Anyone running > 2.2.14pre15 kernels with syslog receiving remote logs? Anyone willing to try?
I've been trying, unsuccessfully, to run kernels > 2.2.14pre15 on a
box that's receiving remote syslog information.
After a relatively short period of time, the Recv-Q in netstat for the
syslog port starts to get big (80K or so), and the box starts to flake
out as syslog seems to get clogged up on /dev/log as well (local
messages are seriously delayed, entered out-of-order), processes seem
to hang waiting to write syslog info, etc.
Kill syslog, and suddenly it is un-constipated.
I've reported this on linux-kernel, but I thought it might be useful
to ask if anyone else has seen this---Alan Cox's already gotten back
to me, but I don't know how far Debian and RedHat's syslogs may have
diverged, so I thought it worthwhile to see if anyone has/can
reproduce this with different hardware.
BTW, syslog-ng seems to take care of this problem, but I can't get my
other boxes to log to a remote host, it complains about "resource
temporarily unavailable" when trying to bind AF_INET. or some such.
Haven't had a chance to take the message down verbatim.
Mike.
Reply to: