In our environments with distributed server rooms there is a good chance
that we loose the connection to the RSA/DRAC/whatever used for stonith
when we loose the connection between the nodes. Usually we use two
independent Networks running over redundant switches for such
environments.
Now we use a dummy resource that only runs where there is network
connection to the outside world and promotes the DRBD when it is running.
This way we make sure, that no changes are possible to the DRBD while
there is no network connection at all. As soon as there is connection to
the outside world, it will be promoted. DRBD and DOPD are configured with
very low timeouts. (100 ms). If the DRBD connection is lost, the secondary
node is outdated, so no failover will happen at all, but the cluster will
go down as a whole to protect the Data. So far it seems quite stable this
way. Maybe this helps you. Any suggestions on that setup are greatly
appreciated.
Robert Köppl
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
