of self-help information and technical flashes. The
MustGather Web page
(http://www-01.ibm.com/support/docview.wss?uid=swg21229861)
contains diagnostic hints and tips that will aid in diagnosing and
solving problems, as well of details of the documentation required by
the WebSphere MQ support teams to diagnose problems.
Before you "Submit your problem" to IBM Software Support, ensure
that your company has an active IBM software maintenance contract, and
that you are authorized to submit problems to IBM. The type of software
maintenance contract that you need depends on the type of product you
have:
For IBM distributed software products (including, but not limited to,
Tivoli(R), Lotus(R), and Rational(R) products, as well as DB2(R) and
WebSphere products that run on Windows or UNIX(R) operating systems),
enroll in Passport Advantage(R) in one of the following ways:
Online: Go to the Passport Advantage Web site at
http://www.lotus.com/services/passport.nsf/WebDocs/Passport_Advantage_Home,
and click "How to Enroll".
By phone: For the phone number to call in your country, go to the "Contacts"
page of the IBM Software Support Handbook at
www.ibm.com/support/handbook, click 'contacts' and then click the name of
your geographic region.
For customers with Subscription and Support (S & S) contracts, go to the
Software Service Request Web site at
http://www.ibm.com/support/servicerequest.
For customers with IBMLink(TM), CATIA, Linux(R), S/390(R), iSeries(TM),
pSeries(R), zSeries(R), and other support agreements, go to the IBM Support
Line Web site at
http://www.ibm.com/services/us/index.wss/so/its/a1000030/dt006.
For IBM eServer(TM)) software products (including, but not limited to,
DB2(R) and WebSphere products that run in zSeries, pSeries, and iSeries
environments), you can purchase a software maintenance agreement by working
directly with an IBM sales representative or an IBM Business Partner.
For more information about support for eServer software products, go to the
IBM Technical Support Advantage Web site at
http://www.ibm.com/servers/eserver/techsupport.html.
If you are not sure what type of software maintenance contract you need,
call 1-800-IBMSERV (1-800-426-7378) in the United States. From other
countries, go to the "Contacts" page of the IBM Software Support
Handbook at www.ibm.com/support/handbook, click 'contacts' and then
click the name of your geographic region. for phone numbers of people
who provide support for your location.
To contact IBM Software support, follow these steps:
1. "Determine the business impact of your problem"
2. "Describe your problem and gather background information"
3. "Submit your problem"
Determine the business impact of your problem
When you report a problem to IBM, you are asked to supply a severity
level. Therefore, you need to understand and assess the business impact
of the problem that you are reporting. Use the following criteria:
---------------------------------------------------------------+
Severity 1 The problem has a critical
business impact: You are unable
to use the program, resulting in
a critical impact on operations.
This condition requires an
immediate solution.
---------------------------------------------------------------+
Severity 2 This problem has a significant
business impact: The program is
usable, but it is severely
limited.
---------------------------------------------------------------+
Severity 3 The problem has some business
impact: The program is usable,
but less significant features
(not critical to operations) are
unavailable.
---------------------------------------------------------------+
Severity 4 The problem has minimal business
impact: The problem causes
little impact on operations, or
a reasonable circumvention to
the problem was implemented.
---------------------------------------------------------------+
Describe your problem and gather background information
When describing a problem to IBM, be as specific as possible. Include
all relevant background information so that IBM Software Support
specialists can help you solve the problem efficiently. See the
MustGather Web page
http://www-01.ibm.com/support/docview.wss?uid=swg21229861 for
details of the documentation required. To save time, know the answers to
these questions:
What software versions were you running when the problem occurred?
Do you have logs, traces, and messages that are related to the problem
symptoms? IBM Software Support is likely to ask for this information.
Can you re-create the problem? If so, what steps do you perform to
re-create the problem? Did you make any changes to the system? For example,
did you make changes to the hardware, operating system, networking software,
or other system components? Are you currently using a workaround for the
problem? If so, please be prepared to describe the workaround when you
report the problem.
Submit your problem
You can submit your problem to IBM Software Support in one of two ways:
Online: Go to the Submit and track problems tab on the IBM Software Support
site at http://www.ibm.com/software/support/probsub.html. Type your
information into the appropriate problem submission tool.
By phone: For the phone number to call in your country, go to the "Contacts"
page of the IBM Software Support Handbook at
www.ibm.com/support/handbook, click 'contacts' and then click the name
of your geographic region.
If the problem you submit is for a software defect or for missing or
inaccurate documentation, IBM Software Support creates an Authorized
Program Analysis Report (APAR). The APAR describes the problem in
detail. Whenever possible, IBM Software Support provides a workaround
that you can implement until the APAR is resolved and a fix is
delivered. IBM publishes resolved APARs on the Software Support Web site
daily, so that other users who experience the same problem can benefit
from the same resolution.
NOTICES AND TRADEMARKS
======================
IBM may not offer the products, services, or features discussed in this
document in all countries. Consult your local IBM representative for
information on the products and services currently available in your
area. Any reference to an IBM product, program, or service is not
intended to state or imply that only that IBM product, program, or
service may be used. Any functionally equivalent product, program, or
service that does not infringe any IBM intellectual property right may
be used instead. However, it is the user's responsibility to evaluate
and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject
matter described in this document. The furnishing of this document does
not give you any license to these patents. You can send license
inquiries, in writing, to:
IBM Director of Licensing
IBM Corporation
North Castle Drive
Armonk, NY 10504-1785
U.S.A.
For license inquiries regarding double-byte (DBCS) information, contact
the IBM Intellectual Property Department in your country/region or send
inquiries, in writing, to:
IBM World Trade Asia Corporation
Licensing
2-31 Roppongi 3-chome, Minato-ku
Tokyo 106, Japan
The following paragraph does not apply to the United Kingdom or any
other country/region where such provisions are inconsistent with local
law: INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS
PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
Some states do not allow disclaimer of express or implied warranties in
certain transactions; therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical
errors. Changes are periodically made to the information herein; these
changes will be incorporated in new editions of the publication. IBM may
make improvements and/or changes in the product(s) and/or the program(s)
described in this publication at any time without notice.
Any references in this information to non-IBM Web sites are provided for
convenience only and do not in any manner serve as an endorsement of
those Web sites. The materials at those Web sites are not part of the
materials for this IBM product, and use of those Web sites is at your
own risk.
IBM may use or distribute any of the information you supply in any way
it believes appropriate without incurring any obligation to you.
Licensees of this program who wish to have information about it for the
purpose of enabling: the exchange of information between
independently created programs and other programs (including this one)
and (ii) the mutual use of the information that has been exchanged,
should contact:
IBM Canada Limited
Office of the Lab Director
8200 Warden Avenue
Markham, Ontario
L6G 1C7
CANADA
Such information may be available, subject to appropriate terms and
conditions, including in some cases payment of a fee.
The licensed program described in this document and all licensed
material available for it are provided by IBM under terms of the IBM
Customer Agreement, IBM International Program License Agreement, or any
equivalent agreement between us.
Any performance data contained herein was determined in a controlled
environment. Therefore, the results obtained in other operating
environments may vary significantly. Some measurements may have been
made on development-level systems, and there is no guarantee that these
measurements will be the same on generally available systems.
Furthermore, some measurements may have been estimated through
extrapolation. Actual results may vary. Users of this document should
verify the applicable data for their specific environment.
Information concerning non-IBM products was obtained from the suppliers
of those products, their published announcements, or other publicly
available sources. IBM has not tested those products and cannot confirm
the accuracy of performance, compatibility, or any other claims related
to non-IBM products. Questions on the capabilities of non-IBM products
should be addressed to the suppliers of those products.
All statements regarding IBM's future direction or intent are subject to
change or withdrawal without notice, and represent goals and objectives
only.
This information may contain examples of data and reports used in daily
business operations. To illustrate them as completely as possible, the
examples include the names of individuals, companies, brands, and
products. All of these names are fictitious, and any similarity to the
names and addresses used by an actual business enterprise is entirely
coincidental.
This information may contain sample application programs, in source
language, which illustrate programming techniques on various operating
platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM for the purposes of developing, using,
marketing, or distributing application programs conforming to the
application programming interface for the operating platform for which
the sample programs are written. These examples have not been thoroughly
tested under all conditions. IBM, therefore, cannot guarantee or imply
reliability, serviceability, or function of these programs.
Trademarks
The following terms are trademarks of International Business
Machines Corporation in the United States, other countries,
or both: DB2, eServer, IBM IBMLink, iSeries, Lotus, MQSeries, pSeries,
Passport Advantage, Rational, s/390, SupportPac, Tivoli, WebSphere, zSeries.
UNIX is a registered trademark of The Open Group in the United States
and other countries.
Microsoft Windows is a trademark or registered trademark of Microsoft
Corporation in the United States, other countries, or both.
Java and all Java-based trademarks and logos are trademarks or registered
trademarks of Sun Microsystems, Inc. in the United States, other countries,
or both.
Linux is a trademark of Linus Torvalds in the United States, other
countries, or both.
Other company, product or service names may be the trademarks
or service marks of others
Was this topic helpful?
Document Information
Modified date:
06 April 2021
UID
ibm16440307