HW-Upgrade, replace 860c i2 with 860c i6 with minimum downtime

[Christian Schulze]

Hi,

I just want to get feedback about my steps to Upgrade my existing Blade-HW.

Scenario:

2-Node Cluster with BL860c i2 in two different C3000 Enclosures. Each enclosure has Storages (P2000 and MSA2040) connected, mirroring between these is done via LVM.

New HW is very similar to old one, regarding number of mezzanine-cards

My approach would be:

1. move all Applications to Node1, freeing up Node2 (Means a short downtime, so needs to be scheduled=. Disable automatic clusterstart on node2.
2. create an Ignite Image of Node2 to Ignite-Server.
3. shut down Node2 and unplug it from C3000
4. plug the new HW-Blade in the same slot
5. reconfigure SAN-Connections in the FC-Controllers of the C3000 to reflect the new HW
6. Passthrough does not need to be reconfigured for LAN-Ports
7. boot new HW from Igniteserver
8. check if everything is OK
9. join this node to cluster, check ...
10. move all apps to the new node (second downtime)
11. repeat steps on Node1

Did I miss something important ?

Would it be an option, to switch internal diskdrives from old HW to new HW instead of Ignite ? Thus minimizing downtime. Then when Cluster is working, replace internal diskdrives one by one with the new ones.

Thanks,

Christian

[Torsten.]

I think if you install the latest HW patches first, booting the new server from the old disks should work.

[Christian Schulze]

Thanks,

this is good news,

HPUX is pretty fresh, update_ux was run in August, only the latest Patchupdate (QPKG from Sept/Oct 2017) is not yet applied.

I think we go the disk-swapping-way. Because we are in some sort of pressure to switch the Nodes. Load on this Cluster was increasing so fast, that we can run on one Node only for a short time (some days max with issues).

Christian