Topic
  • 7 replies
  • Latest Post - ‏2007-06-21T21:35:21Z by SystemAdmin
SystemAdmin
SystemAdmin
2340 Posts

Pinned topic MQ Workflow -Deadlocks

‏2006-11-13T09:54:03Z |
Hi All,
Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.

I am getting deadlocks every now and then in our production workflow server. I am not able to trace out what causing these deadlocks.Can anyone please tell me the reason?.

Here is the deadlock deatils:
11/13/06 15:15:29 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:15:31 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:39:06 FMC10510I Execution server instance stopped.
11/13/06 15:39:07 FMC10500I Execution server instance started.
11/13/06 15:40:59 FMC10510I Execution server instance stopped.
11/13/06 15:40:59 FMC10500I Execution server instance started.
11/13/06 15:47:22 FMC10510I Execution server instance stopped.
11/13/06 15:47:22 FMC10500I Execution server instance started.
11/13/06 15:48:53 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:48:55 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:48:57 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:48:59 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:49:01 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:49:03 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:50:30 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 15:54:58 FMC10510I Execution server instance stopped.
11/13/06 15:54:59 FMC10500I Execution server instance started.
11/13/06 16:00:42 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:00:44 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:00:46 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:00:48 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:00:50 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:00:52 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.
11/13/06 16:26:49 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout. The message will be retried.

Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000

Thanks and regards,
MBCA.

Updated on 2007-06-21T21:35:21Z at 2007-06-21T21:35:21Z by SystemAdmin
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-14T06:09:58Z  
    christadimaib@gmail.com wrote:
    > Hi All,
    > Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.
    >
    > Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >
    > Thanks and regards,
    > MBCA.
    It seems you're missing some fixpacks, for MQWF, for DB/2 and for AIX.
    AIX should be ML11 (ML10 has 3 flavours, so you can't be shure which),
    For DB/2 V6.1 FP8 is recommendedand, I request to update to DB/2 V7.1
    Which Fixpacklevel WF hpuld be, I must take a look on my machines.
    Can take a day, cause I'm on filed.

    At all: Your System is out of date, so your without support if you need it.
    But: This behaviour (Deadlocks) is well documented in Fixpack-Readmes.

    You can find them on ftp.software.ibm.com and so on.

    HTH
    Frank
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-14T06:11:44Z  
    christadimaib@gmail.com wrote:
    > Hi All,
    > Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.
    >
    > Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >
    > Thanks and regards,
    > MBCA.
    It seems you're missing some fixpacks, for MQWF, for DB/2 and for AIX.
    AIX should be ML11 (ML10 has 3 flavours, so you can't be shure which),
    For DB/2 V6.1 FP8 is recommendedand, I request to update to DB/2 V7.1
    Which Fixpacklevel WF hpuld be, I must take a look on my machines.
    Can take a day, cause I'm on filed.

    At all: Your System is out of date, so your without support if you need it.
    But: This behaviour (Deadlocks) is well documented in Fixpack-Readmes.

    You can find them on ftp.software.ibm.com and so on.

    HTH
    Frank
    Frank Schletz wrote:

    > christadimaib@gmail.com wrote:
    >> Hi All,
    >> Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.
    >>
    >> Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >>
    >> Thanks and regards,
    >> MBCA.
    >
    >

    And sorry for the writing errors, I'm without coffee and
    searching my glasses ;-)
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-14T16:16:09Z  
    Hi,

    please make sure that you move to a supported environment (MQWF 3.5 at
    least)
    asap before going agead with your production system. Deadlock problems
    have
    almost completely disappeared in later MQWF releases, timeout problems are
    a
    matter of DB tuning, but should definitely not occur with 10 users -
    except that your
    application uses MQWF tables, too.

    Volker Hoss
    IBM WebSphere Process Server Development
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-14T19:37:08Z  
    christadimaib@gmail.com wrote:

    > Hi All,
    > Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.
    >
    > Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >

    If you can't upgrade that easy:

    I got a running production environment with

    AIX 4.3.3 ML9 (11 also works)
    DB/2 V7.1.0.40
    MQS WF 3.4.0.30
    MQSeries 5.3.0.5 (working with 5.1 and up)

    without any problems, no deathlocks.

    As I mentioned befor: You need fixes!
    Deadlocks fixed in former releases.
    Check readmes.

    HTH
    Frank

    PS: Hoss is rigth, there's no support
    But: It's not easy to move up to WF3.5, but
    easy to move up to 3.4.
    Think about it and take the easy way
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-15T07:18:49Z  
    christadimaib@gmail.com wrote:

    > Hi All,
    > Last week our workflow server went for production, right now only 10 users are allowed to access the server and load is very low.
    >
    > Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >

    If you can't upgrade that easy:

    I got a running production environment with

    AIX 4.3.3 ML9 (11 also works)
    DB/2 V7.1.0.40
    MQS WF 3.4.0.30
    MQSeries 5.3.0.5 (working with 5.1 and up)

    without any problems, no deathlocks.

    As I mentioned befor: You need fixes!
    Deadlocks fixed in former releases.
    Check readmes.

    HTH
    Frank

    PS: Hoss is rigth, there's no support
    But: It's not easy to move up to WF3.5, but
    easy to move up to 3.4.
    Think about it and take the easy way
    Frank,

    how do you get fixes/ServicePacs for unsupported versions?
    I don't know about DB2 and AIX but I doubt that you can do it for MQWF!
    And even if you have it, what if you run into problems when applying
    service?
    I discourage this way.

    Volker Hoss
    IBM WebSphere Process Server Development
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2006-11-15T19:09:57Z  
    Frank,

    how do you get fixes/ServicePacs for unsupported versions?
    I don't know about DB2 and AIX but I doubt that you can do it for MQWF!
    And even if you have it, what if you run into problems when applying
    service?
    I discourage this way.

    Volker Hoss
    IBM WebSphere Process Server Development
    hos@de.ibm.com wrote:

    > Frank,
    >
    > how do you get fixes/ServicePacs for unsupported versions?
    > I don't know about DB2 and AIX but I doubt that you can do it for MQWF!
    > And even if you have it, what if you run into problems when applying
    > service?
    > I discourage this way.

    Yes, you must. Cause you are IBM :-)

    What I say is, in my environment, there are no
    errors, which need support. Software without bugs
    which need to be fixed.
    Many customers think "never touch a running system"
    And I talk about version, which run!

    If customer is interessted in support, he must
    upgrade. But this isn't easy all times, so I
    say: First go up to a level without relevant
    bugs, then think about going to a higher version.

    Regards
    Frank
  • SystemAdmin
    SystemAdmin
    2340 Posts

    Re: MQ Workflow -Deadlocks

    ‏2007-06-21T21:35:21Z  
    christadimaib@gmail.com wrote:
    > Hi All,
    > Last week our workflow server went for production, right now only 10 users are allowed to access the
    > server and load is very low.
    >
    > I am getting deadlocks every now and then in our production workflow server. I am not able to trace out what
    > causing these deadlocks.Can anyone please tell me the reason?.
    >
    > Here is the deadlock deatils:
    > 11/13/06 15:39:07 FMC10500I Execution server instance started.
    > 11/13/06 15:54:59 FMC10500I Execution server instance started.
    > 11/13/06 16:00:42 FMC31100W The message ProgramFinished could not be processed because of a database deadlock or timeout.
    > The message will be retried.
    >
    > Server details: MQ Workflow version 3.3.2.0, AIX 4.3.3.0, & DB2 06.01.0000
    >
    > Thanks and regards,
    > MBCA.

    Re-Post 7months late.

    Found something about this error and want to document this here
    so everyone can found it if ever happend again.
    Maybe this information help someone somewhere in time.

    About: Warning FMC31100W in fmcsys.log
    sometimes followed by a FMC31130W

    If so, it could be because of some "old" FMCIMREF.BND package which
    is no longer used but still in database and responsible for errors.
    Also if runstats / rebind does not help take a look at "invalid" packages.

    Solution: Drop the package FMC31130W

    Comment: Often it is usefull to "clean up" the database and remove old and
    useless things (packages, indexes, whatever) which can cause
    unpredictable behavior.

    Remember: Always take a (restorable) backup!

    Yust for the files.