Assuming I have 3 nodes with Proxmox cluster with HA and linstor storage cluster in HCI mode. The issue I wanted to understand is, if there is an issue with the node/host the Proxmox/corosync fencing might shutdown the node/host to avoid any data inconsistency between the node/host. In this situation what happens to the Linstor storage cluster resources since one of the node/host is shutdown or rebooted etc., Generally in a converged environment the Storage cluster is separate. I hope you get my query correct…
Usually, each resource in your Linstor cluster will be placed on at least 2 nodes, depending on your place-count setting. So, if on the node that is shut down or rebooted any resources were primary, they will be demoted to secondary and the resources on the other node(s) become primary. In simpler terms, your Linstor storage cluster will not be negatively affected by the node being rebooted.
In a production setup, make sure that the controller is made highly available (HA). Because if it’s located on the node that is being shut down, you won’t be able to create new resources until the node is back up again.
I just want to add that “under the hood”, each virtual disk will take advantage of DRBD’s quorum.
In your example scenario (losing one node), your resources (virtual disks) will not lose quorum. They can continue to be used in your cluster from the remaining nodes.
HI,Thanks for the update, however it didnt answer my query…While individually Proxmox has its own way to handle and Linstor has its own way the node failures…
But I was trying to understand is, in the HCI mode how Proxmox and Linstor will handle the node failure together, since both run’s together in all the 3 nodes..Unlike in Converged mode Promox is separate Infra(Hardware) while the Storage is separate Infra…If there is proxmox node failure, there is no way Linstor is involved here. which I mentioned in my first post..