Application core dumps ENOMEM in trace

[Paul Clayton]

RP7420 8 cpu's 6gb ram 24gb swap.
Back end EMC disk 2gb Fibre
HP-UX 11i patched December 2004.

Guys,

We have a customer with a server of the above config. Our application will not run on this server for some strange reason, and we are now at our wits end.
Initially the server was built with Dec 2003 OS install CD. At the time of analysis, we detect that there was a problem surrounding /usr/lib/libc, and libdld. Further investigation showed that it was patch related, and the December 2004 cluster patch and PHKL_25233.depot PHKL_29696.depot PHKL_31993.depot PHSS_26793.depot
PHKL_25652.depot PHKL_29704.depot PHKL_32061.depot PHSS_30968.depot
PHKL_27094.depot PHKL_30034.depot PHNE_30367.depot PHSS_31221.depot
PHKL_27096.depot PHKL_30035.depot PHNE_31247.depot
PHKL_28122.depot PHKL_31003.depot PHSS_26792.depot

This fixed the majority of problems which would report ENOMEM in a tusc trace. Subsequent tracing showed a marked reduction in ENOMEM errors, but found that these ENOMEM errors where now appearing when the oracle (9205) database was being addressed. We have addressed most of the obvious stuff, but are at an end. If anyone has had an obsquere problem like this and managed to fix it I would appreciate assistance.