You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cloudstack.apache.org by Ivan Rodriguez <iv...@gmail.com> on 2012/11/05 10:12:19 UTC
cloudstack DataCenter Failover ???
Dear Fellows,
We are thinking in the possibility of having cloudstack in 2 different
datacenters, and have some sort of failover datacenter configuration,
imagining that we have a really low latency link may be less tan 2 ms, and
that we can do synchronous replication of the storage primary and
secondary, may be asynchronous should work to, I was thinking that we
could replicate the mysql cloudstack config into a master/slave and use may
be HA or some other process, to detect if the primary cloudstack is down
then change our slave to master and let the agents connect to the
management console, after that cloudstack should be able to fired up all
the vm's and since the network will be and exact copy should work in theory
right ?,
Is anybody doing something similar ?
Cheers
Ivan
Re: cloudstack DataCenter Failover ???
Posted by Caleb Call <ca...@me.com>.
We are implementing HA/Clustering using Red Hat Clustering (Ricci and Luci) for our management server. In our POC it seems to work great, we'll be moving it to our production Cloudsack probably this week or next. I'm sure this same thing could be used cross datacenters with the proper configuration. One thing is that the same IP range will have to exist in both data centers. Otherwise all your instances will be fired up with new IPs and you wouldn't really be able to do a true HA of the management node/mysql database as the configs would be different.
On Nov 5, 2012, at 2:12 AM, Ivan Rodriguez <iv...@gmail.com> wrote:
> Dear Fellows,
>
>
> We are thinking in the possibility of having cloudstack in 2 different
> datacenters, and have some sort of failover datacenter configuration,
> imagining that we have a really low latency link may be less tan 2 ms, and
> that we can do synchronous replication of the storage primary and
> secondary, may be asynchronous should work to, I was thinking that we
> could replicate the mysql cloudstack config into a master/slave and use may
> be HA or some other process, to detect if the primary cloudstack is down
> then change our slave to master and let the agents connect to the
> management console, after that cloudstack should be able to fired up all
> the vm's and since the network will be and exact copy should work in theory
> right ?,
>
> Is anybody doing something similar ?
>
>
> Cheers
> Ivan
Re: cloudstack DataCenter Failover ???
Posted by Chiradeep Vittal <Ch...@citrix.com>.
Are the hypervisor and storage server addresses identical in the standby
datacenter?
When the switchover happens, the state of the VMs in the database will be
'Running', but the actual state will be 'stopped'.
You might have to manually restart all the vms.
On 11/5/12 1:12 AM, "Ivan Rodriguez" <iv...@gmail.com> wrote:
>Dear Fellows,
>
>
>We are thinking in the possibility of having cloudstack in 2 different
>datacenters, and have some sort of failover datacenter configuration,
>imagining that we have a really low latency link may be less tan 2 ms,
>and
>that we can do synchronous replication of the storage primary and
>secondary, may be asynchronous should work to, I was thinking that we
>could replicate the mysql cloudstack config into a master/slave and use
>may
>be HA or some other process, to detect if the primary cloudstack is down
>then change our slave to master and let the agents connect to the
>management console, after that cloudstack should be able to fired up all
>the vm's and since the network will be and exact copy should work in
>theory
>right ?,
>
>Is anybody doing something similar ?
>
>
>Cheers
>Ivan