automatic pool resizing when cluster grows
Bug #1819898 reported by
James Page
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Ceph Monitor Charm |
Triaged
|
Wishlist
|
Unassigned |
Bug Description
pool pg calculation is performed during deployment based on the expected number of OSD's at the point of deployment.
This is a fixed point in time; capacity may be added to the cloud over time, and pg's need to be re-calculated to avoid any weighting weirdness at some future point when the deployment has grown x fold in size.
An action could be used to re-assess all of the previous processed broker requests - based on the new size of the cluster pg configuration for pools could then be increased if needed.
Decreasing should probably not be supported (only very recent ceph versions support pg merging).
To post a comment you must log in.
We could also flag a major discrepancy between 'expected- osd-count' and the actual osd count in juju status to trigger the operator to run the action.