12-14-2011 12:54 AM
Hello,
I'm using two RHEL6.1 servers with veritas 5.1SP1PR2 cluster file system. Cluster starts ok and I can switch services to other node etc but when I execute commands hastop -local or hastop -all then the cluster hangs.
It even makes one of the nodes to hang totally and no commands cannot be executed as root.
What I can see is that it hangs when it does the cfsumount for shared mount points. Some of them (changes time to time so not the same mount points) are hanging and appead again to the mount list but not as real mounts but as copies of root /. example below
-bash-4.1# df -k
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
8256952 5710140 2127384 73% /
tmpfs 18503412 424 18502988 1% /dev/shm
/dev/sda1 253871 37102 203662 16% /boot
/dev/mapper/VolGroup00-LogVol04
12385456 863508 10892804 8% /opt
/dev/mapper/VolGroup00-LogVol02
8256952 428260 7409264 6% /tmp
/dev/mapper/VolGroup00-LogVol03
12385456 5429820 6326492 47% /var
tmpfs 4 0 4 0% /dev/vx
/dev/vx/dsk/mountdg/mymount
8256952 5710140 2127384 73% /mymount
From here you can see that the data values are the same between my cfsmount (mymount) and root /. I can survive from this using command umount /mymount
It throws an error
umount: /mymount: not mounted
but still ater this cluster continues to go down. This is just a workaround and I do not want to leave it like this.
Any ideas how to fix this? Is there a patch for this or should I change something on RHEL or in veritas?
br, JP
12-15-2011 12:56 AM
Please post engine_A.log? You can find it in /var/VRTSvcs/log/
/var/log/messages might also help.