IBM Support

OA57508: NEW FUNCTION: - ZFS HIGH AVAILABILITY SUPPORT

A fix is available

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as new function.

Error description

  • This apar will provide new zFS High Availability support.
    

Local fix

Problem summary

  • ****************************************************************
    * USERS AFFECTED:                                              *
    * zFS Version 2 Release 3                                      *
    * zFS FMID:HZFS430                                             *
    *                                                              *
    * zFS Version 2 Release 4                                      *
    * zFS FMID:HZFS440                                             *
    ****************************************************************
    * PROBLEM DESCRIPTION:                                         *
    * APAR OA57508 enables the high                                *
    * availability function on V2R3 and V2R4                       *
    * systems.  With the PTF applied on all                        *
    * V2R3 and V2R4 systems, you can take                          *
    * advantage of the zFS high availability                       *
    * support on those systems.                                    *
    ****************************************************************
    * RECOMMENDATION:                                              *
    * APPLY PTF                                                    *
    ****************************************************************
    With the zFS high availability option, if the file system owner
    experiences an outage, applications that are accessing file
    systems on other systems are not affected. You can use a
    mount parameter or the IOEFSPRM option to designate a zFS
    sysplex-aware file system as high availability. To take
     advantage of the zFS high availability support, apply the PTF
    for APAR OA57508 on all V2R3 and V2R4 systems. For
    read/write sysplex-aware file systems, consider whether they
    should be high availability. If you are concerned about
    application availability after a system experiences an outage,
    consider using the high availability option.
    

Problem conclusion

  • APAR OA57508 enables the high availability function on V2R3 and
    V2R4 systems.  With the PTF applied on all V2R3 and V2R4
    systems, you can take advantage of the zFS high availability
    support on those systems.
    
    Updates are provided for the the following books :
    
          z/OS Version 2 Release 3 Distributed File System
    Administration SC23-6887-30
          z/OS Version 2 Release 3 Distributed File Service Messages
    and Codes  SC23-6885-30
          z/OS Version 2 Release 4  File System Administration
    SC23-6887-40
    
    
    The following information is being added to the z/OS
    Version 2 Release 3 Distributed File System Administration
    book:
    
    The  z/OS Version 2 Release 3 Distributed File System
    Administration and Messages and Code updates can
    also be found at the following url:
    
    http://publibz.boulder.ibm.com/zoslib/pdf/OA57508.pdf
    
    In the Summary of changes , the following is being added:
    
    The high availability option
    
    With the zFS high availability option, if the file system owner
    experiences an outage, applications that are accessing file
    systems on other systems are not affected. You can use a
    mount parameter or the IOEFSPRM option to designate
    a zFS sysplex-aware file system as high availability. To
    take advantage of the zFS high availability support, apply
    the PTF for APAR OA57508 on all V2R3 and V2R4 systems.
    
    For read/write sysplex-aware file systems, consider
    whether they should be high availability. If you are
    concerned about application availability after a system
    experiences an outage, consider using the high
    availability option.
    
    In Chapter 4. Creating and managing zFS file systems using
    compatibility mode aggregates
    
    The following information  is being added in Changing zFS
    attributes on a mounted zFS compatibility mode file system:
    
    zFS attributes are assigned to a zFS compatibility mode file
    system when it is mounted. The attributes can be set by
    specifying a zFS MOUNT parameter or they can be set from
    the zFS default values of the system where the primary
    mount occurs. These attributes, which are generally only
    meaningful for read/write mounted file systems, include
    HA.
    
    You can query the current default value of a zFS attribute by
    issuing the zfsadm configquery command. For example, to
    query the default value of the following attributes, issue the
    following command:
    
    zfsadm configquery -ha
    
    Also, as the following example shows, you can change the
    zFS default values by issuing the zfsadm config command:
    
    zfsadm config -ha on
    
    Specifying the high availability option for read/write
    sysplex-aware file systems With the zFS high
    availability option, if the file system owner
    experiences an outage, applications from other
    systems that are accessing that particular file
    system are not affected. File systems such as
    WebSphere Application Server and CICS that
    do not make frequent directory operations
    from non-owning systems might find the
    high availability option useful.
    
    To designate a zFS file system as high availability,
    you have two choices:
    
    - Specify the HA option in the MOUNT parameter of the
    BPXPRMxx parmlib member.
    
    -  Use the HA=ON option in the IOEFSPRM file.
    
    To dynamically enable or disable the high availability
    function for a file system, use the zfsadm chaggr command.
    
    Important:  Do not use the HA option or have the IOEFSPRM
    option default to ON unless all sysplex members have the
    PTF for APAR OA57508 applied.
    
    Usage notes:
    
    1. You cannot use the high availability option if there
    are systems in the sysplex that do not have the PTF
    for APAR OA57508 applied.  Even if the IOEFSPRM HA
    option is set so that high availability is enabled by
    default or if HA is specified in the mount parameter,
    zFS will still consider the file system to be non-high
    availability.
    
    2. If a high availability file system is mounted in the sysplex,
    
    systems in the sysplex that do not have the PTF for APAR
    OA57508 applied will not be allowed to initialize. If those
    systems must be initialized, you will have to either unmount
    the high availability file systems or use the zfsadm chaggr
    command to remove the high availability option from the
    file systems. The zfsadm fsinfo command can be used to
    determine which file systems are mounted with the high
    availability option.
    
    3. In a high availability file system, applications that are
    creating or updating files will not see errors if the owning
    system goes down. However, if the application is working
    with a FIFO special file, it will see errors if the owning
    system goes down.
    
    4. To ensure that applications do not receive errors, use the
    high availability option when you are mounting any file
    system that is accessed by applications in a parallel sysplex.
    Also, include the file systems that are at a higher level in the
    
    mount tree. The high availability option is not needed for file
    systems that are mounted read-only.
    
    5. The high availability option is not needed for file systems
    that are mounted with the noautomove and unmount
    automove options. Those file systems cannot be accessed
    by other systems if the owning system is terminated.
    
    6. If the high availability option is used, disk synchronization
    
    of the internal log file in the file system occurs more often.
     An increase in the response time of metadata (any file system
    data that is not the contents of user files) operations for
    requests from sysplex client systems will occur. Applications
    on the owning system do not need to synchronize the log file.
    However, because the log file is shared, workloads that
    frequently update the directories from non-owning systems
    of the file system will experience slower performance on
    both the client systems and the owning system. If
    performance is critical, the high availability option might
    not be a good choice.
    
    In Chapter 11. zFS commands
    
    The MOUNT command section is being updated to include:
    
    The HA | NOHA options were added to the MOUNT command.
    
    HA | NOHA
    Specifies whether the system will provide high availability for
    applications on non-owning systems for a sysplex-aware file
    system when the owning system experiences an outage.
    The default is the HA specification in the IOEFSPRM file.
    
    Using the HA mount option increases the directory
    response time from non-owning systems.
    
    The zfsadm chaggr command section is being updated to include:
    
    zfsadm chaggr
    
    The -ha | -noha options were added to the zfsadm chaggr command.
    
    -ha | -noha
    
    Specifies whether the aggregate should provide the high
    availability.
    The -ha and -noha options will cause a samemode remount to be
    issued if the aggregate is mounted read/write. If the aggregate
    is mounted read-only, only the mount parameters are updated.
    
    The -ha and -noha options are mutually exclusive.
    
    The zfsadm chaggr command section is being updated to include:
    
    
    zfsadm config
    
    The -ha option was added to the zfsadm config command.
    
    -ha ON | OFF
    Specifies whether the high availability option is enabled by
    default for mounts of sysplex-aware file
    systems.
    
    The zfsadm configquery command section is being
    updated to include:
    
    zfsadm configquery
    The -ha option was added to the zfsadm configquery command.
    
    -ha
    Displays whether the high availability option is enabled by
    default for mounts of sysplex-aware file systems.
    
    The zfsadm fsinfo command section is being updated to include:
    
    zfsadm fsinfo
    
    Aggregates can be selected by use of the -select option. HA and
    NOHA were added to the list of criteria for selecting
    aggregates.
    
      HA shows aggregates that are mounted with the high
       availability option.
    
       NOHA shows aggregates that are not mounted with
       the high availability option.
    
    When you use the -owner option, the displayed information
    has the file system status as part of the output.
    
      HA indicates that the aggregate is mounted with
      the high availability option.
    
    In Chapter 12. The zFS configuration options file (IOEPRMxx or
    IOEFSPRM). the following information is being added:
    
    IOEFSPRM
    
    HA
    Specifies whether high availability is enabled by default for
    mounts of sysplex-aware file systems.
    
    The expected value is ON or OFF. For example:
    
    HA = ON
    
    The default value is OFF.
    
    In Chapter 13. zFS application programming interface
    information, the following information is being added:
    
    pfsctl (BPX1PCT)
    
    Two subcommands (opcodes) were added: Query HA and
    Set HA.
    
    For           Command               Subcommands(opcodes)
    Configuration  ZFSCALL_CONFIG     Query HA (269)
                                      (0x40000006)
    
    For           Command               Subcommands(opcodes)
    Configuration    ZFSCALL_CONFIG    Set HA (268)
                                        (0x40000006)
    
    
    
    
    Change Aggregate Attributes
    The ch_ha_flags attribute was added.
    ch_ha_flags     char      Indicates if aggregate should be high
    availability.
    
      1 - File system should be high availability.
      2 - File system should not be high availability.
    
    List Detailed File System Information
    
    For fr_selection, these lines were added:
    x20000000 - Show aggregates that are high availability.
    x40000000 - Show aggregates that are not high availability.
    
    In the z/OS Version 2 Release 3 Distributed File Service
    Messages and Codes  (SC23-6885-30), the following
    messages are being added:
    
    IOEZ01028E Unexpected errors during takeover processing
                         for HA file system filesystem_name with
                        new  owner system_name.
    
    Explanation
    zFS encountered unexpected errors while resolving
    pending operations that were in progress when the
    original owner of the file system went down. This
    means the applications that requested those
    operations will receive EIO (122) errors from zFS.
    
    In the message text:
    filesystem_name
        The name of the file system.
    system_name
        The name of the system.
    System action
        The program continues.
    Severity
        svc_c_sev_error
    
    Administrator Response
    Contact your service representative.
    
    
    IOEZ01029E The HA attribute is not assigned to file system
                         filesystem_name because system system_name
                         does not support high availability file
                         systems.
    
    Explanation
    The HA attribute was to be assigned to the specified file
    system either because the HA mount parameter
    was specified or the IOEFSPRM configuration option
    HA=ON is specified. The HA attribute can only be used
    when all members of a sysplex support high availability
    file systems.
    
    In the message text:
    file_system
        The name of the file system.
    system_name
        The name of the system that does not support high
    availability file systems.
    System action
        The program continues.
    Severity
        svc_c_sev_error
    
    
    Administrator Response
    To use the HA file system attribute, ensure that all members
    in the sysplex have OA57508 applied. Then you can enable
    the HA attribute for this file system by using the zfsadm
    chaggr command with the -ha option. Alternatively, you
    could unmount the file system, and mount it again with
    the HA mount parameter. Or, you can issue the zfsadm
    config command with -ha on specified, then unmount
    and mount the file system again.
    
    
    IOEZ01030E Initializing system system_name will not be allowed
    to join the sysplex because it does not support high
    availability
    file systems.
    
    Explanation
    There is at least one high availability (HA) mounted file system
    
    in the sysplex. HA support requires that all members of a
    sysplex support HA.
    
    In the message text:
    system_name
    Name of the initializing system that does not support HA.
    
    System action
    The program continues.
    
    Severity
    svc_c_sev_warning
    
    Administrator Response
    zFS on the initializing system will terminate. To correct this
    problem, either unmount or change the HA status of all
    file systems that are currently mounted HA. The zfsadm
    chaggr command can be used to change the HA status
    of a file system.
    
    The following reason codes have been added:
    
    Reason codes
    
    
    
    6CA0 A sysplex client without high availability mount
    support (HA) is trying to connect to an HA-mounted
    file system.
    
    
    Action:
    Retry the operation when the client system is at a level
    that supports high availability mount.
    
    6CA2 An invalid value was specified for the -ha
    aggregate attribute.
    
    Action:
    The -ha aggregate attribute takes a value of either ON or
    OFF. When using the Change Aggregate Attributes API,
    use the corresponding value, which is either 1 (for ON)
    or 2 (for OFF).
    
    
    6CA3 The aggregate attribute cannot be changed to both
    -ha and -noha.
    
    Action:
    Limit your request to one attribute value and try again.
    
    6CA5 The HA aggregate attribute cannot be used in a
    sysplex where some systems do not support high
     availability file systems.
    
    Action:
    Ensure that APAR OA57508 is applied to all systems
    in the sysplex and try the command again.
    
    
    The following information is being updated  in z/OS
    Version 2 Release 4  File System Administration SC23-6887-40:
    
     In Chapter 5. Using zFS in a shared file system
     environment in the section:  Specifying the high
     availability option for read/write
    sysplex-aware filesystems
    
    The following is being updated:
    
    High availability option for read/write sysplex-aware file
    systems
    With the zFS high availability option, if the file system owner
    experiences an outage, applications from other systems that
    are accessing that particular file system are not affected. File
    
    systems such as WebSphere Application Server and CICS that do
    not make frequent directory operations from non-owning
    systems might find the high availability option useful.
    
    To designate a zFS file system as high availability, you have
    two choices:
    Specify the HA option in the MOUNT parameter of the
    BPXPRMxx parmlib member. See zfsmount - Mount a file
    system into the z/OS UNIX hierarchy MOUNT.
    
    Use the HA=ON option in the IOEFSPRM file. See IOEFSPRM.
    To dynamically enable or disable the high availability function
    for a file system, use the zfsadm chaggr command. See
    zfsadmchaggr - Changes the attributes of an aggregatezfsadm
    chaggr.
    
    Important: Do not use the HA option or have the IOEFSPRM
    option default to ON unless all sysplex members are at the
    2.4 level or at the 2.3 level with APAR OA57508 applied.
    
    Usage notes:
    1. You cannot use the high availability option if there are
    systems in the sysplex that are not at the 2.4 level or at the
    2.3 level with APAR OA57508 applied. Even if the IOEFSPRM
    HA option is set so that high availability is enabled by
    default and if HA is specified in the mount parameter, zFS
    will still consider the file system to be non-high availability.
    
    2. If a high availability file system is mounted in the
    sysplex, be aware that systems in the sysplex that are not
    at the 2.4 level or at the 2.3 level with APAR OA57508
    applied are not allowed to initialize. If those systems must
    be initialized, you will have to either unmount the high
    availability file systems or use the zfsadm chaggr
    command to remove the high availability option from
    the file systems. The zfsadm fsinfo command can be used
    to determine which file systems are mounted with the high
    availability option.
    
    3. In a high availability file system, applications that are
    creating or updating files will not see errors if the owning
    system goes down. However, if the application is working
    with a FIFO special file, it will see errors if the owning
    system goes down.
    
    4. To ensure that applications do not receive errors, use
    the high availability option when you are mounting any file
    system that is accessed by applications in a parallel sysplex.
    Also, include the file systems in a higher level in the mount
    tree. The high availability option is not needed for file
    systems that are mounted read-only.
    
    5. The high availability option is not needed for file systems
    that are mounted with the noautomove and unmount
    automove options. Those file systems are not accessed by
    other systems if the owning system is terminated.
    
    6. If the high availability option is used, disk
    synchronization of the internal log file in the file system
    occurs more often. An increase in the response time of
    metadata (any file system data that is not the contents
    of user files) operations for requests from sysplex client
    systems will occur. Applications on the owning system
    do not need to synchronize the log file. However,
    because the log file is shared, workloads that frequently
    update the directories from non-owning systems of the
    file system will experience slower performance on both
    the client systems and the owning system. If
    performance is critical, the high availability option might
    not be a good choice.
    
    After the activation of OA57508, the following
    messages may be generated:
    IOEZ01028E
    IOEZ01029E
    IOEZ01030E
    

Temporary fix

Comments

APAR Information

  • APAR number

    OA57508

  • Reported component name

    ZFS ZSERIES FIL

  • Reported component ID

    5696EFS00

  • Reported release

    440

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    NoHIPER

  • Special Attention

    YesSpecatt / New Function / Xsystem

  • Submitted date

    2019-05-07

  • Closed date

    2019-12-12

  • Last modified date

    2020-01-03

  • APAR is sysrouted FROM one or more of the following:

  • APAR is sysrouted TO one or more of the following:

    UJ01635 UJ01636

Modules/Macros

  • IOEFSUTL IOEZHADD IOEZHMSG IOEZADM  IOEDUMPF IOEFSCM  IOEFSTHD
    IOEZHCK1 IOEFSKN  IOEAGFMT IOEAGSLV
    

Publications Referenced
SC23688730SC23688740SC23688530  

Fix information

  • Fixed component name

    ZFS ZSERIES FIL

  • Fixed component ID

    5696EFS00

Applicable component levels

  • R430 PSY UJ01635

       UP19/12/18 P F912  

  • R440 PSY UJ01636

       UP19/12/18 P F912  

Fix is available

  • Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG19M","label":"APARs - z\/OS environment"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"440","Edition":"","Line of Business":{"code":"","label":""}},{"Business Unit":{"code":null,"label":null},"Product":{"code":"SG19O","label":"APARs - MVS environment"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"440","Edition":"","Line of Business":{"code":"","label":""}}]

Document Information

Modified date:
03 January 2020