Krishna Krishnapriyan
2008-Apr-16 15:11 UTC
[Ocfs2-users] ocfs2 processes that need to be up
Hi, Any information on the following situation will be appreciated. I am not on the ocfg2-users mailing list. So, please reply to me directly. We run our install rests on the farm rac clusters and we have been checking whether ocfs2 is up by looking for processes [ocfs2vote-x], [ocfs2cmt-x], [ocfs2vote-x], [ocfs2cmt-x],[ocfs2_wq] where x is a digit. Our observation was that x was usually 0 but sometimes a larger digit. Of late we have been noticing that x appears to be a larger multi-digit number. Questions: What is the significance of the specific digit/number x? Is it supposed to be within some range for ocfs2 to be working correctly? Rather than doing a check of the processes to verify that ocfs2 is up, is there a command that can be run, that verifies the health of ocfs2? Thanks, --Krishna
An HTML attachment was scrubbed... URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20080416/4b5ef3c6/attachment-0001.html
this is a ocfs2 superblock id that gets incremented for each ocfs2 mount. It gets reset whenever ocfs2 modules are re-loaded(on restarting the node or /etc/init.d/o2cb restart). This is expected behavior and will not cause any problems until it grows beyond 64K. thanks, --Srini Madhu Velukur wrote:> Srinivas - To validate ocfs2 process and setup on any given set of > cluster nodes, we do following validation to make sure ocfs2 is > working as expected before doing clusterware install: > > ------------------------------------------------------------------------ > # /sbin/lsmod | grep ocfs > ocfs2 351172 2 > debugfs 14216 2 ocfs2 > ocfs2_dlmfs 27144 1 > ocfs2_dlm 199176 2 ocfs2,ocfs2_dlmfs > ocfs2_nodemanager 159360 26 ocfs2,ocfs2_dlmfs,ocfs2_dlm > configfs 28044 2 ocfs2_nodemanager > jbd 59609 2 ocfs2,ext3 > > # *ps -efw | grep ocfs* > root 4232 12 0 2007 ? 00:00:00 [ocfs2_wq] > root 4233 12 0 2007 ? 00:00:51 [ocfs2vote*-0*] > root 4238 12 0 2007 ? 00:00:00 [ocfs2cmt*-0*] > root 4243 11 0 2007 ? 00:00:00 [ocfs2vote*-1*] > root 4248 11 0 2007 ? 00:00:00 [ocfs2cmt*-1*] > -------------------------------------------------------------------------- > > From past few weeks, we are observing a waived behavior on ocfs > process out, it shows multi-digit for each daemon process and this > count is getting increased for each node reboot. We do node reboot [as > and when required] as WA to overcome bug #5960630 that we had > discussed in past. Can you let know whether below o/p is expected > behavior? > > # ps -ef | grep ocfs2 > root 4309 12 0 Apr15 ? 00:00:00 [ocfs2_wq] > root 19441 13 0 05:59 ? 00:00:00 [ocfs2vote-*52*] > root 19446 12 0 05:59 ? 00:00:00 [ocfs2cmt-*52*] > root 19454 11 0 05:59 ? 00:00:00 [ocfs2vote-*53*] > root 19459 11 0 05:59 ? 00:00:00 [ocfs2cmt-*53*] > > Above problem is observed on following farm nodes - stbcf01, stbcf03, > stbcf05, stbcf07, stbcf11, stbcf13, stbcf15, stbce09, stbce13 > > Thanks, > Madhu > > Krishna Krishnapriyan wrote: >> Hi, >> >> Any information on the following situation will be appreciated. I am >> not on the ocfg2-users mailing list. So, please reply to me directly. >> >> We run our install rests on the farm rac clusters and we have been >> checking whether ocfs2 is up by looking for processes [ocfs2vote-x], >> [ocfs2cmt-x], [ocfs2vote-x], [ocfs2cmt-x],[ocfs2_wq] where x is a >> digit. Our observation was that x was usually 0 but sometimes a >> larger digit. Of late we have been noticing that x appears to be a >> larger multi-digit number. >> >> Questions: What is the significance of the specific digit/number x? >> Is it supposed to be within some range for ocfs2 to be working >> correctly? Rather than doing a check of the processes to verify that >> ocfs2 is up, is there a command that can be run, that verifies the >> health of ocfs2? >> >> >> Thanks, >> --Krishna >