Monitor MDS load

[jeroenmaelbrancke]

Monitor MDS load: it should evenly (re)balance the MDS master and slaves over all nodes, and where required deploy extra MDS servers.

A starting point: https://github.com/openvstorage/framework/blob/master/ovs/lib/mdsservice.py#L795

[jeroenmaelbrancke]

issue can be closed: https://github.com/openvstorage/openvstorage-monitoring/issues/85

[JeffreyDevloo]

@wimpers What type of check would you need in the healthcheck

[wimpers]

For the moment I would

• Check if the # MDS servers is higher than the threshold 'capacity' and warning if above maxload (in % of capacity)

@khenderick more suggestions?

[khenderick]

@wimpers, that sounds OK, if we need more information we can always add it. But we'll need some kind of "waiting period" to give the mds checkup logic to move volumes around if a lot of vdisks are created at once. We should e.g. only warn users when servers are overloaded for a longer time.