[ceph-users] Re: Replace OSD while cluster is recovering?

On 1/3/25 04:12, Gustavo Garcia Rondina wrote:

Thank you for the suggestion. I started `ceph pg repair {pgid}` inconsistent PGs but so far, no effect to be seen. Is it possible to monitor the progress of the repairs? With `ceph progress` I can't see it, and for some reason `ceph -w` is hanging.

We sometimes issue `ceph pg repair` and nothing happens until we `ceph osd set noscrub` and nodeepscrub. Once repair starts, we unset those again. Also we use osd_scrub_auto_repair which helps avoid some of the manual work, while osd_scrub_auto_repair_num_errors means we don't go a long time without noticing disks which are dying. -- Gregory Orange System Administrator, Scientific Platforms Team Pawsey Supercomputing Centre, CSIRO