This post is more than 5 years old
15 Posts
0
5103
July 28th, 2017 15:00
moving iscsi connections from old server to new
We encountered some issues, so this is more of a question of whether Dell requires/recommends a different approach to moving EQ volumes from one server to another.
We attempted to replace an old 2003 server with a 2008R2 server this past week. The server has close to 200 separate volumes.
My process was to disable (not stop) the iSCSI Initiator service on the old server so that it wouldn't disconnect any volumes while the server was up, but it wouldn't attempt to reconnect any volumes upon reboot. I changed the name of the server at the same time so we could reuse that name for the replacement server. After rebooting, both SAN HQ and Group Manager confirmed there were no connections to any volumes on the SANs.
After changing the name of the replacement server and rebooting, I added the target portal IP and manually connected all 200 volumes.
At that point, the disks were all showing in Disk Management as "offline". I started bringing them online one by one. After around 20 volumes, Disk Management stopped responding. I waited for 15 minutes or so, then tried killing the VDS service to see if it would recover. After another 5 minutes or so, I was able to get Disk Management to load again. I brought maybe 4 more disks online before Disk Management stopped responding again and never recovered until a reboot (which messed up all of my mount point paths which hadn't been bound yet in iSCSI Initiator). The mount point paths originally had been carried over from the old server. This worked fine in my test environment several times.
From Dell's perspective, is there anything I should have done differently? I've mostly only seen suggestions about making sure all old connections were gone before making connections from a different server. And I believe my method accomplished that. Any tips/suggestions are welcome.
Thanks



jesseroscoe
15 Posts
0
July 28th, 2017 18:00
These are not volumes from a cluster, just a single server. I wanted the old server to still be accessible for a while on the network with it's new name, just without the iSCSI disks. And since the replacement server used the same name and IQN as the old server (we use IQN for access control), I didn't have to make any ACL changes on the volumes. I verified all volumes were showing zero connections before connecting them to the new server.
I felt like this should have been a safe approach, but I just wanted to make sure there wasn't something I needed to do differently.
Thanks
jesseroscoe
15 Posts
0
July 28th, 2017 19:00
Good to know. Thank you.