ADSM-L

Re: recreate bitvectors

2001-05-03 14:13:00
Subject: Re: recreate bitvectors
From: David_Bohm AT TIVOLI DOT COM
Date: Thu, 3 May 2001 11:13:41 -0700
The APAR actually applies to all TSM server platforms.  For the MVS server
a ZAP
is available.  Although the problem is in common code, the symptom does not
occur on the Windows platform.  The MVS and the Solaris servers will hang.
The
AIX server will crash.  I am not sure what the HP server will do.  On the
Windows server
the command will complete successfully.  The original fix was tested on the
Windows
platform which is why this problem was not noticed.  This has now also been
tested
on MVS since the ZAP was tested there.  This problem will be fixed in the
next
fixtest level.

I am sure that this APAR is the problem even if you are not running MVS
since the
problem is in common code.  Although it is common code the symptom on the
other platforms are different because of the interaction between the common
code
and the platform specific code.

My tests recreated the bitvector in a matter of seconds, although my disk
storage pool
volumes were small.  If the volumes are large and contain a lot of data I
would expect
the time to take something in the range of a few hours at the most, not
days.  This
assumes that the server is very busy with processing.  Until the new
fixtests are
available, this command will only work on the Windows server or with the
ZAP on
the MVS server.

David Bohm
TSM server development
email - dbohm AT tivoli DOT com


Bill Colwell <bcolwell AT DRAPER DOT COM>@VM.MARIST.EDU> on 05/03/2001 07:09:36 
AM

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

Sent by:  "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>


To:   ADSM-L AT VM.MARIST DOT EDU
cc:
Subject:  Re: recreate bitvectors



Saud - if your server is os/390 then this apar is your problem.

Item PQ48420

  APAR Identifier ...... PQ48420       Last Changed..01/05/02
  HANG ACQUIRE MUTEX ISSUING RECREATE BITVECTORS  IC29878


  Symptom ...... WS WAIT              Status ........... OPEN
  Severity ................... 1      Date Closed .........
  Component .......... 5697TSMVS      Duplicate of ........
  Reported Release ......... 371      Fixed Release ............
  Component Name TIVOLI STR MGR       Special Notice
  Current Target Date ..01/05/02      Flags
  SCP ...................
  Platform ............

  Status Detail: CREATE - APAR solution creation has begun.

  PE PTF List:

  PTF List:


  Parent APAR:
  Child APAR list:


  ERROR DESCRIPTION:
  On MVS platform, issuing RECREATE BITVECTORS as part of
  recommended fix for IC29878, server hangs attempt to
  reaccess a mutex that it already has acquired.
  --
  Server will not respond to commands.



  LOCAL FIX:
  ZAP will be provided for MVS customers
  Following ZAP has been tested with 4132 and 3745 fixtest
  --
    NAME     DSMSERV DSUTIL$
    VERIFY   2186 05EF
    VERIFY   2190 478035A0
    REP      2186 0700
    REP      2190 47F035A0
  The customer will also get following message twice for each
  bitvector object that needs to be recreated.
  ANR9999D TMLOCK(734): Attempt to release unheld shared lock.
  These messages can be ignored.  The fix for those messages
  are already included in the GA code and in the next patch level.


In <20010503123404.B27654 AT auckland.ac DOT nz>, on 05/03/01
   at 10:09 AM, Suad Musovich <suad AT CCU1.AUCKLAND.AC DOT NZ> said:

>Has anyone run the special "recreate bitvectors" introduced in a patch
>recently?

>I have fruitlessly watch my system for 4 days waiting for the process
>to end but nothing seems to be happening(system has been sitting at idle)

>If so, how long did it take? how big was your database? etc...

>Suad
>--

--
--------------------------
--------------------------
Bill Colwell
Bill Colwell
C. S. Draper Lab
Cambridge, Ma.
bcolwell AT draper DOT com
--------------------------
=========================================================================
<Prev in Thread] Current Thread [Next in Thread>