VXVM Mount Error in Service Guard - Urgent

[gajula]

Hi All,

While i am failing over package from one node to other node, We are not able to mount one of the FS, the errors as as follows.

Apr 1 11:35:49 - Node "cmfpts01": Mounting /dev/vg01/lvol1 at /u90
Apr 1 11:35:49 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/u01 at /u01
Apr 1 11:35:49 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/u02 at /u02
Apr 1 11:35:50 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/u03 at /u03
Apr 1 11:35:50 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/u04 at /u04
Apr 1 11:35:51 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/u05 at /u05
Apr 1 11:35:51 - Node "cmfpts01": Mounting /dev/vx/dsk/appl/abb at /abb
UX:vxfs mount: ERROR: V-3-23739: vxfs module is not loaded or mount point not supported inside cluster file system.
ERROR: Function check_and_mount
ERROR: Failed to mount /dev/vx/dsk/appl/abb
Apr 1 11:35:52 - Node "cmfpts01": Unmounting filesystem on /dev/vx/dsk/appl/u05

i tried to seach in this forum but nothing which pointed to my issue, could some one help ??

[Duncan Edmonstone]

what's strange here is that this error is associated with mounting a CFS filesystem in a nested mount point - but looking at the output above you don't seem to be doing that... it would be interesting to see if there are any differnces betrween your mount options... what do you get from:

cfsmntadm display -v /u05

compare with:

cfsmntadm display -v /abb

HTH

Duncan

[gajula]

Hi,

Thanks for the reply,

But I cant find the command cfsmntadm in my server,

This is VxVM configure unders MC service guard cluster,

Regards,
Govind.

[Duncan Edmonstone]

SO you don't have CFS installed on this system? That's even stranger then... care to share the relevant section of the Serviceguard package control file? (i.e. the bit that defines the filesystem mounts)

HTH

Duncan

[gajula]

Hi,

Thanks for your efforts in looking into it,
here is the control script.

#
#VXVM_DG[0]=""

VXVM_DG[0]=appl

#


LV[0]=/dev/vg01/lvol1; FS[0]=/u90; FS_MOUNT_OPT[0]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[0]=""; FS_FSCK_OPT[0]=""; FS_TYPE[0]="vx
fs"

LV[1]=/dev/vx/dsk/appl/u01; FS[1]=/u01; FS_MOUNT_OPT[1]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[1]=""; FS_FSCK_OPT[1]=""; FS_TYPE[1
]="vxfs"

LV[2]=/dev/vx/dsk/appl/u02; FS[2]=/u02; FS_MOUNT_OPT[2]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[2]=""; FS_FSCK_OPT[2]=""; FS_TYPE[2
]="vxfs"

LV[3]=/dev/vx/dsk/appl/u03; FS[3]=/u03; FS_MOUNT_OPT[3]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[3]=""; FS_FSCK_OPT[3]=""; FS_TYPE[3
]="vxfs"

LV[4]=/dev/vx/dsk/appl/u04; FS[4]=/u04; FS_MOUNT_OPT[4]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[4]=""; FS_FSCK_OPT[4]=""; FS_TYPE[4
]="vxfs"

LV[5]=/dev/vx/dsk/appl/u05; FS[5]=/u05; FS_MOUNT_OPT[5]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[5]=""; FS_FSCK_OPT[5]=""; FS_TYPE[5
]="vxfs"

LV[6]=/dev/vx/dsk/appl/abb; FS[6]=/abb; FS_MOUNT_OPT[6]="-o rw,suid,delaylog,datainlog"; FS_UMOUNT_OPT[6]=""; FS_FSCK_OPT[6]=""; FS_TYPE[6]="vxfs"

LV[7]=/dev/vx/dsk/appl/vendor; FS[7]=/vendor; FS_MOUNT_OPT[7]="-o rw,suid,largefiles,delaylog,datainlog"; FS_UMOUNT_OPT[7]=""; FS_FSCK_OPT[7]=""; FS_
TYPE[7]="vxfs"


Regards,
Govind.

[Duncan Edmonstone]

well the only difference I see is the lack of largefiles flag for the /abb filesystem, but that shouldn't prevent it mounting.

What do you see from the following?

fstyp -v /dev/vx/dsk/appl/abb

mkfs -F vxfs -m /dev/vx/dsk/appl/abb



HTH

Duncan

[Duncan Edmonstone]

also confirm that when the package is in a halted state the /abb filesystem isn't mounted... (make sure someone hasn't put it in /etc/fstab)

HTH

Duncan

[gajula]

Hi Duncan,

the package is running on adoptive node at the moment, I did check that on other node. both the control scripts are same,

Even though the file system is not mounted with largefiles option enabled i am sure that should not be a problem,

This FS is mounted and working fine on adoptive node but getting these errors while failback.

Thanks & Regards,
Govinda.

[Duncan Edmonstone]

so stupid I know, but the /abb directory does exist on the adoptive node doesn't it?

aslo check that the mount point isn't busy for other reasons:

fuser /abb

should return

# fuser /abb
/abb:


indicating no processes accessing the directory.

HTH

Duncan

[Leslie Fischer]

Hi,

Try running the processes manually.

Stop your appl, run fuser on all the FS's, unmount the FS's, then deport the dg-group.

Then try to import the dg-group on the failover server and mount the FS's manually?

This has help me when I have had problems.