On Thu, Oct 14, 2010 at 11:42:07AM -0700, Mark Foster wrote: >On 10/14/2010 11:25 AM, Jesse Keating wrote: >> On 10/14/2010 11:06 AM, Glenn Stone wrote: >>> You guys ever seen RHEL4's syslogd just hang inexplicably? RHEL4.8, mysql >>> server, the only oddball thing is we've got a third-party app using syslog's >>> standard "local6" facility to log to a file on an OCFS filesystem. A >>> bog-standard "service syslogd restart" restores functionality quickly and >>> without complaint or trace in the logs, just "restart successful"...
>Glenn, you may want to try using strace(1) on the syslogd process. >That often yields some valuable clue. Just in general or when it's hung? Customer tells me it only happens about once a month, and it's being written to reasonably constantly... (he suggested ten minutes as a valid timeout to raise a monitoring error if nothing had been written to the log in question...) It's one of those pesky things that will inevitably take a long time to debug, but sure puts your five-nines effort in a bind... -- Glenn
