Detecting corrupt log files

[rabs]

Friends,

I am managing a project where there are frequent datastage crashes which require a server reboot. We are investigating and following this up with IBM.

One side effect of having to reboot the server mid-batch execution is that the log files of jobs which are executing often become corrupt, returning BLINK and other errors.

Is it possible to run some sort of uv/basic/other command to detect all the corrupted log files?

The idea would be to then clear these files or somehow recreate them programatically in order to make sure they are not corrupt.

This would form the basis of a cleanup/restart script which we would run when we have to restart.

Do you think this is possible or have you done something similar before?

Cheers,
Rabs

[WoMaWil]

(1) Use Auto-Purge to delete the log-files after up to 2 or 3 runs

(2) Are you on windows or are you on unix?

[ray.wurlod]

The uvbackup command can do what you require. Search the forum for details.