DSXchange

One of our job is failing with the below given error message:

"node_node2: Fatal Error: Unable to start ORCHESTRATE process on node node2 (hp823npc): APT_PMPlayer::APT_PMPlayer: fork() failed, Not enough space"

There are other jobs which are loading more data than failing job, but they run fine. The difference between these jobs is that failing jobs has more number of stages (Stages used are Oracle Enterprise, JOIN, SORT, Change Capture, Datasets, Lookup. There are multiple stages of each kind).

The physical space available is more than enough. What could be the problem.

Is there any mechanism or rule to know about the memory to be used by a Stage in a job?

This error is basically telling you that there is not sufficient paging-space pages available to fork an additional process. Normally the parent process will retry at lest 10 times (AIX) before failing. This means that after 100ms the process will fail. You can increase the number of retries to some value > 10 like 15 so that there is more of a chance to fork successfully given peak workloads etc... The parameter on AIX is pacefork=nn.

I would also check how much space has been allocated for paging etc...

If I want to increase the page size how much should I increase? And Is there any rule/process to find out the page size required for a job?