Loading ...
Sorry, an error occurred while loading the content.

RE: PeopleSoft DBA Forum Application server with multiple tuxedo queues.

Expand Messages
  • David Kurtz
    Why are there other differences in the commands line? Server 1 (pid 31842) has two -s parameters to specify the advertised services, server 23 (pid 17433)
    Message 1 of 3 , Aug 2, 2013
    • 0 Attachment
      Why are there other differences in the commands line?
      Server 1 (pid 31842) has two -s parameters to specify the advertised
      services, server 23 (pid 17433) doesn't.
      Is the CLOPT the same for both APPQ queues?

      -R and -m are not in the Tuxedo documentation for servopts at
      http://docs.oracle.com/cd/E13161_01/tuxedo/docs10gr3/rf5/rf5.html#wp1003290

      -R won't be recycle count because that is not a Tuxedo function, it doesn't
      appear in the UBB. It is something done by PeopleSoft to mitigate the
      effects of memory leaks in the app servers (the Tuxedo people from BEA were
      very critical of this in the early days of PeopleTools 7 when they saw it).

      PeopleSoft count services and when the process hits the recycle count the
      server shuts down and Tuxedo restarts it.
      I think the -R parameter appears when the server has handled a service. It
      is not in the command line, so the process description is updated by the
      process as it runs. I can't find to what it correlates - I'd guess
      something to do with the return to the JSL.
      I don't know what -m does either, but it always seems to be 0.


      regards
      _________________________
      David Kurtz
      tel: +44 (0)7771 760660
      mailto:david.kurtz@...




      -----Original Message-----
      From: psftdba@yahoogroups.com [mailto:psftdba@yahoogroups.com] On Behalf Of
      tc_at_120maple
      Sent: 02 August 2013 17:37
      To: psftdba@yahoogroups.com
      Subject: PeopleSoft DBA Forum Application server with multiple tuxedo
      queues.

      I have added a second tuxedo queue to my application servers (using DK's
      most excellent guides) and now I see interesting information for the two
      queues when I do a "ps -ef" command.

      Red Hat linux ES5
      ==================
      #ps -ef | grep PSAPPSR[V] | less
      ... (this is a partial list, there are 10 processes per queue)
      psuser 17433 1 0 Jul30 ? 00:01:45 PSAPPSRV -C dom=psprod_56602
      -g 99 -i 23 -u sf-psleg-032 -U
      /peoplesoft/sys/software/pt8.49/appserv/psprod/LOGS/TUXLOG -m 0 -R 20823 -o
      LOGS/PSAPPQ2.stdout -e LOGS/PSAPPQ2.stderr -- -C psappsrv.cfg -D psprod -S
      PSAPPSRV
      psuser 31842 1 0 Jul03 ? 00:03:30 PSAPPSRV -C dom=psprod_56602
      -g 99 -i 1 -u sf-psleg-032 -U
      /peoplesoft/sys/software/pt8.49/appserv/psprod/LOGS/TUXLOG -m 0 -o
      LOGS/PSAPPQ1.stdout -e LOGS/PSAPPQ1.stderr -s@../psappsrv.lst
      -s@../psqcksrv.lst -- -C psappsrv.cfg -D psprod -S PSAPPSRV
      ...
      ====================
      Spreading out the above, you will see two PSAPPSRV processes each that use a
      different tux queue. The interesting thing is that the one with pid 17433
      shows different parameters than the 31842 one.

      I suspect, but would like verification on two things:
      #1: The two queues (though they are load balanced: "LDBAL Y") do not both
      receive tasks equally because the first one has never been busy enough to
      queue any traffic. I suspect that, if the server was busy enough, the load
      would have gotten to the second queue.

      #2: The parameter "-R 20823" in pid 17433 indicates that this server has
      been restarted by the BBL - probably due to reaching the Recycle Count. I
      kind of can verify this by restarting the services on the server - this will
      then show all PSAPPSRV processes like the second pid: 31842.

      I'd like confirmation on the above, or better information, and I'd also like
      to be assured that the restarted server processes get the exact same
      parameters as the start-up ones.

      Thanks!
      -T



      ------------------------------------

      PeopleSoft for the Oracle DBA is published by Apress - see
      http://www.psftdba.com The PeopleSoft DBA Forum is managed by
      http://www.go-faster.co.uk Yahoo! Groups Links
    • tc_at_120maple
      Thanks for the reply. ... CLOPT is identical for both queues (the exception is the err & out log file names) ... I too could not find a definition for the -R
      Message 2 of 3 , Aug 2, 2013
      • 0 Attachment
        Thanks for the reply.

        > Is the CLOPT the same for both APPQ queues?
        >

        CLOPT is identical for both queues (the exception is the err & out log file names)

        >
        > -R and -m are not in the Tuxedo documentation for servopts at
        > http://docs.oracle.com/cd/E13161_01/tuxedo/docs10gr3/rf5/rf5.html#wp1003290

        I too could not find a definition for the "-R" option. I suspect that it represents a recycled process though.

        From the info I found (see below) I surmise that when the PSAPPSRV process gets restarted due to a recycle count being exceeded, the restarted process gets the "-R <old_pid>" option along with it. This may be just for documentation, or for internal use by tuxedo - not sure.

        From the following you can see that:

        In the APPSRV_0730.LOG file, the PSAPPSRV process with pid 20823 was recycled after 5K services at 07/30/13 09:52:33

        In the TUXLOG.073013 at 09:52:43 (right after 20823 recycles), the PSAPPSRV process 17433 is started.

        And, sure enough, in the ps -ef line for PSAPPSRV process 17433, we see the option "-R 20823" !


        ==================================================================
        From a grep through logs for 20823:
        ===================================
        APPSRV_0730.LOG:PSAPPSRV.20823 (5000) [07/30/13 09:52:33 CMP2@...](0) Recycling server after 5000 services

        From a grep through logs for 17433:
        ===================================
        TUXLOG.073013:095243.sf-psleg-032!PSAPPSRV.17433.4105078464.0: 07-30-2013: Tuxedo Version 9.1, 32-bit
        TUXLOG.073013:095243.sf-psleg-032!PSAPPSRV.17433.4105078464.0: LIBTUX_CAT:262: INFO: Standard main starting
        TUXLOG.073013:095244.sf-psleg-032!restartsrv.17432.4155403136.-2: server APPSRV/23: CMDTUX_CAT:580: INFO: A server process has restarted: 17433

        From a ps -ef :
        ===============
        psuser 17433 1 0 Jul30 ? 00:03:44 PSAPPSRV -C dom=psprod_56602 -g 99 -i 23 -u sf-psleg-032 -U /peoplesoft/sys/software/pt8.49/appserv/psprod/LOGS/TUXLOG -m 0 -R 20823 -o LOGS/PSAPPQ2.stdout -e LOGS/PSAPPQ2.stderr -- -C psappsrv.cfg -D psprod -S PSAPPSRV
        psuser 31842 1 0 Jul03 ? 00:03:35 PSAPPSRV -C dom=psprod_56602 -g 99 -i 1 -u sf-psleg-032 -U /peoplesoft/sys/software/pt8.49/appserv/psprod/LOGS/TUXLOG -m 0 -o LOGS/PSAPPQ1.stdout -e LOGS/PSAPPQ1.stderr -s@../psappsrv.lst -s@../psqcksrv.lst -- -C psappsrv.cfg -D psprod -S PSAPPSRV
        =================================================================

        So the newly restarted "command line" represented by the "ps -ef" command is different after the process gets recycled. The question
        is:

        Does it get the same command line opts? It seems to, as everything appears to continue to function ok. Perhaps the "-R <old_pid>" is a way that the new process inherits some of the old processes environment?

        Anyhow, this is what I am seeing. More info might be good to know.

        Thanks again,
        -T
      Your message has been successfully submitted and would be delivered to recipients shortly.