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

Re: slapd eating up resource on 2.3.14



You might be able to nail down the thread that's causing the CPU spins
with a prstat -L.  (or prstat -mL might also be helpful) You could
then gcore the process to examin what the program was doing at that
time, or truss the pid and look for the offending LWP(s) to see what
they're up to.

You may have to recompile with debugging (CC_FLAGS += -g ?), and I'm
not exactly sure how gcore works with gdb, but I'd be interested to
see if it's one offending thread, or a bunch.

_Matt

On 1/5/06, Spicer, Kevin <KevinS@bmrb.co.uk> wrote:
> Hi,
>
> I've been running 2.3.14 on my test setup for a few days with no
> problems, last night I rolled it out to my production systems as an
> upgrade from 2.3.11 (shortly before 2.3.15 was released!) I'm now
> experiencing a problem that I can't track down.  After a period of
> running (minutes or hours) slapd starts to chew up CPU and become very
> slow to respond, at times slapd constantly uses 100% of one CPU (these
> are dual CPU systems).  I had no such problem on 2.3.11 (although I'm
> not ruling out a configuration error on my part).  The only change I
> made to the configuration was to comment out a 'threads 500' directive
> (thanks for the helpful warning message about that BTW!) - that isn't
> related, as reinstating it doesn't solve the problem.  I've rerun
> slapindex in case it was an indexing issue but that also hasn't helped.
> I'm seeing this on several servers and it tends to happen when there is
> a reasonable amount of user activity on the server (the user load isn't
> very high though, just the problem doesn't appear to happen when the
> server is mostly idle).
>
> Can anyone offer any insights as to what I might look at to troubleshoot
> this?  I've been unable to find anything similar in list
> archives/FAQ/ITS - although I might not be using the right search terms
> ;)  Logging at level 256 I don't see anything out of the ordinary - I
> haven't tried a higher log level yet, but will if anyone can suggest
> what I should look for.
>
> My setup, briefly
>
> 4 Solaris 9 servers
> ol 2.3.14
> db 4.2.52
>
> Each server has 4 databases organised as a master with three subsidiary
> db's glued on, each server acts as master for one of these with the
> other three replicated onto it using slurpd.  I'm also using the ppolicy
> overlay.
>
> compiled with '--prefix=/usr/local' '--enable-bdb' '--enable-crypt'
> '--with-threads' '--with-tls' '--without-kerberos' '--enable-wrappers'
> '--enable-modules' '--enable-ppolicy=mod'
>
>
>
> --
> Kevin Spicer
> Unix Systems Specialist
> Millward Brown UK Limited
>
> =================================================================
>
> BMRB wins two BMRA awards - http://www.bmrb.co.uk
> _________________________________________________________________
> This message (and any attachment) is intended only for the
> recipient and may contain confidential and/or privileged
> material.  If you have received this in error, please contact the
> sender and delete this message immediately.  Disclosure, copying
> or other action taken in respect of this email or in
> reliance on it is prohibited.  BMRB Limited accepts no liability
> in relation to any personal emails, or content of any email which
> does not directly relate to our business.
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>