Proxmox 2.1 synchro storage

[snarf77]

Hi all,

I was using proxmox 1.9 and updated to 2.0 beta for a while in cluster configuration. At that time I had to configure by hand a drbd storage to host virtual machine and I performed VM migration manually in case of physical node problem.

I tried today the 2.1 version with two servers and wanted to test the HA feature.

I must confess I don't understand fully of this works compared to previsou version and most documentation I found is always related to 2.0 version so I never know if this is still up to date or not.

My main question is consequently fairly simple: how do we configure a synchronised file system between two cluster node, is that automatic in 2.1 when you select "managed by HA" on a VM ? or do you still need to manage the storage syncrho manually ?

If included, how do you check to synchro status ?

Today I just created one VM and choose managed by HA, but stopping the cluster node on which I create the VM caused no migration and I was unable to relaod it from other node. I guess, I was probably not synchronised on both nodes but really don't know how to check this ?

At last, when I did it the manual way for 2.0 version, I had dedicated an additionnal network interface for drbd syncrho in order to save the rest of the bandwith for the real business use. Where can we configure this in 2.1 ? (if possible of course ...)

Thanks in advance for your help and congratulations to the dev team for such a wonderful product !!

Keep on going

Snarf

 

[udo]

Hi all,

I was using proxmox 1.9 and updated to 2.0 beta for a while in cluster configuration. At that time I had to configure by hand a drbd storage to host virtual machine and I performed VM migration manually in case of physical node problem.

I tried today the 2.1 version with two servers and wanted to test the HA feature.

Hi,
HA with 2 nodes are not realy recommended because you get easy trouble without quorum.

I must confess I don't understand fully of this works compared to previsou version and most documentation I found is always related to 2.0 version so I never know if this is still up to date or not.

2.x is complety different to 1.x - but between 2.1 and 2.0 are not much differences (more stuff from the roadmap is included).

My main question is consequently fairly simple: how do we configure a synchronised file system between two cluster node, is that automatic in 2.1 when you select "managed by HA" on a VM ? or do you still need to manage the storage syncrho manually ?

Managed by HA mean only, that this VM will be restartet on a other node if the node, where this VM currently is running, fails.
You mean the storage switch "shared", which mean, that all nodes see the same content - this is the same on pve 1.x.

If included, how do you check to synchro status ?

syncro status? you must use an storage which are the same on all nodes - this can be lvm on FC/iSCSI/DRBD or NFS (lvm isn't an filesystem) or an clusterfilesystem.

Today I just created one VM and choose managed by HA, but stopping the cluster node on which I create the VM caused no migration and I was unable to relaod it from other node. I guess, I was probably not synchronised on both nodes but really don't know how to check this ?

No - if your DRBD isn't working your VM will started on the other node with old data - pve don't care about the drbd-status.
The HA fails, because with only one node you lost the quorum and /etc/pve is readonly. No changes are possible on the clusterconfig until you get back the quorum.

At last, when I did it the manual way for 2.0 version, I had dedicated an additionnal network interface for drbd syncrho in order to save the rest of the bandwith for the real business use. Where can we configure this in 2.1 ? (if possible of course ...)

Datacenter -> select Node -> Network

Udo

 

[snarf77]

Hi Udo,

Many thanks for these clear and precise answers I really appreciate.

It appears I was dreaming a little bit and I had understood that the HA feature would include the DRBD management which was the point I wanted to avoid as I'm not an IT expert and found this a little bit touchy to monitor and manage.
Unfortunately it was only a dream...perhpas on a future release ?? So between 2.0 and 2.1 (concerning the redundancy management I mean) the only difference is the 'automatic' restart in case of failover, is that correct ?

Concerning the quorum I totally share your point and I use three machine in production but did the test only with two as this only for discovery purpose...

Thanks you

Snarf