Recompiled with
(a) add --enable-cassert to the configure command;
(b) do make COPT=-g all instead of just make all.
On the same machine that the backend runs on.
Still same behaviour. Saw small notice in log file:
proc_exit(0) [#0]
shmem_exit(0) [#0]
exit(0)
proc_exit(0) [#0]
shmem_exit(0) [#0]
exit(0)
Still no core.
I'm not really sure there is a linkage with the ordering thing -
remember that this worked fine a week ago before I upgraded to 6.5.3.
Unless the backend reaches out to different libraries than in 6.5.1, I
think it's unlikely to be RH6n libraries. Also, even tough RH libs
can be odd, this is not a redhat machine - it's VA linux, which uses
RPMs but is in many aspects different than VA linux.
So I'm still stumped. For laughs, I think I'll try it on a much less
powerful RH60 box and see what I get.
Let me know if there's anything else I can do to help diagnose.
Karl