Configure HA between Version 8 node and version 9 node

zandhaas

New Member
Aug 19, 2024
9
2
3
I have a home lab with two proxmox nodes in a cluster. one node is installed with version 8.4.1 and the other with version 9.05.
I have a "Quorum" docker container on my Synology for the third quorum.

I now want to configure HA between the two nodes. I am logged in on the version 9 node to configure all off this. But it seems not to work. If I configure the "start" or "Stop" in the Request state field the HA rule ends in the "Freeze" state.

I have configured the HA resources for the test CT i'm using for this test. And also crested a affinity rule for this CT.

When I try to simulate a failover by stopping the CT (HA rule in stopped state). Nothing seems to happen.

Is this behaviour because I use a 8.4.1 node and a 9.05 node or do I mis something during the configuration?
 
When I try to simulate a failover by stopping the CT (HA rule in stopped state). Nothing seems to happen.

When you stop a Ressource via the normal mechanism HA knows that. You ordered it to stop and this wish is obeyed.

To test HA: pull the network cord of the hardware node which currently runs the Ressource in question, in "Request State: started". After a minute (or two) two things should happen:
  • that node will "fence" itself = it reboots hard - make sure to shutdown (or migrate) everything in beforehand, except the proband
  • that Ressource will get started on the other node
 
  • Like
Reactions: Johannes S
When you stop a Ressource via the normal mechanism HA knows that. You ordered it to stop and this wish is obeyed.

To test HA: pull the network cord of the hardware node which currently runs the Ressource in question, in "Request State: started". After a minute (or two) two things should happen:
  • that node will "fence" itself = it reboots hard - make sure to shutdown (or migrate) everything in beforehand, except the proband
  • that Ressource will get started on the other node
Thank you for your hints. While running the version 8 node combined with the version 9 node in one cluster it did not work. THe nodes I configured for HA ended all in the HA-Freeze status. and the failover did not work. I dicided to do the upgrade from pve 8 to pve 9 earlier as planned. So no both nodes are running PVE 9. And now HA is working perfectly. When I disconnect the new node with just one lxc on it for now this lxc autmaticaly is migrated to the remaining node. And I can also use the migrate command to migate lxc's without first replicating them.

So i short it seems the differences between version 8 and 9 are to big for at least the HA function to work correctly in a (very small) cluster environment