I see a lot of memory problems with back-bdb when a particular
configuration is set to allow too many threads.
I've never specified a threads statement in slapd.conf. The default is
32 threads. Is that too high on some machines?
Yes, my master runs at 9. Stress testing shows a peak at 8 - 10, so I
picked 9. (This is a 2xPro200 with 384Mb and UW SCSI, LDAP database
striped across two drives. Albeit it isn't a current machine, but
testing newer machines shows that threads seems to top out about ~12 on
x86 boxes).
When you
spawn so many threads that all of the free RAM in your address space is
consumed, slapd aborts because it cannot allocate any more dynamic
memory for its normal operation.
This is not the case with my machine. There must be something else wrong.
I've never had slapd crap out, but it does get stuttery if you push up
the thread count and utilization too high. But I allocate a great deal
of swap, RAMx4.