[Date Prev][Date Next] [Chronological] [Thread] [Top]

"Deferring operation" messages, slow operation?



Good day,

We've been trying out OpenLDAP 2.1.22, and are considering migrating some
data to it.  The machine in question is a dual zeon 2.8GHZ (4 procs w/
hyperthreading) w/ 2.5GB of RAM, running Red Hat 9.  We installed openldap
by recompiling the source RPM from Rawhide (openldap-2.1.22-8) which uses
Berkeley DB by default.  The RPM version of BDB is "4.0.14", but the version
reported by slapd when it starts is "Sleepycat Software: Berkeley DB 4.1.25:
(December 19, 2002)" - the latter which I presume is correct.

We're using ldbm instead of bdb because we have aliases in our data, and bdb
doesn't support aliases yet, apparently.

Unfortunately, the server's performance is so far noticeably worse than our
2.0.27 servers w/ gdbm which is even running on slower hardware.  In
particular, when testing the server with queries, the message "deferring
operation" shows up very, very frequently.  Searching on Google and the
mailing list archives comes up with mostly unanswered pleas for help (and
there's nothing in the FAQ), with the exception of one user who had a very
large amount of threads running.  But, when testing, the number of threads
only hits 10 - 20 threads, and slapd is using less than 1% of the memory on
the machine.  The machine isn't doing anything else.

The configuration is pretty standard, although we did set "cachesize" to
100000 on both the old and new servers.  This name doesn't appear in the
2.1.22 man page, so we're not even sure if it's supported anymore.

Could someone tell us what this message means, and if it indicates a problem
with the server?  Is there anything else that might cause the general
slowness of it?

Thanks in advance!

============================
Darren Gamble
Planner, Regional Services
Shaw Cablesystems GP
630 - 3rd Avenue SW
Calgary, Alberta, Canada
T2P 4L4
(403) 781-4948