Bug #14247
closedscrubbing and deep scrubbing on OSD startup
0%
Description
This was observed with sage's wip-bluestore branch, but may also be present in master. On creation, OSDs are immediately starting to scrub and deep-scrub PGs in the initially created pools, despite the following ceph.conf values being set:
osd scrub load threshold = 0.01
osd scrub min interval = 137438953472
osd scrub max interval = 137438953472
osd deep scrub interval = 137438953472
osd max scrubs = 16
If I can get through the performance tests I'm running I'll try to grab some logs.
Updated by Samuel Just over 8 years ago
e6dcf1419a45c6a94c1c543269fa70483f163b41 is suspicious. reg_next_scrub now checks pool.info.opts for the pool settings, but doesn't check the return. If they are missing, scrub*interval are 0 in reg_pg_scrub. ScrubJob() seems to compensate by checking for 0 and using the default, but that does seem sketchy.
Updated by Samuel Just over 8 years ago
a828a4df54f040a39f54c4cdecde234376301be5
Big values for min (such as people tend to set when they want to not scrub) overflow the int and go very negative.
Updated by Samuel Just over 8 years ago
- Assignee set to Samuel Just
- Priority changed from Normal to Urgent