file system super-block update failed when mounting a lun on 2 systems

[scytar]

Hello,

i present a lun (DMX-3 Disk Array) to 2 servers (HP 11iv3):
- the first server use it in read-write mode (vgchange -a y, mount -o=rw)
- the second is in read only mode (vgchgid on pv for more than one copy,vgchange -a r,mount -o=ro)

when activating vg in read-only mode on the second server, the first one is disabling the corresponding FS:

*** bdf gives:

/toto: I/O error

*** syslog gives:

Nov 18 11:36:38 hptoto vmunix: vxfs: WARNING: msgcnt 62 mesg 025: V-2-25: vx_wsuper - /dev/vgtoto/lvtoto file system super-block update failed
Nov 18 11:36:38 hptoto vmunix: vxfs: WARNING: msgcnt 63 mesg 031: V-2-31: vx_disable - /dev/vgtoto/lvtoto file system disabled


what's wrong with my method ?

[likid0]

vgchgid on pv for more than one copy??

if you change the vgid of the disk, then on the firdt node is not going to recognize the disk as part of the vg.

you can use vgchgid when using cloned disks(no the same disk) after a flashcopy or BCopy.

if you wan't to have both vg active, with service guard, you can do:

on your first node vgchange -a e /dev/vg and mount it rw

On the second node vgchange -a r /dev/vg mount it read-only

[Emil Velez]

Dont do it.. YOu cannot read a filesystem read write on one node and read only on another node unless it is a cluster file system. Node 1 will update a file sysetm structure and node 2 will have invalid structures

you can use LVsplit to split a mirrored LV and mount the split off copy to another node or you can createa snapshot in the disk array and mount the unique snapshot lun to another node. What you want to do will not work.

[scytar]

In my case, the lun is a clone of a clustered filesystem. I have to use vgchgid because i have 2 clones of the same source to present to the second server at the same time. So is my method still wrong ?

[scytar]

is vgchgid really modifying informations on disks or is it only a server-side information located in pv definition ?