06-16-2017 07:41 AM
Hi all,
I have a very hard scenario in my company:
AS-IS
- two node SFHA 5.1SP1RP4 and LINUX 6.5 installed
TO-BE
- SFHA 6.2.1 and Linux 7.2
Which is the best practice to doing this with minimal outage:
Reinstalling with new release and restore vcs/vxvm configuration, or is it possible to add new cluster node with different os and switch services and reinstalling the second nod with new os? I don't known that configuration is supported, for me should be first time this complex environment.
Best regards
Rick
Solved! Go to Solution.
06-17-2017 01:19 AM
HAD does not support connections from a different version of binaries, engine_A.log will show that and won't allow a node with different VCS version to join the cluster and you can't install VCS 5.1 on RHEL 7.2. With these limitations you have limited options especially if you are looking to minimise the downtime
The best option I see is to upgrade SF/HA to 6.2.1 in-place with RHEL 6.5, then build a new temporary node with RHEL 7.2 (or 6.5) and SF/HA 6.2.1 (you will need some patches to get it working but it is supported), join it to the cluster (this part is not supported if you use RHEL 7.2 but will work as long as HAD binary version is the same across all nodes), failover the workload to the new node and rebuild remaining nodes with RHEL 7.2 and SF/HA 6.2.1. Then failback the workload to RHEL 7.2 nodes and remove the temporary node from the cluster.
If you want a fully supported process, then just build a parallel system on new nodes with RHEL 7.2 and SF/HA 6.2.1 and switch the disks over to it.
I've done a few similar upgrades recently and both options worked fine.
06-17-2017 01:19 AM
HAD does not support connections from a different version of binaries, engine_A.log will show that and won't allow a node with different VCS version to join the cluster and you can't install VCS 5.1 on RHEL 7.2. With these limitations you have limited options especially if you are looking to minimise the downtime
The best option I see is to upgrade SF/HA to 6.2.1 in-place with RHEL 6.5, then build a new temporary node with RHEL 7.2 (or 6.5) and SF/HA 6.2.1 (you will need some patches to get it working but it is supported), join it to the cluster (this part is not supported if you use RHEL 7.2 but will work as long as HAD binary version is the same across all nodes), failover the workload to the new node and rebuild remaining nodes with RHEL 7.2 and SF/HA 6.2.1. Then failback the workload to RHEL 7.2 nodes and remove the temporary node from the cluster.
If you want a fully supported process, then just build a parallel system on new nodes with RHEL 7.2 and SF/HA 6.2.1 and switch the disks over to it.
I've done a few similar upgrades recently and both options worked fine.
06-19-2017 01:04 AM
Hi,
Thanks for your response, I have just only one question:
You suggest to adding a new node to cluster to support upgrading, but if I shoud to run the cluster with one node and reinstall other node to RedHat 7.2...what about this?
Best Regards
Rick
06-28-2017 03:07 AM - edited 06-28-2017 03:09 AM
Yes, as long as you can have the same version of VCS across all systems. As I said before, you can't use 5.1 on RHEL 7.2 so you need to find a version that works on both old RHEL and new RHEL