Am I correct abaut MC Service Guard switchover idea

[MQ'ski]

Hello.
Is it correct what I think.

1. In normal situation when AUTO_RUN is enabled and switching enabled while any of nodes boots package will start automatically on the active node?

2. If in any case one of active nodes in cluster goes down of any reason (stop, crash etc) AUTO_RUN goes to deisabled until administratively changed to active.
This idea is to prevent switching back if previously active node has some failures.

3. Different meaning is FAILBACK_POLICY that defines:
If original node comes back (assumed repaired, rebooted,recovered) it will switch package back to this node...
Attention!... how it is related to AUTO_RUN because I assume it will be still disabled state?
BR

[Srikanth Arunachalam · Accepted solution]

Hi Marcin,

1. AUTO_RUN set to YES is sufficient for package to start automatically. In the event of failure the package will be started on adoptive node (if switching is enabled this will make sure about node failure).

2. Yes

3. FAILBACK_POLICY can be MANUAL or AUTOMATIC. If MANUAL, no attempt will be made to move the package back to primary node when currently running on adoptive node.

So, if FAILBACK_POLICY=MANUAL and AUTO_RUN=ENABLED, it required your internvetion to start the package on primary node (this gives better option for you to decide and switch the package to primary after repair to it).

if FAILBACK_POLICY=AUTOMATIC and AUTO_RUN=ENABLED. during restoration of package service, it will automatically attempt to start package in primary node, even if the primary node is not yet repaired.

Thanks,
Srikanth

[Steven E. Protter]

Shalom,

1. Yes.

2. Depends on the package configuration scripts.

3. If you have a preferred node you can configure the cluster to run on that node when available. I do this to keep my applications running on a faster server with more cpu's and memory.

SEP

[MQ'ski]

Thank you for answer.