Sébastien CRAMATTE schrieb: > Hello > > Efectively, I've found an How to > http://linux-vserver.org/Vserver+DRBD > > But it's incomplete... > > Well I think that I'm going to install an active/passive failover > but how can I manage IPs of each vserver ? > What happens if a service crash inside a vserver ? How heartbeat > detect this ?
Hearbeat seems to be a good solution for high availability on a host basis, so if node-a dies make node-b active. That's at least what we use it for. For monitoring of services and automated actions based on their status we use monit (http://www.tildeslash.com/monit/index.php) within the vserver, but there are several other tools like that around. If you use vservers and start/stop vservers on the corresponding hosts once heartbeat detects an outage vserver configuration will take care of the ip takeover if you use the same configuration on all nodes. But of course there are a lot more configurations that could be thought of. you could monitor services on the other node from the root server and - ind case of outage - bind additional ips to your vservers and start the corresponding services. Or you can do that from within if you give your vservers enough privileges to acquire additional ips. With that and shared storage systems being cheaply available you can do nice setups. We have a two node ha cluster which has active vservers on each node, their filesystems are on the shared storage (but theiy don't share data, so we don't need an cluster fs) and in case of an outage the other node takes over the other nodes vservers within a couple of seconds. You need nodes that can handle the complete load on their own, but it's far more efficient than running the nodes in an true active-passive failover setup (which is what you'll have with drdb). Bye, Oliver > > Thanks for your help > > Bye > > > Oliver Welter a écrit : >> Hi Sebastien, >> >>> I search for a config example of heartbeat + drdb + vserver 2.x under >>> kernel 2.6 >> >> There are two HowTos on the Wiki >> http://linux-vserver.org/Documentation >> >> It depends on your setup and there are at least two different >> approaches to manage the system - you can either start the vserver >> from heartbeat and control everything else from the vServer scripts >> (thats what the author of the list post seems to do) or you control >> all ressources from heartbeat... >> >> Oli >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Vserver mailing list >> Vserver@list.linux-vserver.org >> http://list.linux-vserver.org/mailman/listinfo/vserver >> > > _______________________________________________ > Vserver mailing list > Vserver@list.linux-vserver.org > http://list.linux-vserver.org/mailman/listinfo/vserver _______________________________________________ Vserver mailing list Vserver@list.linux-vserver.org http://list.linux-vserver.org/mailman/listinfo/vserver