I posted the below to the sparks-discuss list but only heard the
crickets chirping. So, I'm hoping the wider audience may be able to
help:

I've joined the sparks list just to bring up the fact that even with
B54, we are seeing major problems with nscd handling ldap requests.
Prior to the sparks inclusion, we had perfect stability. Post that, we
had the core dumping on nscd.

Now with B54, it doesn't appear that its dumping core, but rather the
other error which would normally lead up to the core dump is
happening:

Dec  1 08:23:02 sram nscd[266]: [ID 293258 user.warning] libsldap:
Status: 7  Mesg: Session error no available conn.
Dec  1 08:25:50 sram last message repeated 4 times
Dec  1 08:26:57 sram nscd[266]: [ID 293258 user.warning] libsldap:
Status: 7  Mesg: Session error no available conn.


We are using LDAP in one task only: netgroup parsing (since solaris
does not support flat file for netgroups -- urgh)

So, when our system reboots and sometime during operation, we have all
our NFS mounts silently fail on the server (no log messages) except
for these nscd failure messages. The clients of course get permission
denied.

Furthermore, if nscd is disabled, mountd makes those complaints, so it
looks like there are connection pooling issues with libslap that have
somehow changed with SPARKS integration. I looked into /etc/nscd.conf,
and it appears to not cache for netgroups, so no options there seem
relevant, and no options to ldapclient seem relevant either.
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to