On 8 Sep 2002, Gilad Ben-Yossef wrote: > On Sat, 2002-09-07 at 22:14, Tzafrir Cohen wrote: > > > > > TThe system probably had enough swap available, and there were no > > OOM-killer messages in the log. Syslog seemed to work and the xinetd > > master process managed to do something. So I tend to rule out any > > out-of-memory situation. Seems like some kernel bug. > > > It does seem like some kind of severe kernel memory leak (ouch!) but I > dio not think ext3 actually has anything to do with it - it's a symptom > not the cause. The messages you see simpyl state that the ext3 code > could not allocate memory. This does not mean that it is it to blame for > not freeing it. > > Can you try to give the output of vmstat on the machine during normal > operation and when it reaches this state?
I'm not sure this is reproducable. In case it is: you mean: just before the system goes bezerk, or when it is already in that state? If the latter: I would like to avoid a script that runs 'vmstat>log' from cron, or even 'vmstat>log;sleep 5 minutes' in a loop, because of the chance there won't be memory to launch vmstat. Could anybody please give me the equivalent perl code? From which files in /proc does vmstat get its data? -- Tzafrir Cohen mailto:[EMAIL PROTECTED] http://www.technion.ac.il/~tzafrir ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
