1825891
Members
3015
Online
109689
Solutions
SOLVED
you only run "vgchange -a s ..." if this is a SHARED logical volume used with SLVM (usually as part of a RAC cluster). If this is just a standard Serviceguard cluster where a logical volume is only ever active on one node in the cluster at any one time, you can omit this step.
HTH
Duncan
I am an HPE Employee
HTH
Duncan
I am an HPE Employee
...seems to be a lot of poor advice being handed out on this thread...
first off, when you are running the vgexport on your primary node to generate the map file, you need to make nsure you include the "-s" option on the vgexport. This will ensure that as well as the lvol names, the volume group ID is also writrten to the map file (think of the volume group ID or VGID as like a disk signature)
Then when importing the volume group on the other node use the "-s" option on the vgimport and the vgimport command will scan all the disks visible to the system and look for that VGID - that way it knows what disks need to be part of the volume group and you don't have to list them on the command line.
As for the point about device files being the same on all cluster nodes. In my experience they are different as often as they are the same. Serviceguard really doesn't care if a LUN is seen as c3t5d0 on one node and c7t5d0 on another node - it can cope with that without a problem.
HTH
Duncan
I am an HPE Employee
first off, when you are running the vgexport on your primary node to generate the map file, you need to make nsure you include the "-s" option on the vgexport. This will ensure that as well as the lvol names, the volume group ID is also writrten to the map file (think of the volume group ID or VGID as like a disk signature)
Then when importing the volume group on the other node use the "-s" option on the vgimport and the vgimport command will scan all the disks visible to the system and look for that VGID - that way it knows what disks need to be part of the volume group and you don't have to list them on the command line.
As for the point about device files being the same on all cluster nodes. In my experience they are different as often as they are the same. Serviceguard really doesn't care if a LUN is seen as c3t5d0 on one node and c7t5d0 on another node - it can cope with that without a problem.
HTH
Duncan
I am an HPE Employee
Ganesan's post was closest to correct - I've altered it slightly (I don't know what your VG is called or what VG minor numbers you use so you'll need to adjust those for your own system):
on node1
Create the logical volume as per LVM commands. After creation,
#vgexport -p -s -v -m /tmp/vg02.map vg02
It will just create a map file /tmp/vg02.map
Copy this map file to second node.(node2)
on node2. (Assuming VG02 is not activated here)
Note down the minor number of vg02
#vgexport /dev/vg02
#mkdir /dev/vg02
#mknod /dev/vg02/group c 64 0x020000
#vgimport -v -s -m /tmp/vg02.map vg02
It will take a little longer as it has to scan all the disks on the system, but should work.
HTH
Duncan
I am an HPE Employee
on node1
Create the logical volume as per LVM commands. After creation,
#vgexport -p -s -v -m /tmp/vg02.map vg02
It will just create a map file /tmp/vg02.map
Copy this map file to second node.(node2)
on node2. (Assuming VG02 is not activated here)
Note down the minor number of vg02
#vgexport /dev/vg02
#mkdir /dev/vg02
#mknod /dev/vg02/group c 64 0x020000
#vgimport -v -s -m /tmp/vg02.map vg02
It will take a little longer as it has to scan all the disks on the system, but should work.
HTH
Duncan
I am an HPE Employee
weertman
about volume corruption...
yes you're corerct it wouldn't actually be "corrupted" it just wouldn't work correctly - either the VG wouldn't activate or the new lvol would be missing. But there would be no actual data corruption.
I guess it depends on the interperetation of the word "corruption". You could ceratinly say that the volume group configuration was "corrupted" as it would be missing data, but the actual contents of your logical volumes and filesystems wouldn't get corrupted.
HTH
Duncan
I am an HPE Employee
about volume corruption...
yes you're corerct it wouldn't actually be "corrupted" it just wouldn't work correctly - either the VG wouldn't activate or the new lvol would be missing. But there would be no actual data corruption.
I guess it depends on the interperetation of the word "corruption". You could ceratinly say that the volume group configuration was "corrupted" as it would be missing data, but the actual contents of your logical volumes and filesystems wouldn't get corrupted.
HTH
Duncan
I am an HPE Employee