Hi Guys,

DB2 V9.1.3 on Win 2k3

We have a recent issue of backup failling because of disk becoming full.

Till yesterday as I know, Image size was varying within 165GB - 185GB. The disk size is 200GB.
Here is database size info.
Today morning it failed just 47 mins after its normal completion time (1:00 am - 3:00am)
Assuming that there might have been too many activities during weekend night, I just restarted ONLINE BACKUP again. This time it Failed again around 6:00am taking about 50 minutes more than normal.
I restarted again without investigating, just in hurry to get mirror up in least time. This time again it failed after 3 hours.

I am not really expert of backup & restore. The database size seems as before. What factors affect ONLINE BACKUP ?? Can it be real-time applications that misbehave during backup resulting bigger backup image ??.

DATA satying the same, what factors force DB2 to generate bigger image ??.

here is command used and result

While monitoring remotely I observed too many commit_Active 's take place at same time.

I believe these COMMIT_ACT at same TIMEstamp originate from DISK or NETWORK inconsistency ?? - not sure !

Please hep me figure out the cause, so that I may be able to take ONLINE backup.

Thanks
DBFinder

Online backup doesn't include logs by default in v9.1 so that can't a contributing factor. DB2 doesn't backup temporary tablespaces so even if applications are using a lot of tempspace, it shouldn't affect the backup image size. DB2 will backup DMS tablespaces up to the HWM so if HWM increased so will the backup image. The HWM can go up even without adding more data (in some cases, reorg may cause HWM to go up).

Any way to reduce HWM. Since I had noticed some HWM are over 20G above the actual data size.

If there is a way then I can reduce sige signifcantly. I have over 10 tablespaces with unreasonably high HWMs.

Regards
DBFinder

Please take a look at the following technote:
IBM - Lowering the High-Water Mark of a Tablespace

Thanks Bella

Two things.

1. As Bell pointed out your backup images are worthless with out log files. Unless you store log files some place else and know which one belongs to which back up. So, if you keep doing online backup include logs.

2. From personal experience with HWM, the only way to lower them is to rebuild table and its tablespaces. It might have been fixed in 9.5, I just haven't tried it yet.

I've been able to lower the HWM in different versions of DB2. It really depends on what's holding it.

Woman's touch

hahaha

Using db2dart ?? Offline ??

The technote is a bit old... v9 manual states the following:

"In general, db2dart requests are to be run when the database is offline. However, for the /DHWM and /LHWM actions, an offline database is not strictly required. The report could be generated without the database being offline, but the reliability of the results will vary depending on how much write/update activity has occurred recently (less activity implies more reliable results). "


Starting from v9.1 fp4, you can use db2dart /dhwm and /lhwm while the db is online: IBM - IZ04647: DB2DART /DHWM AND /LHWM COULD BE RUN WHEN DB IS ACTIVE/INCONSISTENT (ONLINE) TO GENERATE REPORTS FROM V91FP4.

I was wondering why you are not using the COMPRESS in the backup? It definitely should be able to 'squeeze in' the backup.

OUPS. that is a good point that we all have missed