A Pacemaker resource to detect network partitioning is needed

[Sergey Kolekonov]

Currently there's a problem that some Pacemaker resources can be turned into "Unmanaged state" due to network partitioning.
The possible solution here is to create another resource which will detect such situations. Then all affected resources can be collocated with this resource, so they will be gracefully turned off when the network partitioning situation exists.

[Bogdan Dobrelya (bogdando)]

The description is too vague, please elaborate the cases with unmanaged resources due to partitions

[Bogdan Dobrelya (bogdando)]

"another resource which will detect such situations. Then all affected resources can be collocated with this resource, so they will be gracefully turned off when the network partitioning situation exists."

that is that STONITH resources was designed for. It is a known issue Fuel doesn't provide stonith out of box.

[Bogdan Dobrelya (bogdando)]

But if you are about *other* clusters which may be partitioned and which running as pacemaker resources, then indeed, it may be worth to implement such helper resources to monitor cluster's internal health

[Bogdan Dobrelya (bogdando)]

Closing as this issue remained incomplete for more than a month