12-09-2009 02:21 PM
12-10-2009 03:10 PM
12-21-2009 04:44 AM
Hi, eshaffer,
remote groups are not very usfull in this case.
But there is a running solution for your problem. This only has two restrictions:
1. SG1 is not manually switchable, of SG2 is running on the same node as SG1 (means on C0), you have to take SG2 offline before you can switch SG1 to C1
2. if you manually switched SG1 to C1 and switch back SG1 to C0 you have to start manually SG2 on C0 as well.
But the automatic failover will take place anyway, and also the automatic startup of SG2 on C0 also takes place when SG1 comes to this node.
And in addition you can have some influence to the sequence of taking SG2 offline when SG1 is failing over to C1: "online local hard" first takes offline SG2 and switches then SG1 to C1, "online local firm" first switches SG1 and then takes offline SG2.
Code is below. Take care for the SystemList attribute and the service group dependency. Please let me know if this is what you want, if not we have to look a little bit closer...
include "types.cf"
cluster test (
UserNames = { admin = ElmElgLimHmmKumGlj }
ClusterAddress = "127.0.0.1"
Administrators = { admin }
)
system C0 (
)
system C1 (
)
group SG1 (
SystemList = { C0 = 0, C1 = 1 }
AutoStartList = { C0, C1 }
)
Phantom SG1_phantom (
)
group SG2 (
SystemList = { C0 = 0 }
AutoStartList = { C0 }
)
Phantom SG2_phantom (
)
requires group SG1 online local hard