Dear All,
It appears that at 14:00 on 18/02/2005 something (mysqld?) went out of
control on my RT server and used all the available memory:
From /var/log/messages:
Feb 18 14:00:04 bogeyman kernel: Out of Memory: Killed process 14270
(mysqld-max).
The sar output also shows that between 13:00 and 14:00 all the swap
disappeared:
bogeyman:/var/log/sa # sar -r -f sa.2005_02_18 -e 16:00:00
Linux 2.6.5-7.111-default (bogeyman) 18/02/05
00:00:01 kbmemfree kbmemused %memused kbbuffers kbcached kbswpfree
kbswpused %swpused kbswpcad
11:00:01 5992 380500 98.45 8348 67184 206388
315716 60.47 26728
12:00:00 6916 379576 98.21 7344 139252 113040
409064 78.35 34188
13:00:00 7824 378668 97.98 9064 145464 24
522080 100.00 43712
14:00:24 5920 380572 98.47 1644 187816 0
522104 100.00 16060
15:00:00 43220 343272 88.82 27736 134072 522104
0 0.00 0
16:00:00 25996 360496 93.27 30432 136144 522104
0 0.00 0
Average: 12146 374346 96.86 37745 70729 206825
315279 60.39 24524
The machine came back up, fsck did it’s thing but now whenever I try to
take ownership of a ticket in RT I get:
Could not change owner.
and an entry in /var/log/messages that reads:
Feb 22 09:38:11 bogeyman RT: Couldn’t delete cached group submember 4689
(/opt/rt3/lib/RT/GroupMember_Overlay.pm:324)
So, it seems I have a database problem, does anyone know how I can fix
this?
I’m using RT 3.2.2, MySQL 4.0.18-32, perl v5.8.3 with fast_cgi on SUSE 9.1
kernel version 2.6.5-7.111-default.
Any advice appreciated.
Many thanks,
Steve.
Stephen Ison
Unix Support
University of Cambridge Computing Service
si202@cam.ac.uk