High Availability and Master-Master clustering

[ronpeled]

Hello OpenNode!

First what you are doing here is great! I love the fact that it is based on the CentOS platform. I am currently using proxmox but would love to move to opennode since it is CentOS based. So keep up the good work!

I got a bit confused by your Roadmap, it is not so clear what is actually released vs what is planned. Anyway, I wanted to know:
1. Does your system support master/master clustering?
2. Does your system support High Availability for both OpenVZ and KVM containers?
3. Does your system support fully automated Backups right out of the box?


I already downloaded your ISO and will give it a try, but I still need to know the above items. Thank you!
-Ron

[activesys]

1. Does your system support master/master clustering?

Out-of-box no (not yet as of ON 1.1) - I believe it can be done with some manual setup probably.

2. Does your system support High Availability for both OpenVZ and KVM containers?

Out-of-box no (not yet as of ON 1.1) - there are some options - lbvm and RH Cluster Suite probably.

3. Does your system support fully automated Backups right out of the box?

Out-of-box no (not yet as of ON 1.1) - actually for now we use safekeep.sf.net on our OpenNode installations and we expect ext4 snapshot based realtime COW backup solution to land in the end of the summer. If interested we can provide Safekeep setup howto and rpm packages.

[activesys]

Forgot that for OpenVZ VM backup purposes there is also vzdump utility included in OpenNode.

[Kynao]

Forgot that for OpenVZ VM backup purposes there is also vzdump utility included in OpenNode.

Cool, is saves me of several situations under Proxmox
In Opennode, is it usable through the interface (defining nodes to be regularly backup) ?
If the question is not understandable i mean the same way we use it in Proxmox.
While cli does the job, a little plus vs Proxmox would be the gui also allow the restore of dumps.

[ilja_l]

Hi, Kyano

it's a very valid request. We don't yet have anything in cli for snapshot/backup management. Thanks, will add it to the todo list.

[epretorious]

I got a bit confused by your Roadmap, it is not so clear what is actually released vs what is planned. Anyway, I wanted to know:
1. Does your system support master/master clustering?
2. Does your system support High Availability for both OpenVZ and KVM containers?
3. Does your system support fully automated Backups right out of the box?

Is master/master clustering similar to VMware vSphere's Fault Tolerance (a la VMware vLockstep)?

[activesys]

I would think that under master/master is actually ment active/active clustering - eg load-balanced HA setup - multiple processing nodes and shared storage.

But Im pleased to say that active/passive HA clustering for OpenVZ and KVM is now running on my test setup - so it will be probably ready by OpenNode 6.0 stable release time.  We are using standard RHCS + some custom modifications for OpenVZ clustering. Currently there is a problem with OpenVZ checkpointing and GFS2 shared storage - blocking VE migration if GFS2 shared storage is used. If that gets corrected upstream we can implement also load-balanced active/active clusters.

[epretorious]

I would think that under master/master is actually ment active/active clustering - eg load-balanced HA setup - multiple processing nodes and shared storage.

Thanks for clarifying.

But Im pleased to say that active/passive HA clustering for OpenVZ and KVM is now running on my test setup - so it will be probably ready by OpenNode 6.0 stable release time... we can implement also load-balanced active/active clusters.

Thanks - I look forward to Version 6.0 [STABLE]! 

[fbmondia]

I would think that under master/master is actually ment active/active clustering - eg load-balanced HA setup - multiple processing nodes and shared storage.

But Im pleased to say that active/passive HA clustering for OpenVZ and KVM is now running on my test setup - so it will be probably ready by OpenNode 6.0 stable release time.  We are using standard RHCS + some custom modifications for OpenVZ clustering. Currently there is a problem with OpenVZ checkpointing and GFS2 shared storage - blocking VE migration if GFS2 shared storage is used. If that gets corrected upstream we can implement also load-balanced active/active clusters.

Hi,

OpenNode 6.0 Stable has just been released. I would like to ask if clustering is already implemented with it. I would further like to clarify if the active/active clustering allows moving vm's from one node to another a.k.a VMware livemotion technology.

Francis

[activesys]

GFS2 OpenVZ checkpointing problem is still unsolved as far as I know - so only active/active clustering support atm with GFS2. But with DRBD or NFS it might be doable.

But live migration is working also with non-shared storage or with NFS shared storage as well.