Re: [Ldapdns-users] Real-world $HANDLERS and $THREADS values with OpenLDAP

[Mrs. B. <mrs...@ni...>]

LDAPDNS has to answer very quickly, and there are race conditions with
the OpenLDAP client libraries.

Some of these appear to disappear if you set HANDLERS=1 and THREADS=128

On Wed, 2005-12-21 at 14:23 +1100, James Wakefield (Sunet Sysadmin)
wrote:
> G'day guys,
> 
> I'm running OpenLDAP 2.2.23 and ldapdns 2.06 (as per Debian stable) and 
> every few minutes syslog is telling me that a handler or two has waited 
> too long.  At this point, DNS lookups time out until ldapdns is 
> restarted.  I am supporting some reasonably intense apps with the LDAP 
> server (mail, web vhosts, RADIUS, NSS for user/group lookups), but those 
> are working without complaint.  Following the FAQ, I tried upping 
> $HANDLERS (to 256 from 128) but to no avail.  The LDAP server is on 
> 127.0.0.1 so I would think it's unlikely to be network latency.  Can 
> someone with a similar setup let me know what values they're using, as 
> restarting ldapdns from a cron job every few minutes is not what I want 
> to be doing.
> 
> Cheers,
>