delay during the finishing stages of DS jobs

[cosec]

Hi All,

We upgraded from Datastage v7.5.2 to v8.1 and noticed that there is a delay in the jobs. For e.g if a job completes in 8s in v7.5.2 now it takes around 14s to complete in v8.1.

After investigation, it was noted that the slowness is during the finishing stages of the job (during the purging of entries)

The differences at the finishing stage is as follows:

Previously in v7.5.2 the job:
Auto-purged 10 log entries

In v8.1:
Job-level auto-purge, 10 log entries removed

Please advise if there is any setting to minimize this delay during the point mentioned above. Any advice would be appreciated ! Thanks !

[ray.wurlod]

Where are you doing your logging? (What are the values of RTLogging and ORLogging in DSParams?)

[cosec]

Hi Ray,

Thanks for your response.

Current value:

RTLogging=0
ORLogging=1


We have changed it to the following and the performance is now not bad (its almost the same as in v7.5.2)

RTLogging=1
ORLogging=0

I was reccommended to change the MFILES setting from 100 to 200 or 250? any impact in doing so ? rather how does MFILES setting affect ?

[ray.wurlod]

MFILES is the number of files (including local repository hashed files) that can be open simultaneously. If you have more files open than this, some are not genuinely open, but are maintained in a "rotating file pool". I usually set this to 1000 on UNIX systems.