Actions
Feature #62019
openceph cli: please report the fullest OSD in addition to overall stats
Status:
Fix Under Review
Priority:
Normal
Assignee:
-
Category:
ceph cli
Target version:
-
% Done:
0%
Description
Currently, both `seph -s` and `ceph df` report some stats on overall free space in the cluster. However, both Ceph courses that I attended state that the cluster is, for all practical purposes, as full as its fullest OSD. Yes, there are warnings such as "nearfull," but they are not enough - the kernel CephFS client switches to synchronous writes then, with the corresponding performance loss.
Hence a feature request: please include into `ceph -s` the name of the fullest OSD (as in: osd.123) and the percentage of the space used on this OSD. Also please provide the same information somewhere in `ceph status -f json`.
Updated by Igor Fedotov 6 months ago
- Status changed from New to Fix Under Review
- Pull request ID set to 54426
IMO this information is better to be provided on per-pool basis. Hence implemented via 'ceph df' command.
Updated by Igor Fedotov 6 months ago
- Subject changed from ceph -s: please report the fullest OSD in addition to overall stats to ceph cli: please report the fullest OSD in addition to overall stats
Actions