Hello,
Woody had the 2.0.14 version of slapd (OpenLDAP) upgraded a while ago,
instead of the official "stable" 2.0.11 version and also not the 2.0.15
most recent version.
Since then, with my LDAP PAM/NSS configured woody box, I get a
proliferation of slapd processes - over 20, as reported by ps. I don't know
if these are threads or forked processes (how do you tell, anyway).
Furthermore, netstat -a reports a ton of ESTABLISHED and CLOSE_WAIT
connections from localhost to localhost:ldap (and a few vice versa). The
CLOSE_WAIT ones actually persist across /etc/init.d/slapd restart, while
the processes do go down in number to just a few.
lsof -a seems to indicate that these CLOSE_WAIT processes are also from
slapd, but really, they're from all sorts of processes - presumably every
process which uses NSS or PAM will have one of these CLOSE_WAITs lying around.
So:
1) How do I keep the number of slapd processes (or threads) in check - that
is, down to just two or three?
2) How do I get the CLOSE_WAIT sockets to actually fully close and disappear?
Thanks,
Doug
