Apparently missing DSD.RUN files

jackdaw · Post by **jackdaw** » Wed Jun 03, 2009 9:43 am

Trying to locate DSD.RUN files in 1st step to track down why job just cut-off entailing a reboot of the server. Director auto-purged log shows 5 entries, spanning the date when the problem occurred. I used a routine found here to derive the likely file names for all 5. Found 3 but the others don't appear. For Director to show the detail there must be some phantom DSD.RUN_ files I believe. What is likely to have happened ? My interest here is in trying to understand what's going on - the problem has been resolved - albeit clumsily, to no (apparent) harm.

chulett · Post by **chulett** » Wed Jun 03, 2009 9:55 am

So... have you tried grep'ing through the contents of the project's &PH& directory for the job number in question?

jackdaw · Post by **jackdaw** » Wed Jun 03, 2009 10:35 am

Thanks, yes. The job number for the server job changed after the problem was resolved, presumably because it was re-compiled, so I've grep'd both job numbers. I've got a match for some (with the date and time for the job from Director), but others that are shown in the Director log, that I've derived the DSD.RUN file name for, aren't found. I can read the entries OK so there must be a file somewhere for them, but if so it doesn't appear closely related to the expected date and time.

Hope that makes sense !

chulett · Post by **chulett** » Wed Jun 03, 2009 10:39 am

As far as I know, the information in the job logs are sourced from the &PH& entries, not linked to them. Meaning they existed at one time but are not necessarily still there, especially if you have any kind of automatic pruning going on. Can you not get whatever it is you need from the log entry itself?

jackdaw · Post by **jackdaw** » Wed Jun 03, 2009 10:43 am

Oh, OK.
No, there is nothing, the process doesn't finish, there's no terminal Control entry, no error. It just hung and the developer couldn't reset or re-compile because of a locked process, so killed the 'service' which availed naught, then had the server re-booted, then compiled the job again, and it worked fine in sequence, as before. Apparently it's happened a couple of times.

jackdaw · Post by **jackdaw** » Thu Jun 04, 2009 9:52 am

The RT_LOG for this job shows 4 copies of the same sequence of severity / events and text, so it shows the same job as having started and finished 4 times. They are copies except that the order in which the transforms are done is different each time. Needless to say, by design the job is meant to be run once from a sub-sequence. How do I look at the job controller to see how this is configured ?

chulett · Post by **chulett** » Thu Jun 04, 2009 10:11 am

Not quite sure what you are asking. If your job was run under the control of another job, then you should be able to right-click in the log view and select 'Related log' to see the log for the controller. And from there open the job design.

jackdaw · Post by **jackdaw** » Thu Jun 04, 2009 10:20 am

I'm still stuck with this situation where a log in Director for a server job is incomplete. (And I'm looking at this because the job just hung until we re-booted the server). Having found the RT_LOG for the job in question I saw there were 4 iterations of messages in it, but the job is only supposed to run once, and that's how it appears in Designer. So I looked at the Director log for the sequencer that calls the job, and it too is unfinished. The sequence job in Designer only shows one call to the server job, so I was wondering if I could look behind the scenes of the sequencer to see if what was actually happening showed up as different to what was apparently happening.
I'm sorry if this is all a bit like pulling teeth.....

chulett · Post by **chulett** » Thu Jun 04, 2009 10:53 am

That's ok, I think I have a couple of molars left if you look all the way in the back... [choke,gack] Sorry.

If the job crashed or hung and was killed, I can see both logs being incomplete and complicating this. So... these four iterations - are they complete iterations from Starting to Finished with different start times? Do they overlap? Something else entirely? Any chance this is a multi-instance job and you're looking at the base log rather than the individual views the specific invocations would show?

I'm confused by what "the order in which the transforms are done" statement might mean, not sure how you get something like that without changing and recompiling the job between each run, but I would assume that's not the case. Perhaps it would help to print those log entries to a file and then paste the contents here so we can all see it?