[Date Prev][Date Next]
Re: CPU utilisation = 99% => death; help, please!
John Madden wrote:
> > we have set up OpenLDAP v2.0.15 on RH Linux 7.1 on a 2 CPU 800MHz Acer
> > box. We are authenticating about 5000 users with LDAP, with automount
> > home directories. We are providing password, group and automount info
> > to our laboratories. We have up to 200 users on the system at once.
> > LDAP stops several times a day. I watched this happen on /usr/bin/top
> > today; usage went up to 99%. I am working arround this using a cron
> > job that checks the output of ps and restarts ldap if necessary, but we
> > have a serious problem, and after reading all the documentation, am
> > unsure what to do next. I am now running 2.4.10-ac12, but have tried
> > many kernels.
> I understand 2.4.10 has serious issues with network load and simply dying,
> so you might try 2.2.19 as it's much more stable than anything in 2.4, but I
> doubt that's the problem here.
I am using Alan Cox's kernels, which I have found to be okayish; they do not
include the wildest changes introduced into Linus' kernels. For what it's
worth, slapd has not died since I rebooted into 2.4.10-ac12.
> First off, show us some more debugging. Is it a segfault? Is it being
> killed by the kernel OOM killer?
That is my question: how to identify what is happening? I noticed that memory
consumption was modest while the CPU load was so high.
> As for the 99% cpu thing, you probably need some more indexes.
Thank you! Right, I think that is probably the right thing to investigate.
How can I find what the server is looking up? How do I get stats on what it is
I know these are basic FAQ type questions, but none of the many docs I've read
have given me much of an indication.
Nick Urbanik RHCE firstname.lastname@example.org
Dept. of Information & Communications Technology
Hong Kong Institute of Vocational Education (Tsing Yi)
Tel: (852) 2436 8576, (852) 2436 8579 Fax: (852) 2436 8526
PGP: 53 B6 6D 73 52 EE 1F EE EC F8 21 98 45 1C 23 7B ID: 7529555D