Originally posted by: SystemAdmin

We have an HACMP 6.1 Cluster in the production site and a backup server in the DR site. We use EMC Recoverpoint for replication (each LUN is replicated to a second LUN in DR site). The HACMP VGs are of type Enhanced Concurrent. On the backup server when we try to mount filesystems, we get this error :

1. mount /echange

mount: /dev/echangelv on /echange
0506-365 Cannot mount guarded filesystem.
The filesystem is potentially mounted on another node.

(importvg, varyonvg operations are OK)

The problem does not arise with normal VG type (not Enhanced Concurrent).

We must varyoffvg and umount filesystems of the production site to mount those of the DR site.

We have opened a call with EMC support; they said it’s an IBM AIX issue.

Thank you for your help

Originally posted by: SystemAdmin

The solution is very simple :

1. mount -o noguard /echange

from man mount :

noguard

Mount the filesystem regardless of the current mountguard setting which would otherwise guard the filesystem against unsupported concurrent mounts in a PowerHA or other clustering environment. If mountguard is enabled by the chfs or crfs command, the filesystem cannot be mounted if it appears to be mounted on another node or system. Specifying the noguard option temporarily overrides the mountguard setting.

Originally posted by: cplarre

In my case I was not using EMC Recoverpoint, instead I am cloning the LUNs with EMC snapview commands sent to CLARiiON storage. The volume groups were not imported with importvg, instead the pvids were reset on hdiskpower devices (chdev -l hdiskpowerXX -a pv=clear ; chdev -l hdiskpowerXX -a pv=yes) and a recreatevg command was issued specifying the complete list of hdiskpower devices. The issue was fixed running an fsck on the filesystems and mounting them ro (read only) later.

Originally posted by: de78

Hi All, I experienced exactly the same issue, so thank you for having solved it.

Just two questions about this:

1) What's the difference between importvg and recreatevg? I tried both commands and had the same result.

2) Both commands modify the file systems mount point adding them a /fs, for example /data file system become /fs/data

Is there a way to avoid that?

Originally posted by: SiddharthSingh

personally I prefer importvg over recreatevg .

it seems, recreatevg basically calls importvg at backend .

but I really hate when recreatevg prefixes /fs/ before every LV/FS recreated on target node .

to avoid pre-fixing '/fs' , I had to rename VG and the filesystem.
Would really appreciate if someone could suggest any solution to avoid this irritating feature .

 

loving importvg,

Sidd

 

Originally posted by: wycky

Hallo,

 

recreatevg -y vgname -Y NA -L / hdiskx 

 

Steven