All,
I have a name server (BIND 9.2.4 as packaged with CentOS 4.6 i386)
which is reasonably busy (30,000,000 queries per day/350qps average)
on a Celeron 2.6GHz with 1GB RAM. Typical CPU utilization is around
25%. However, in the last week or two it will periodically increase
CPU utulization gradually until it is using 100%. I didn't catch it
the first time until it was at 100%, but I have caught it climbing
before it maxed out another 4 times. This does not seem to happen on a
regualr interval or time of day. It takes several hours for it to
climb from 25% to 70% or 80% when all I have to do to "fix" it is to
stop and restart the named process.
While investigating this I realized the number of queries had
climbed significantly in the last year and decided to perform an
upgrade of both the HW and SW. The new server will be CentOS 5 with
the packaged BIND 9.3.3 on a new quad-core Xeon with 8GB RAM. I
realize I can get more performance form BIND 9.4 or newer, but I'd
prefer to stay with a binary package for easy patching.
My question: Has anyone had the above described "CPU leak" kind of
behavior with named ? Any guess as to what is the cause ?
Thanks,
Mike
--
************************************************************
Michael J. McCafferty
Principal, Security Engineer
M5 Hosting
858-576-7325 Voice
http://www.m5hosting.com
************************************************************
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
