[ILUG] Corrupted Process(?)
gavinmcc at netsoc.ucd.ie
Wed May 9 10:59:01 IST 2001
need a little help with this if anyone knows. We have a machine
running Debian 2.2, kernel 2.4.2, dual cpu PIII. It's used for both
computation and general surfing/email etc.
We have two long running computations running on it, which have a
few weeks left and cannot be stopped, so a reboot is out. It would appera
that Netscape crahseed on one of the users and left a corrupted or
unreadable /proc entry. As a result, ps and top both freeze. Whenever
the directory in /proc is read, the process reading it seems to freeze.
As a result we now have a load avg of 40:
gavin at buggeredmachine:/home/gavin$ uptime
10:54am up 50 days, 9:07, 17 users, load average: 41.09, 41.22, 41.24
All else, seems to be fine, but we'd like to correct this without
a reboot. Do you think the cpu is wasting much time on the corrupted
processes. Do you think the computation (nice 19) is still using most of
the cpu or not (it normally would use 80%)? Can anyone see what has
caused this, and in particular, does anyone have a solution. Does it seem
likely that netscape could do this?
Thanks in advance, mail me back if you want any more info. Please
cc my address as I am only on the summary email thingy.
More information about the ILUG