On Wed, Aug 06, 2008 at 09:36:02AM +0200, Carlo Granisso wrote: > 1) I need to make an HA system with no main node but "distributed" > control system:
Heartbeat itself will elect a "master" server which serves to coordinate the actions of the cluster, however any server is capable of filling this role and it will be moved elsewhere if the current Designated Controller fails. > goal is to have a pool of N production servers with one (or more) > standby machine. When one of the N servers goes down, the other in > standby must become up. If by this you mean the standby servers will be switched off, and will be powered on when needed, then no I don't believe this is something heartbeat tries to solve. You would need to add some extra plumbing if this is what you want. If you just mean the services will not be active on the standby nodes, then yes. Usually you let heartbeat manage an IP address alias which your clients access, and that address gets moved around between the servers; but the specifics are application-specific. > All machines can have it's indipendent disk and, for the moment, > sync data is not important. It's possibile this solution with > heartbeat? Heartbeat will synchronise its own configuration across all nodes as part of its regular operation. Further synchronisation of data beyond that is left as an exercise to the reader. > 2) Somewhere I've seen that with heartbeat is possible to have HA > with more than two nodes for a service (e.g. httpd or mysql) but I > haven't found tutorials that can help me (only for two nodes). It > seems that from v2 is it possible by enabling in ha.cf "crm" > parameter. Can you help me with some example? Multi-node clusters are supported. You might find reading through this document to be very helpful: http://clusterlabs.org/mw/Image:Configuration_Explained.pdf Essentially you configure "resources" for the services you want to provide, and then you define rules using a fairly flexible syntax to specify which nodes you want them to run on (or where you don't want them to run, as the case may be). You will also find that using the latest packages from here: http://download.opensuse.org/repositories/server:/ha-clustering/ will give you the latest stable builds and thus people on this list will be better able to help you out if you have issues. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
