cluster with quorum disk, if it breaks cluster stops working

[mgiammarco]

Hello,
I have followed your guides and built a two servers cluster with quorum disk.
Unfortunately my quorum disk has broken yesterday.
I now see that in the servers rgmanager does not start anymore and so VMs cannot start.
So the quorum disk now is a single point of failure.
I need that qdisk is only a tie breaker, how can I do it?

Thanks,
Mario

 

[tom]

Hello,
I have followed your guides and built a two servers cluster with quorum disk.

Just to note, the recommendation is NOT a two node with quorum disk - use at least 3 nodes for HA.

Unfortunately my quorum disk has broken yesterday.
I now see that in the servers rgmanager does not start anymore and so VMs cannot start.
So the quorum disk now is a single point of failure.
I need that qdisk is only a tie breaker, how can I do it?

Thanks,
Mario

fix the quorum disk or why is this not possible?

 

[mgiammarco]

Just to note, the recommendation is NOT a two node with quorum disk - use at least 3 nodes for HA.



fix the quorum disk or why is this not possible?

If I have three nodes I do not need a quorum disk as tie breaker.

I am fixing the quorum disk but, while I am fixing it, the cluster is down...
So, tell me, what is the purpose to buy two servers to have an HA cluster if at first little problem all cluster is down?

 

[Raymond Burns]

But your cluster is down with a important broken piece of a not recommended 2 - node HA Cluster. How much do you expect from a workflow that is stated as not 100% efficient.
The purpose, as I see it, as long as all the parts are working in their respective manner, and the node fails, the quorum be there to bring the other server up.
It seems to me that the quorum disks is acting as some type of master for the cluster. If it's down, everything is down. What's wrong with establishing a 3-node cluster now? You see the ill effects from the workflow that's not recommended.

 

[tom]

If I have three nodes I do not need a quorum disk as tie breaker.

I am fixing the quorum disk but, while I am fixing it, the cluster is down...
So, tell me, what is the purpose to buy two servers to have an HA cluster if at first little problem all cluster is down?

Again, the recommendation is a three node cluster, no one tells you a two node cluster is a great idea.

3 nodes (or two and quorum disk) means, your cluster have three votes. So as long as two nodes are online, your services are up. As I did not know your system in detail, I have no idea why your cluster is down because you lost one vote. Find it out.