Hi, you're probably searching for this config option [0]: # ceph config help osd_scrub_auto_repair osd_scrub_auto_repair - Automatically repair damaged objects detected during scrub (bool, advanced) Default: false Can update at runtime: true Services: [osd] I like to be informed when an inconsistent PG occurs (yes, we do get email notifications) and have the chance to look into potential issues. If it's automatically repaired, the information can get lost (logrotation) or nobody even noticed it. I guess it depends on your personal preference. Regards, Eugen [0] https://docs.ceph.com/en/latest/rados/configuration/osd-config-ref/#confval-... Zitat von lejeczek <peljasz@yahoo.co.uk>:
Hi guys.
I get PG inconsistencies constantly. I have a small, 3-node lab cluster, eg. -> $ ceph health detail HEALTH_ERR 4 scrub errors; Possible data damage: 4 pgs inconsistent [ERR] OSD_SCRUB_ERRORS: 4 scrub errors [ERR] PG_DAMAGED: Possible data damage: 4 pgs inconsistent pg 5.15 is active+clean+inconsistent, acting [5,3,4] pg 5.4f is active+clean+inconsistent, acting [5,4,3] pg 5.64 is active+clean+inconsistent, acting [5,3,4] pg 5.78 is active+clean+inconsistent, acting [3,4,5]
I read but I failed to find a doc/howto on relevant tweaking/setting - could you point me to such a doc, if one exists, where it is covered? Also, are there perhaps different best-practices on small VS large cluster & manual VS auto PG heal/repair? Say - should PG repair be always set/delegated to self/auto VS to never?
many thanks, L. _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io