[Ocfs2-users] Problem mounting volume on second node after uncontrolled reboot.

[Sunil Mushran]

Could be that the device has been claimed by some other entity. Like lvm, multipath, etc. 

On Apr 13, 2011, at 4:08 AM, Asle Næss <Asle.Naess at telio.no> wrote:

> Hi, 
> 
> I have the following setup: 
> 
> 2 linux nodes connected to shared SAN, both nodes have shared OCFS2 volume as /dev/sdb1
> Both servers are running CentOS 5.5
> 
> Node 1 has volume mounted RW.
> Node 2 has volume mounted RO.
> 
> After an unscheduled reboot of node 2 it is no longer able to mount filesystem: 
> 
> # /etc/init.d/ocfs2 start
> Starting Oracle Cluster File System (OCFS2) 
> mount.ocfs2: Device or resource busy while mounting /dev/sdb1 on /data/. 
> Check 'dmesg' for more information on this error.
>                                                          [FAILED]
> 
> dmesg has no information of why it failed.
> 
> I've used strace on mount command and believe this is the relevant problem: 
> 
> mount("/dev/sdb1", "/tmp/sadgfdgerwterfdsvcew/", "ocfs2", 0, "heartbeat=local") = -1 EBUSY (Device or resource busy)
> 
> Note that I've tried mounting it on "/tmp/sadgfdgerwterfdsvcew/" which is a random directory created just before mounting to rule out the possibility of a busy mount point.
> 
> Controlled reboot of node 2 did not resolve problem.
> 
> I believe this show that only node 1 has filesystem mounted: 
> # echo "slotmap" | debugfs.ocfs2 -n /dev/sdb1
>    Slot#   Node#
>        0       1
> 
> All commands shown here is done on node 2. 
> I'm also able to read disk with other tools on server.
> 
> Please advise on how to proceed with debugging and resolving this issue.
> 
> Regards, 
> Asle Ness
> 
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users