All of a sudden a set of jobs collecting DS log data started performing poorly. It now takes 3 and half hours as opposed to few minutes it took before (jobs run normally for the last couple of years). I assume this should have something to do with possibly corrupted or overloaded DS run-time repository (log files, indexes and such). Could you please suggest any techniques and/or methods to clean it up?
Anything else that might be useful to keep DS repository and DS run-time environment in a healthy state?
Thanks!
Vladimir
Sudden degradation in performance of job collecting logs
Moderators: chulett, rschirm, roy
Start off with a simple test - go to the project directory and list the files ordered by descending size. The data for the most common file type, DYNAMIC, is contained in a DATA.30 and OVER.30 file inside a directory which is the hashed file name. So, are there very large log files? If yes, you can use the TCL command "LIST DS_JOBS WITH JOBNO EQ {LogFileNumber}" to get the respective job names, then clear those files from the director.
It makes sense to make sure that your autopurge settings are appropriate for each file. The project level settings just give a default value for a job; so it is possible (and common) to have single jobs that have different settings and where the logs can grow indefinately.
If you don't have large log files then you need to look elsewhere for performance bottlenecks.
It makes sense to make sure that your autopurge settings are appropriate for each file. The project level settings just give a default value for a job; so it is possible (and common) to have single jobs that have different settings and where the logs can grow indefinately.
If you don't have large log files then you need to look elsewhere for performance bottlenecks.
</a>