Hi,
I have a rather urgent problem with a large program we're working on.
It does a bunch of processing to generate many sets of data files in
big loop. Each time one set is generated, it calls a number of system()
calls to spawn rcp to copy the files to a remote system. At some point
after many such files are generated and copied, another system("rcp...")
generates the message "Not enough core" and dies with doing the copy.
At this time, we can't find any good reason why such a message should
come up from a subprocess. The O/S is OSF/1 v1.3a, 128 MB main memory,
about 200 MB swap space, the program shows less than 8M and approx. 1.7
from ps aux, and vmstat shows 8137 in the free column.
So the question is: why is this happening? It looks like the system
should have plenty of memory. Does the concept of "JOB" (as in VMS)
exist, where all processes in a JOB are sharing the same memory resource
quotas? If so, should each system() simply create a process, which
should free up its resources when it completes the command?
Are there any system parameters that can be tuned to at least
temporarily alleviate the problem?
Any help would be greatly appreciated.
eyc
Received on Tue Jan 31 1995 - 16:54:48 NZDT