fsync on large files

Alan Curry (pacman-kernel@cqc.com)
Fri, 12 Feb 1999 22:36:51 -0500 (EST)

I'm coming late to this topic, but with a good reason: I just figured out
today that syslogd's fsyncs have been the cause of some major performance
problems on an ISP's central server. Load averages have been unreasonably
high and gradually getting worse. Today we realized that there is a weekly
cycle to it, and it matches the cycle of the log rotation of
/var/log/messages. As this log file grows (currently 36 megs with 2 days left
before rotation) and beyond, with syslogd fsync'ing every line it writes,
syslogd hangs for long periods of time, and when syslogd is hung, lots of
other stuff hangs.

Previously I deleted all the messages on the large-file-fsync thread without
paying much attention, and now it seems to have died off just when I realized
I needed it. What I could find in the archive did not seem to be a conclusive
answer to what is now a pretty urgent question: would 2.2.1 do better than
2.0.36 is doing here? And if I'm 2.2-phobic (which I'm not, but some people
are), will 2.0.37 fix this problem?

Given that all of the stuff on this machine is pretty close to redhat 5.x
standard issue, including the syslogd and log rotation configuration, are
there lots of other people out there suffering without knowing the cause,
like I was?

Alan Curry

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/