Getting UNIX Signal KILL....

Post questions here relative to DataStage Enterprise/PX Edition for such areas as Parallel job design, Parallel datasets, BuildOps, Wrappers, etc.

Moderators: chulett, rschirm, roy

Post Reply
stiwari
Participant
Posts: 64
Joined: Wed Apr 05, 2006 3:45 am

Getting UNIX Signal KILL....

Post by stiwari »

Hi All,

I am loading 1.2 billion to my fact table, I have used LOOKUP stage to join all the dimentions. When I kicked off the job it has started at very good seed but when it had reached to .5 billion, It got aborted ...Log file showing UNIX Signal SIGKILL 9...that means that job has aborted forcefully....And when i looked into status of unix box....one folder with /Dev/fstst....it is showing used memory 100%...My source and target is DB2.

Can any one help me....what could be the reason

Thanks in Advance

stiwari
kumar_s
Charter Member
Charter Member
Posts: 5245
Joined: Thu Jun 16, 2005 11:00 pm

Post by kumar_s »

That might be your temp directory. The Scratch will get used during run time. Are you using lookup fileset or sparse lookup?
You may need to increase the disk space for the directory.
Impossible doesn't mean 'it is not possible' actually means... 'NOBODY HAS DONE IT SO FAR'
stiwari
Participant
Posts: 64
Joined: Wed Apr 05, 2006 3:45 am

Post by stiwari »

Kumar,

I am using lookup type as normal......Is it really matter that i have to use lookup type that won't be using Scratch file this much....


stiwari
ray.wurlod
Participant
Posts: 54607
Joined: Wed Oct 23, 2002 10:52 pm
Location: Sydney, Australia
Contact:

Post by ray.wurlod »

Technically, of course, every UNIX signal can be generated by a kill command. In this case a process has - as you noted - been forcefully terminated with a SIGKILL signal (kill -9).

Yes, the Lookup stage does use scratch space if it needs to. The Lookup stage in normal mode is (generates) a composite operator - one to read in ALL the lookup data (for example into a virtual Data Set), the second to perform the actual lookup processing. Building a virtual Data Set consumes memory.

You may need to find a better strategy for combining your Data Sets.
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
kumar_s
Charter Member
Charter Member
Posts: 5245
Joined: Thu Jun 16, 2005 11:00 pm

Post by kumar_s »

Lookup will be always efficient when the lookup data can be made to fit into memory. Try to increase the scratch space else try out some other strategy like converting lookup into join.
Impossible doesn't mean 'it is not possible' actually means... 'NOBODY HAS DONE IT SO FAR'
Amarpreet
Participant
Posts: 20
Joined: Tue Apr 04, 2006 11:45 pm
Location: India
Contact:

Post by Amarpreet »

Can anyone tell, for what purpose the folder /dev/fstst/ is being used? Is it used for some temporary loading by lookup stage?
kumar_s
Charter Member
Charter Member
Posts: 5245
Joined: Thu Jun 16, 2005 11:00 pm

Post by kumar_s »

Check in your DS Adminstrator setting for environmental varialbe TEMP_DIR, if it is pointing to this directry, it is used as scratch space.
Else check your Config file, you can find if it is been used as one of the resource pool.
Impossible doesn't mean 'it is not possible' actually means... 'NOBODY HAS DONE IT SO FAR'
ray.wurlod
Participant
Posts: 54607
Joined: Wed Oct 23, 2002 10:52 pm
Location: Sydney, Australia
Contact:

Post by ray.wurlod »

Your UNIX administrator will be able to tell you whether that is a device that you are not supposed to be using, or part of the file system. To me the name "fstst" suggests "file system test" which in turn suggests that you may not be supposed to be there.
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
Post Reply