RAM not getting released on Datastage server

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
ArndW
Participant
Posts: 16318
Joined: Tue Nov 16, 2004 9:08 am
Location: Germany
Contact:

Post by ArndW »

Any memory used by DataStage jobs is always returned to the OS when the jobs finish. This has nothing to do with DataStage, but with how UNIX works - once a process completes and logs out, any and all memory is released. It is unlikely that such a serious memory leak exists, particularly on AIX.

The AIX OS itself has some very tricky memory management attributes with it's various high and low water thresholds and the dynamic re-allocation of unused physical memory for disk I/O buffers and subsequent freeing up of those buffers on demand. Basically, the OS will attempt to use as much of the physical memory as it can, for as long as it can. If users don't need it, the OS will take it!

How are you measuring your memory usage?
psk_270185
Participant
Posts: 21
Joined: Wed Mar 12, 2008 4:05 am

RAM not getting released on Datastage server

Post by psk_270185 »

Thanks for the reply.

We are checking it using nmon utility on AIX.

Regards,
Santosh
ArndW
Participant
Posts: 16318
Joined: Tue Nov 16, 2004 9:08 am
Location: Germany
Contact:

Post by ArndW »

Which part of NMON are you using to determine memory usage?
psk_270185
Participant
Posts: 21
Joined: Wed Mar 12, 2008 4:05 am

RAM not getting released on Datastage server

Post by psk_270185 »

Hi,

We used the m part.. i.e teh memory utilization part to check RAM utilization.

And our jobs get aborted once RAM gets finished.

Regards,
Santosh
ArndW
Participant
Posts: 16318
Joined: Tue Nov 16, 2004 9:08 am
Location: Germany
Contact:

Post by ArndW »

The memory (M) page of NMON will show you some relevent information, but you need to know where to look and it does not give you process-based virtual memory usage details. A screenshot might help here, or the values for Min/maxperm, Min/maxfree, Pinned. But there is no direct correlation visible on this page that would support your assumption that a memory leak might exist in DS or the OS.
Post Reply