Summary RE: Legato timing out after patch kit 3

Cohen.Jessica_at_ic.gc.ca
Date: 05/27/04

  • Next message: chriss: "Ξασχενθε" 
    Date: Thu, 27 May 2004 17:04:31 -0400
To: tru64-unix-managers@ornl.gov

    There is a patch kit: T64KIT0021340-V51BB24-E-20040120 that supposedly
    fixes this problem, which is a TCP timeout issue and seems to match the
    symptoms I noticed.

    Thanks to David J. DeWolfe, Iain Barker, and mcaplin@miami.edu for letting
    me know about this patch kit.

    -----Original Message-----
    From: tru64-unix-managers-owner@ornl.gov
    [mailto:tru64-unix-managers-owner@ornl.gov]On Behalf Of Cohen, Jessica:
    DGRB
    Sent: Thursday, May 27, 2004 1:48 PM
    To: tru64-unix-managers@ornl.gov
    Subject: Legato timing out after patch kit 3

     We have two ES40's that were recently upgraded to 5.1b, and then to patch
    kit 3. The 5.1b upgrade went fine, and our backups that evening went
    smootlhly, but our backups since we applied patch kit 3 (which we needed for
    some of our Oracle products) have been quite erratic. When I tried testing
    them during the day they went fine, and it isn't consistent which file
    systems successfully save and which don't, although the smallers ones are
    more likely to.

    An example of the problem is shown below from the savegroup completion
    report:

    --- Unsuccessful Save Sets ---

    * dino:/clones/home 1 retry attempted
    * dino:/clones 1 retry attempted

    * enzo:/clones/home/dba/oracle/data/temp 1 retry attempted
    * enzo:/clones/home/dba/oracle/data/index 1 retry attempted

    --- Successful Save Sets ---

      dino: /clones/home/dba/oracle/data/index level=incr, 14 GB 00:26:19 53
    files
      dino: /clones/usr level=incr, 399 KB 00:00:25 11
    files
    * dino:/clones/var 1 retry attempted
      dino: /clones/var level=incr, 14 MB 00:00:46 59
    files
      dino: /clones/home/dba/oracle/data/temp level=incr, 7166 MB 00:13:23
    22 files
      dino: /clones/home/dba/oracle/product level=incr, 2404 MB 00:05:31 310
    files
      rye: index:dino level=9, 966 KB 00:00:03 17
    files

      enzo: /clones level=incr, 0 KB 00:00:05 0
    files
      enzo: /clones/usr level=incr, 12 MB 00:00:38 81
    files
      enzo: /clones/var level=incr, 33 MB 00:00:14 72
    files
      enzo: /clones/home level=incr, 0 KB 00:00:30 0
    files
      enzo: /clones/home/dba/oracle/product level=incr, 2872 MB 00:05:21 345
    files
      rye: index:enzo level=9, 363 KB 00:00:03 5
    files

    And then the cron outputs these types of errors:

    * enzo:/clones/home/dba/oracle/data/temp ! no output
    05/26/04 01:37:55 savegrp: enzo:/clones/home/dba/oracle/data/temp will retry
    1 more time(s)
    * dino:/clones/home ! no output
    05/26/04 01:46:52 savegrp: dino:/clones/home will retry 1 more time(s)
    * dino:/clones/var ! no output

    or in the case of this morning these types:

    * enzo:/clones/home/dba/oracle/data/index lost connection to server, exiting
    05/27/04 00:41:10 savegrp: enzo:/clones/home/dba/oracle/data/index will
    retry 1 more time(s)
    * dino:/clones/home/dba/oracle/data/index lost connection to server, exiting
    05/27/04 00:47:25 savegrp: dino:/clones/home/dba/oracle/data/index will
    retry 1 more time(s)
    * enzo:/clones/home/dba/oracle/data/temp lost connection to server, exiting

    I was running Legato 6.1.3, I upgraded yesterday to 7.1, though this did not
    solve the problem.
    I haven't ruled out a network thing, but the discrete changes between "when
    it last worked perfectly" to "when we had issues" is the change from
    unpatched 5.1b to installing patch kit 3, and adding the following
    Oracle-requested changes to sysconfigtab:
    ipc:
    shm_mni=256
    shm_seg=256
    ssm_threshold=0
    rdg:
    msg_size=32768
    max_objs=5120
    max_async_req=256
    max_sessions=5000
    rdg_max_auto_msg_wires=0
    rdg_auto_msg_wires=0
    rt:
    aio_task_max_num=8193
    vm:
    new_wire_method=0

    We aren't clustered, so I don't believe the rdg does anything other than
    keep the Oracle installation program from squawking.

    Anyone seen any similar issues, either networky or legato-y after patch 3?

  • Next message: chriss: "Ξασχενθε"

    Relevant Pages