From: Mike Knight, ISSC Aerospace, (314)234-5096, KNIGHTM at ISSCVM
Inet:                                             KNIGHTM AT VNET.IBM DOT COM
Subject: ADSM Recovery log growing

ADSM does the mirroring for both the log and database, primary + one copy.
Both the log and database mirrors are normal read and parallel write.
In the test last night, one backup was running, intermittently using a
  tape for large files.  Migration grabbed the same tape, the backup went
  into media wait (no data sent) until the migration ended 2 hours later.
  Then the backup sent the one large file to tape and resumed backing up to
  disk.  During media wait, the log got up to 280MB.
There was one backup and one migration running last night, nothing else.
Interestingly, the migration filled the first tape and started a second,
  the backup was then in media wait for the new tape.

         Mike (not ADSM support) Knight

On Sat, 6 Jul 1996 at 11:23, Dwight Cook asked:
>
>     Is AIX doing the mirroring or ADSM ?
>     and what about the DB is it mirrored ?  If ADSM mirrored are you
>     normal & parallel ? are there other tasks running that might not be
>     letting the commits happen ?
>
>Original query 07/05/96 at 15:33 by Mike Knight:
>>
>>I have noticed a condition where the ADSM recovery log keeps growing
>>when a client backup process is waiting for access to a tape being used
>>by a migration process.  The log's %Util goes up until either the migration
>>process releases the tape or the log hits 99.9% which causes the server to
>>perform "recovery log compression".  Then the log goes back to below 1%.
>>
>>I can recreate this regularly by backing up two large Solaris clients
>>from scratch.  Has anybody else seen this?  We have a similar, but less
>>well defined problem on one ADSM Version 1 server which causes the
>>server to hang and all sessions to fail.
>>
>>We are running ADSM/6000 2.1.0.7 on AIX 4.1.4 with collocation on.
>>Backups go to disk with files over 5MB direct to tape.  The recovery
>>log is 800MB on mirrored raw disks.  TXNGROUPMAX=40, MOVEBATCHSIZE=128,
>>MOVESIZETHRESH=100, TXNBYYELIMIT=25600, LOGMODE=NORMAL.