Home » RDBMS Server » Backup & Recovery » Server crash when exporting database
Server crash when exporting database [message #192655] |
Wed, 13 September 2006 03:00  |
aline
Messages: 92 Registered: February 2002
|
Member |
|
|
Hi,
My configuration:
2 server (production and standby) with 8GB of RAM
linux rh4
oracle9.2.0.7
My problem
all night, I run an export of the production database on my standby site. Sometimes (maybe one per week) my server crash when I run the export.
The command of the export is very simple:
$EXP user/PWD FILE=$SAVELOCALDIR/${USER}1.$BACKUPFILENAME,$SAVELOCALDIR/${USER}2.$BACKUPFILENAME,$SAVELOCALDIR/${USER}3.$BACKUPFILENAME,$SAVELOCALDIR/${USER}4.$BACKUPFILENAME,$SAVELOCALDIR/${USER}5.$BACKUPFILENAME,$SAVELOCALDIR/${USER}6.$BACKUPFILENAME,$SAVELOCALDIR/${USER}7.$BACKUPFILENAME,$SAVELOCALDIR/${USER}8.$BACKUPFILENAME,$SAVELOCALDIR/${USER}9.$BACKUPFILENAME,$SAVELOCALDIR/${USER}10.$BACKUPFILENAME,$SAVELOCALDIR/${USER}11.$BACKUPFILENAME,$SAVELOCALDIR/${USER}12.$BACKUPFILENAME,$SAVELOCALDIR/${USER}13.$BACKUPFILENAME ,$SAVELOCALDIR/${USER}14.$BACKUPFILENAME,$SAVELOCALDIR/${USER}15.$BACKUPFILENAME,$SAVELOCALDIR/${USER}16.$BACKUPFILENAME,$SAVELOCALDIR/${USER}17.$BACKUPFILENAME,$SAVELOCALDIR/${USER}18.$BACKUPFILENAME LOG=$DIRSAVELOG/$USER.$BACKUPNAMELOG FILESIZE=1900MB owner=$USER consistent=Y direct=Y statistics = none
The server crashed with this message:
kernel panic: out of memory and no killable processes.
The size of the sga on the standby site is about 4Gb
SQL> show sga
Total System Global Area 4816866536 bytes
Fixed Size 748776 bytes
Variable Size 301989888 bytes
Database Buffers 4513071104 bytes
Redo Buffers 1056768 bytes
and of course, on this server, their is nothing else than those processes.
I asked metalink with this problem but they have no idea about. So you are my last chance!
|
|
|
|
|
|
|
Re: Server crash when exporting database [message #192778 is a reply to message #192737] |
Wed, 13 September 2006 09:20   |
aline
Messages: 92 Registered: February 2002
|
Member |
|
|
bad news!
I followed the instructions of Werner pushitz to change my semaphores parameters (i.e. 60 58880 460 128).
I test an new export and my server crashed!
The last messages I saw was onto /var/log/messages
Sep 13 15:34:12 oracle-reco kernel: warning: many lost ticks.
Sep 13 15:34:12 oracle-reco kernel: Your time source seems to be instable or some driver is hogging interupts
Sep 13 15:34:12 oracle-reco kernel: rip __do_softirq+0x4d/0xd0
Sep 13 15:42:08 oracle-reco kernel: oom-killer: gfp_mask=0x1d2
Sep 13 15:42:08 oracle-reco kernel: Mem-info:
Sep 13 15:42:08 oracle-reco kernel: Node 1 DMA per-cpu: empty
Sep 13 15:42:08 oracle-reco kernel: Node 1 Normal per-cpu:
Sep 13 15:42:08 oracle-reco kernel: cpu 0 hot: low 32, high 96, batch 16
Sep 13 15:42:08 oracle-reco kernel: cpu 0 cold: low 0, high 32, batch 16
Sep 13 15:42:08 oracle-reco kernel: cpu 1 hot: low 32, high 96, batch 16
Sep 13 15:42:08 oracle-reco kernel: cpu 1 cold: low 0, high 32, batch 16
Sep 13 15:42:08 oracle-reco kernel: cpu 2 hot: low 32, high 96, batch 16
Maybe I set wrong semaphores parameters, or something else is wrong, I don't know...
any other idea?
|
|
|
|
Goto Forum:
Current Time: Fri May 02 15:51:02 CDT 2025
|