Re: syslog spiral death
ron@caldera.com
Thu, 24 Oct 1996 08:29:37 -0600 (MDT)
> Hi all,
>
> Have been seeing a problem with a system we have that has quite a
> heavy load - lots of short lived processes (upto 3 per second which can
> run for a few sconds to a few minutes).
>
> The problem is, that each process adds logs to syslog (for
> accounting). Everything usually runs fine, but it appears that when
> the syslog file gets large (say 8G) that syslog starts using more of
> the CPU, which creates a backlog in the processes, which results in a
> higher load, which means more syslog acitivity etc.
[...]
I've seen a similar problem also. Running strace showed that syslogd
was doing fsync() quite often. As the log files grow, fsync() takes
longer and longer to complete. I don't know whether to consider this
an inefficiency in the fsync() implementation or poor behavior on the
part of syslogd for calling fsync() so often... We avoid this problem
by keeping our log files trimmed (via chklogs).
Ron
--
Ron Holt <ron@caldera.com> Caldera, Inc.