# OpenNebula Controller/ Frontend HA

**URL:** <https://forum.opennebula.io/t/opennebula-controller-frontend-ha/327>\
**Category:** Product Support\
**Created:** [February 27, 2015, 2:31pm UTC](https://forum.opennebula.io/t/opennebula-controller-frontend-ha/327 "2015-02-27T14:31:18Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![bechtoldt](https://avatars.discourse-cdn.com/v4/letter/b/f6c823/32.png) [@bechtoldt](https://forum.opennebula.io/u/bechtoldt)\
**Post date:** [February 27, 2015, 2:31pm UTC](https://forum.opennebula.io/t/opennebula-controller-frontend-ha/327/1 "2015-02-27T14:31:18Z")

</div>

Hey guys,

how do you make your OpenNebula frontend (oned + mm\_sched) high available?

There are some bits of information how to setup ONE with pacemaker/ cman/ ccs and other monsters. I’d prefer a way without these tools.

Unfortunately ONE can’t be used in active/ active mode, so we must have one site to be active only.

How are you managing ONE in active/ passive mode?

Arnold

---

<div class="post-metadata">

**Author:** ![DaD](https://yyz1.discourse-cdn.com/flex031/user_avatar/forum.opennebula.io/dad/32/8466_2.png) [@DaD](https://forum.opennebula.io/u/DaD)\
**Post date:** [February 28, 2015, 12:41am UTC](https://forum.opennebula.io/t/opennebula-controller-frontend-ha/327/2 "2015-02-28T00:41:15Z")

</div>

Those monsters are here for good reasons, to make sure “there can be only  
one” to avoid data corruption.

When something gets wrong you need to “Shoot The Other Node In The Head”  
to be certain of it’s state and then take disposition to recover the service.

Maybe heartbeat is a little more easier, I never tried it.

At work, I’m begining to be more “familiar” with corosync(2.3.3) and  
pacemaker(1.1.10) which manage our cLVM/GFS2 over SAN LUNs.

Our frontend is in a pacemaker managed KVM, so it can live migrate from  
node to node.

I will setup a slave with MySQL replication and a virtual IP when I’ll  
have time, so strictly speaking I don’t have an active/passive ONE 😉

So unfortunately, I see no shortcut for an HA setup.

Nodes need to monitor each other and kill the ones which look suspect, better use  
tools dedicated to this as it can became complicated.

Any network outage makes each node a suspect to the others, in which  
case you can make popcorn and looks for this good old spaghetti western  
like film 😉

Regards.