Batch job logfile output

[Art Wiens]

I'm building new v8.3 systems and I've noticed that the content of batch job logfiles don't appear until the job finishes. This does not occur on our current systems. I thought I remembered a logical to control this behaviour, but can't seem to find it. I have been R'ingTF'ingM's but no luck so far.

Cheers,
Art

[Oswald Knoppers_1 · Accepted solution]

Are you looking for the command 'set output_rate'?

This will set the rate at which output is written to the logfile.

Oswald

[labadie_1]

and see this thread on that subject

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=629558&admit=109447626+1221143534251+28353475

[Art Wiens]

Ok, maybe that's what I remember. I tried another simple DCL procedure and it works ... output after a 1 minute default. In the case where it's not working I am using a batch job to initialize shadow set members - no output until it completes. Here's the procedure ... any idea why?

$ type init_dsax.com

$ set ver
$ set noon
$ show time
$ disk1 = p1
$ disk2 = p2
$ disk3 = p3
$ label = p4
$!
$ initialize /shadow=('disk1','disk2','disk3') 'label' -
/cluster=16 /directories=16000 /erase /headers=900000 /nohighwater -
/index=begin /limit /maximum_files=900000 /stru=5 /system
$!
$ show time
$ exit

Art

[Jess Goodman]

The flush of the log file is done via a supervisor mode AST. You could use the TQE freeware program to check if the timer that fires off this AST was somehow cancelled by the INITIALIZE command.

The timer will look like this:

Request type 00 Request is process timer, one-shot, relative
Process BATCH_2003489 Username SYSTEM
EPID 7B80DDCE PID 006E01CE AST 7EE76030 ASTPRM 7FFBE3E0
Due time 11-SEP-2008 15:31:02.05, Time(quad) DC706747 00A7F7AB
No delta time, Delta(quad) 00000000 00000000
RMOD 42 Super AST, EFN 1F ERQPID 7B80DDCE RQPID 006E01CE

[labadie_1]

And you have the Sda extension TQE, available at least with 7.3-2


SDA> tqe
Timer Tracing Utility TQE commands:

TQE LOAD
TQE UNLOAD

TQE START TRACE [/BUFFER=pages]
TQE STOP TRACE

TQE SHOW TRACE [/SUMMARY]
[/IDENTIFICATION=pid]
[/ADDRESS=address]

[Art Wiens]

I know there wasn't much output to see from this particular process ... I hadn't run any other batch jobs yet and thought it might have been a "bigger issue".

I was init'ing a bunch of shadowsets and wanted to do: type init_dsax.log;*/page just to see what time they all started when I noticed the "problem".

Thanks,
Art

[Jon Pinkley]

Initialize/erase is a long running operation, and evidently runs in an inner mode, because if you run it interactively, control-t has no effect. Therefore, it appears that any supervisor mode ASTs are disabled for the duration. So it can't flush once INIT starts. I am not sure what you could do right before the INIT command to initiate a flush, there probably is something.

And init/shadow doesn't provide must useful info about each member anyway (at least in the test I just did on 7.3-2).

So it appears you made an unlucky choice of program that you wanted to see the output of.

Jon

[Art Wiens]

"Initialize/erase is a long running operation ..."

~ 70 minutes for 3 member 36GB "drive". Two members on one EVA8000, one on another.

"And init/shadow doesn't provide must useful info about each member anyway (at least in the test I just did on 7.3-2)."

No diff on v8.3 - no output from init/shadow command.

"So it appears you made an unlucky choice of program that you wanted to see the output of."

As usual ;-)

Cheers,
Art