Archive log compression?

RecoveryOne

ADSM.ORG Senior Member
Joined
Mar 15, 2017
Messages
347
Reaction score
81
Points
0
Was wondering has anyone enabled archive log compression? I seen it was a new feature for TSM 7.1.1, and my instance started out as a fresh 7.1.1.0 install.

I've had a few incidents recently (twice in the past week) where I was nearly out of or out of archive log space and filling up actlog quickly. Database backup in one instance couldn't keep up, and had a server down event.

Current actlog is limited to 200GB on a 256GB filesystem (set to greater than 8% reserve space at recommendation of IBM).
Archlog is 765GB on a dedicated 765GB filesystem.
No options currently available to have a failover log directory unless I steal it from one of my storage pools.

Now, as to why I'm running out of logspace all of a sudden, I can't put a finger on. Normally on a day to day basis I have about 100 to 200GB free in arclog when my database backup finishes and actlog is 1 to 5gb used at the time. No major new ingest or workload being asked. Haven't expired a massive ton of data other than what is normal for my environment on a day to day basis. Guess time to start looking at apars :)

Anyhow, archive log compression, enable it? Don't ? Performance impact on database backups or other activities? Does archive log compression lead to issue if a dsmserv restore db is needed, or cause issues with any sort of upgrades to the server (say to 7.1.8, or 8.1.x)?

I see there's a note to use caution because it can delay active log movement to archive but rarely (at least historically, and not these past few days) active log usage hasn't been an issue.

If it helps the guru's here, my server is decent: S824 16 core smt8 enabled so any cpu hit should be minimal. Most of the time my cpu is greater than 70% idle with a runqueue that rarely pushes 10.

Good thing its a dynamic setting, but....Though I'd ask here to see if anyone is using it and their thoughts.

Thank you all!
 
I haven't used the compression feature so couldn't comment on that, but I had a similar problem with rapid log growth because of online reorgs running. Are you still running 7.1.1? Have you checked actlog for reorgs?
 
On 7.1.7.0 now. When I looked through actlog last reorg was 2 days ago. And have the heavy hitters disabled for online reorg. That was my thought as well, perhaps one of the SD_* tables for containers, but nothing.
 
Back
Top