High avalibility with no extra hardware.

[ozgurerdogan]

I have 6 servers with same configurations. How can I make high avalibility in case of a node fails. I have no ability to add extra hardware? I read many article but they all made me confuse. Can you guide a updated article about this?

 

[mram]

Your question was unclear.

Create a cluster. Add each node to the cluster. You can add more nodes the cluster at any time. Use ceph or some other common storage for high-availability of your VMs.

 

[Xahid]

You need "Shared Storage" for High Availability to work, use either Ceph or DRBD(not recommended) whatever suits your environment.

 

[Alessandro 123]

You can also use gluster (at least 3.8) with sharding enabled.

 

[ozgurerdogan]

Why is DRBD is not recommended? I see some articles about it. Is Ceph or gluster builtin with proxmox and does not need any extra hardware? My wms are about 20 - 100 GB so how does HA manage it? I have currently 6 server: do I place images for vms on a shared storage (on one proxmox server) and rest of servers responsible for running vms (rest of 5 servers) ? Or do I distrubiute 3 servers for shared storage and rest 3 server for managing and running them. Currently I gave 3-4 vms on each server. And total of about 24 vms.

I am also trying to understand load balance when using shared storage. How is cpu and ram usage done?

Thank you for your responds.

 

[Alessandro 123]

Ceph and gluster are directly supported by proxmox.

There is a guide on the wiki
Keep in mind to upgrade gluster packages, do not use the provided one because you WANT to enable sharding on gluster for VM image hosting and the version distributed with proxmox doesn't support sharding.