[omniORB] omniNames crash: Too many open files

Jonathan Buchanan buchanan@olsen.ch
Wed, 08 Mar 2000 13:25:07 +0100


We have had a stable CORBA network for quite a while now but suddenly
today we find omniNames keeps crashing with the following lines in
stderr.log:

Wed Mar  8 13:15:44 2000:

Starting omniNames for the first time.
Wrote initial log file.
Read log file successfully.
Root context is IOR:000000000000002849444c3a6f6d672e6f72672f436f734e616d696e672f4e616d696e67436f6e746578743a312e3000000000010000000000000028000100000000000d3139332e37322e38312e3238000013240000000c38c64470c61337e500000002
Checkpointing Phase 1: Prepare.
Checkpointing Phase 2: Commit.
Checkpointing completed.
open(/dev/zero): Too many open files
libthread panic: alloc_chunk (PID: 2021 LWP 2)
stacktrace:
        ef3c7c5c
        ef3ceab8
        ef3ce868
        ef3c78ec
        ef3cabd0
        ef3cd658
        0

Platform is Solaris 2.7 running omniNames 2.8.0.

Checked disk space, including inode count, but don't see any obvious
problem relating to 'Too many open files'.  Any other ideas anyone?
Jon.