ADSM-L

LOG usage over 100% - backups failing -- update

2000-12-12 14:51:50
Subject: LOG usage over 100% - backups failing -- update
From: Vijay Havaralu <vhavaralu AT CHUBB DOT COM>
Date: Tue, 12 Dec 2000 14:52:10 -0500
Thanks to who all replied to this problem.

As answer to several question in replies,

we are taking adsm-db full backup daily, and it is not the cause of the
problem.
     Incremental backup is implemented 5 days a week, but it forces full
db-backup
     when the changes go above a certain number of updates.
we were not using dbbackup trigger, as backup taken once daily morning,
when backups
     are not run.
As per tsm-support we tried to increase logpoolsize from 2k to 8k and 16k
and it has no effect
     in the speed.
As per tsm-support we tried to use dbbackuptrigger, but it made things
worse at night, as
     when dbbackup started, it started using more log, and backups got
delayed,
     this may be due to that backup incoming rate is fairly larger at 40
gb/hr.
All the clients are incrementally backing properly, and there are no
problems, when
     they are done at a lesser rate.
Cache-hit ratio is above 98%, bufpoolsize above the nominal size of 32768
(above 100k),
     all housekeeping of migrate, expire inventory , reclaims run daily
successfully,
     client backups well distributed over 10 hours to average the rate even
though
     it is still quite high at 40gb/hr.

Some answers are quite commendable and relevant, as per results here.
     Requesting developer to increase log-size upper limit above 5.4 gb
     Reducing actlog retention # of days (already we are keeping only 3
days of actlog only)
     Stopping & starting server, works but not a practical, as lot backup
fail and
          not a logical solution.
     Splitting server to share load over 2nd server. Yes, but can not be
implemented immediately,
           with regards to library sharing, or a new node implementation.


In general it looks like at large size of 100 gb of adsm database, the
db-updates
are getting slower (considering that log-usage is less than 1 %, when
db-size was
70 gb or less).  Needing a growing size of log-records till it is hitting
the maximum
of log-size; A log-failure clears it up immediately, however after the
final backups failed.
Currently trying to reduce process during night or reduce daily backups to
contain log.

Thanks again, vijay.

On Mon, 11 Dec 2000, Vijay Havaralu wrote:

> Hello every one,
> We are running tsm-server ver 3.7.20 on aix 4.3.2 with about 400 gb per
day
> backup.
> Our adsm db-size is close to 100 gb and growing daily.
> Almost each day log is overflowing with 100%, all backups fails, and
server
> recovers & continues
> We are already using max 5.4 gb of log and can not increase any more on
log
> volume.
>
> I appreciate, if anybody has any solutions this problem to reduce log
usage
> or extend log beyond 5.4gb.
>
> Thanks, vijay.
>
<Prev in Thread] Current Thread [Next in Thread>