Page 1 of 1

Clear & PH& logs & RT_LOG

Posted: Tue Aug 17, 2010 6:42 am
by Akumar1
Hi There,

What happens actually if we do not Clear & PH& logs & RT_LOG at regular intervals or if we dont clear the unwanted logs in scratch disk.

Can it cause datstage jobs being HUNG or LOCKED?


thanks in advance for kind input on this.

Akumar

Posted: Tue Aug 17, 2010 7:13 am
by Sainath.Srinivasan
Nothing will happen.

I thought a poster who is level 5 in both Server and PX will be aware of that.

Posted: Tue Aug 17, 2010 10:13 am
by JRodriguez
Well ... in my humble opinion I would clear those logs regulary, DS jobs can and will fail if not under certain conditions

It will be good to state that I'm just level 3, not 5 (Before I get a very rude or judgemental reply from another level 5 poster)

Regards

Posted: Tue Aug 17, 2010 11:15 am
by kumar_s
Yeah... creates an issue when the &PH& gets full.
And RT_LOG, you can purge it with Admin level settings. Either Time based or run based.
Maintenance is all what, we talking about.

Posted: Tue Aug 17, 2010 12:07 pm
by Akumar1
Hi Kumar , thanks for the kind input on this , however I just wanted to know whether this will cause datstage jobs being HUNG/LOCKED or not?

Posted: Tue Aug 17, 2010 12:09 pm
by Akumar1
I believe it should never get hung or get locked because of it.

Posted: Wed Aug 18, 2010 12:51 pm
by mandyli
Hi

Best one clear every month or ever weekly.

Datastage will not hung but bit slow down when you compile or view log.


Thanks
Man

Posted: Wed Aug 18, 2010 3:11 pm
by kduke
Unless you get warnings on every row then a log file will not get filled up. By default they hold 2GB. &PH& is a directory and the files in here are very small. I have seen a lot of places not clear these and they run fine. Logs should be auto purged based on days or number of job runs especially in DEV. Set these up at the begining of a project.

You may want to change your rating. It is usually wise to be under estimated than over estimated on a web site like this. Make yourself a 5 on your resume otherwise you might take abuse from someone on this site. Ray can call himself a 1 and everybody knows he is a 5. The truth is nobody knows it all. So you might be a 5 in many aspects. I have no idea. Besides it is just a number.

Posted: Wed Aug 18, 2010 6:30 pm
by ray.wurlod
The more entries that exist in &PH& and/or logs, the slower job startup times will be, particularly on operating systems that use linear directory file structures. &PH&, being a directory, can not fill with files until there are more than 4G entries (or the disk fills).

If RT_LOGnnn fills (reaches 2GB) it will become corrupted. It will no longer be able to open this log in Director (fortunately Clear Log can be invoked from Status view). A full log will not prevent the job from being opened in Designer. However a full or corrupted log will prevent the job from being run.

Posted: Wed Aug 18, 2010 6:55 pm
by kris007
We have had instances where the RT_LOGnnn files got huge and it slowed the start up times of the parallel jobs literally around 2-4 minutes. We cleared the log files and things came back to normal.