2.3.16/bdb has now run here for more than 24 hours without a failure, so
it's at the very least a whole lot better than 2.3.11 was. (This is on
Linux 2.6, bdb 4.2, 100,000+ objects, ~4 million operations a day.)
And just hours after sending this, slapd hung. This behavior's different
than what I had been seeing before: slapd hangs at 100% cpu usage,
eventually refusing new clients.
At loglevel 256, nothing of any use is revealed. Is there some level I
can safely turn on to help debug this problem? Will the upgrade to
2.3.17 help?
I'm using the default thread count and I noticed that at higher thread
levels, this CPU situation would happen sometimes within a couple of
hours. Should I go even lower than 16, or is this another Linux-2.6
issue where I should consider using 2.4 instead?