AIX

Don't know what the exact name is with HP EVA but I'd check next that the hdisk4 is unreserved using SCSI3 protocol. With IBM Total storage systems the attribute is reserve_policy=no_reserve and with EMC Clarrion the attribute is reserve_lock=no.

I ended up adding the hdisk4 outside on c-spoc.
I did a extendvg to add the hdisk4 to the volume group. Once added, I go back in to smit hacmp & saw the disk.

If I got you correctly you did an extendvg of the shared VG on one cluster node. If that was the case your cluster is out of sync now. It may be that the cluster will re-sync itself during failover - however there is no guarantee that the cldare operation succeeds. The cluster might as well stop processing in the mid of a failover. The more unpropagated changes you collect the more likely is that the cluster won't work as expected when you need it. Better run a verification and synchronisation on the cluster and / or even better a takeover test.

I did a "Synchronize of volume group" after I added the disk from the active node using extendvg.

I then did a "Mirror a Shared Volume Group" from c-spoc. After the mirroring I did a "Synchronize Shared LVM Mirrors".

I performed the failover of cluster to the other nodes & everything looks ok.

Hi,

I would do it that way, it works almost online, but you lose the cluster functionality for the time of migration, the big advantage: you don't need to use cspoc, you can work with standard lvm-tools

stop hacmp on all nodes, make sure the vg is offline everywhere

exportvg on all nodes, backup /etc/filesystems, expect the node you want to do the migration on



go on that node, run varyonvg and mount all filesystems

asign the cluster ip manually, for example with smitty inetalias

this takes you about 5minutes

start the application if you want, else you don't need to mount filesystems or assign the cluster ip


now work with the vg like in non clustered envrionment, extendvg, mklvcopy and so on


when the data is migrated, varyoffvg and remove the cluster ip

read the new disks on all nodes, und use importvg, IMPORTANT: Volume Group major Number must stay the same, check with ls -l /dev/vgname
use importvg -V majornumber -y vgname hdiskX

restore /etc/filesystems



varyoff all vgs

verify and synchronise the cluster best with the option to automatically correct errors, else it will fail because of the changes vgda timestamp

now you can start the cluster





i think I would do it that way

hello,

Did anyone face this strange problem with POWER HA 5.5 ( HACMP )

Two Node Configuration
- Mirrored ROOTVG on both nodes
- HDISK0 with HDISK1
- Broke the Mirror on HDISK0
- Made HDISK1 bootable and changed the bootlist to include only HDISK1
- Deleted HDISK0
- Rebooted the systems

After this I can't start HACMP if the other node is off. If the both nodes are online then only I can see --> smitty cl_start screen otherwise if one node is down and the other is up, when i type --> smitty cl_start ; i get a message...additional software needs to be installed....... etc.etc..

then I get strange messages in my log file

my solution was to re-install AIX and re-install HACMP.

is there any other solution ?