[Ocfs2-users] Periodic hangs

[Emil Noether]

Hi,

I have a SATABoy2 Nexan storage with 8 disks (SATA Hitachi 
HUA721075KLA330) connected to raid 6.  Two image servers and two 
webservers. Image servers are connected to storage via iSCSI (1GBit) and 
webservers are connected via fibre (QLogic ISP2432-based 4Gb). There is 
ocfs2 filesystem on the storage disk. When I disconnect webserver1 
(identical with webserver2) everything is ok. But when I do 
"/etc/init.d/o2cb start", even without mounting the storage disk (so 
webserver is actually doing nothing) my project is down every aprox 30 
minutes for aprox 2 minutes.

To describe what is down: There is no problem on image servers, but 
there is a problem on webserver2. Mounted ocfs2 disk is not responding 
(I can't run even "df" command), so load goes to aprox 400 and number of 
running apaches reaches it's maximum and so on. The web page is not 
responding.

I store all of my logs on local disks so not on ocfs2 disk.

I use 2.6.32 kernel on servers, but I have already tried change it to 
some another, but with no result.

I use ocfs2-tools in version 1.4.1-1.

My distro is Debian Lenny (5.0.6) x64.

My /etc/default/o2cb:
O2CB_ENABLED=true
O2CB_BOOTCLUSTER=ocfs2
O2CB_HEARTBEAT_THRESHOLD=14
O2CB_IDLE_TIMEOUT_MS=10000
O2CB_KEEPALIVE_DELAY_MS=5000
O2CB_RECONNECT_DELAY_MS=2000

My /etc/ocfs2/cluster.conf:
node:
   ip_port = 7777
   ip_address = 10.0.0.111
   number = 0
   name = www1
   cluster = ocfs2

node:
   ip_port = 7777
   ip_address = 10.0.0.112
   number = 1
   name = ww2
   cluster = ocfs2

node:
   ip_port = 7777
   ip_address = 10.0.0.121
   number = 2
   name = img1
   cluster = ocfs2

node:
   ip_port = 7777
   ip_address = 10.0.0.122
   number = 3
   name = img2
   cluster = ocfs2

cluster:
   node_count = 4
   name = ocfs2


Any help is very appreciated,
Best Regards,

Emil Noether
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20101015/e56be2cf/attachment.html