Great !

Thanks Mike. It seems this was the reason.

Short summary of what I did: 

1) Enabled Comstar on the target node
2) Disabled legacy iscsi on the target node
3) Backed up the STMF config on source node (pre 116) 

svccfg export -a stmf > /tmp/stmf.cfg

4) Copied over the file to the other node

scp /tmp/stmf.cfg target:/tmp/stmf.cfg

5) Cleared the Comstar Configuration on the target node

svcadm disable stmf
svccfg delete xtmf
cd /var/svc/manifest
svccfg import system/stmf.xml
svcadm disable stmf

6) Imported the Configuration on the target node

svccfg import /tmp/stmf.cfg

7) Export the pool on the primary side

zpool export mypool 

8) Import the pool on the secondary side

zpool import mypool

9) Imported the lu

stmfadm import-lu /dev/zvol/rdsk/mypool/vzol1

10) Verified mappings 

stmfadm list-lu 
stmfadm list-view -l LU 

11) DONE 

Everything was taken over. Mappings available and the GUID did not change, 
which is importand for linux device mapper multipath clients woth persistent 
mappings.

Also the content stayed the same (verified with md5sum). Alignment is also 
correct (verified with md5sum on the device). All mappings are there also. 

I even tried this with Linux devicemapper client with "queue_if_no_path" and 
switched a virtual IP between the nodes. During the operations above the path 
fails (because of queue_if_no_path this is not critical), but after the IP is 
up (as a last action), paths are online again - 100% transparently. 

Thanks again for the Tipp Mike. 

Regards, 
Robert
-- 
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to