[ClusterLabs] Knowing where a resource is running

[Ryan Thomas]

I’m attempting to implement a resource that is “master” on only one node,
but the other slave nodes know where the resource is running so they can
forward requests to the “master” node.  It seems like this can be
accomplished by creating a multi-state resource with configured with 1
master with the ‘notify’ action enabled, and then have the slave’s listen
to the pre-promote or post-promote notifications.  However, if node A is
the master, and node B is rebooted… when node B starts back up, I don’t
think it will see a pre/post-promote notification because the promote has
already occurred.  Is there a way for the resource on node B to be informed
that node A is already the master in this case?  I know that I could run
‘pcs resource’ or ‘pcs status’ or etc to see the local status of the
resources and parse the output, but I’d prefer a cleaner and less fragile
solution.  Any suggestions?

Thanks!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20180516/41993688/attachment.html>