MC Service Guard

[ghousebagwan]

Hi Guys,

I'm trying to do a cluster setup on two nodes, and need assistance with that. Can someone share the step by step plan or any guidance for that? Both the servers are HP-UX 11.31 virtual machines running on Superdome. Thanks.

Ghouse Bagwan.

[Duncan Edmonstone]

Serviceguard is complex - and what you require is often not a straightforward list of commands as you have to consider what you are actually trying to achieve and your own circumstances, so my real advice is to RTFM:

https://support.hpe.com/hpsc/doc/public/display?docLocale=en_US&docId=emr_na-a00062147en_us&withFrame

If you really want to jump straight in to creating the cluster (I don't advise it), then go straight to the "Configuring the cluster" section on p237

[Torsten.]

Another question is if you want to have serviceguard for the application or just for the VM itself.

[Bill Hassell]

Are the two nodes on the same Superdome? If so, ServiceGuard won't help keep things going. 

Or is your goal to just monitor database(s) and apps and change to the other node when things aren't working?

A big part of the setup will be to define failure modes which can be handled with failover. For instance, if the app/db fails due to a full filesystem, the filesystem cannot be part of the shared storage or the same failure will exist on the other node. Similar problems might relate to kernel parameters that are exceeded. Some kernel limits will be new on the other node and the apps will start OK, others may cause the app/db to fail as the limit does not change on the second node.

There are lots of docs at hpe.com for designing an SG environment and some good paper books on the subject. Mr. Google will be useful but only choose docs from hpe.com.

[ghousebagwan]

Hi Torsten,

I want to have it for Application .

[ghousebagwan]

Hi Bill,

Both the nodes are hosted on different superdomes. I want the node2 as failover when we have any failure(any hardware/software creash) state on node1.