Is it possible for Resource(Application) online on all Nodes in failover SG

[Veritas_Veritas]

CS want his One Application Resource part of Failover SG to be online on both nodes.I asked him to move the resource out of VCS and put in RC scripts on both nodes to make it online,but he want to put it in VCS(failover SG).


Any Help/Discussion would be appreciated.

[Volker_Hermin1]

A non-persistent resource in a failover service group can only be online on one node at a time. So you are right.
There is a - VERY - dirty way to do it inside VCS, that would be based on writing monitor scripts that "know" if the resource is supposed to be online on a system or not, and report accordingly. But I'm not going to help anybody implement that ;)

[Veritas_Veritas]

what you say abt separate Parallel SG for This Application Resource?

[Veritas_Veritas]

Then I have To Do Lot of Work keep in mind the Dependencies of this Resource and SG dependencies.

[Volker_Hermin1]

Parallel SGs are ONLY a good idea if you do NOT use any shared file systems in that service group, i.e. file systems that migrate to the active node. If you do, better forget it.
If you have a service group that uses some file systems, but still want an application to be online on all nodes, then you MIGHT play with a separate, parallel service group that contains only the application resource. But service group dependencies are not trivial to use, and in many cases they do not work the way you would like them to work.
It's probably better to use init scripts or an entry in /etc/inittab if you want to keep the service running on all nodes. In Solaris 10, you can define it as a service and have the OS keep the service running relatively easily. That's certainly better than mangling the VCS configuration.

[Veritas_Veritas]

Volker,

yeah its Difficult and might be Impossible to mangle with VCS configuration in this case.I will try to convice CS the other way.

Anyways, can we discuss more abt parallel SGs,like you mention Shared file systems, sorry,but i didn't get you exactly.

[Volker_Hermin1]

Well, if you integrate a VxVM disk group with volumes, mount points etc. into a service group that is parallel, then there is no protection against importing the disk group on multiple nodes at the same time, mounting the file systems on multiple nodes at the same time, and so on. Corruption will occur.
If all data for the application is stored locally, then you can make it a parallel service group. Otherwise, you shoudn't.

[Veritas_Veritas]

Yeah,thats Correct.Then, wat is the use of Parallel SGs in VCS. where it play it role.

Shared DGs concept should Make use in Parallel SG but thats only above vxvm 4.0.

how the Parallel Sgs survives below Vxvm4.0 then?

[Volker_Hermin1]

Parallel SGs can be used for service groups without data or with replicated data, like e.g. ftp servers or web servers with static data or similar stuff.


P.S. Don't forget you can award points if your questions have been answered.

[Hywel_Mallett]

Volker is right about parallel and failover service groups (and he deserves some points ;) )
If the application resource can be online on multiple nodes at once (without causing data corruption), then you could use a parallel service group. I would favour this over taking it out of VCS control.
But trying to get it to run in parallel, but still in failover VCS service groups is asking fro trouble in my opinion!