Bug #64975
closedcrimson: Health check failed: 9 scrub errors (OSD_SCRUB_ERRORS)" in cluster log'
0%
Description
ERROR 2024-03-15 10:04:01,561 [shard 1:main] osd - pg_epoch 198 pg[2.2( empty local-lis/les=11/12 n=0 ec=11/11 lis/c=11/11 les/c/f=12/12/0 sis=11) [0,1] r=0 lpr=11 crt=0'0 mlcod 0'0 active+clean+scrubbing+deep PGScrubber::emit_chunk_result: Scrub errors found. range: start: MIN, end: MAX, result: chunk_result_t(num_scrub_errors: 1, num_deep_scrub_errors: 0, snapset_errors: [[]], object_errors: [[inconsistent_obj_t(error: , object: //snapmapper, version: 0, shards: {osd_shard_t(osd: 0, shard: -1): shard_info_t(error: INFO_MISSING, size: 0, omap_digest_present: true, omap_digest: 4294967295, data_digest_present: true, data_digest: 4294967295, selected_io: false, primary: true), osd_shard_t(osd: 1, shard: -1): shard_info_t(error: INFO_MISSING, size: 0, omap_digest_present: true, omap_digest: 4294967295, data_digest_present: true, data_digest: 4294967295, selected_io: false, primary: false)}, union_shards: INFO_MISSING)]])
The bug is that scrub isn't skipping the snap mapper object.
Updated by Samuel Just about 1 month ago
Testing fix -- gonna stick the snap mapper keys into the pgmeta object and avoid the problem entirely.
Updated by Samuel Just about 1 month ago
- Status changed from New to Fix Under Review
Updated by Samuel Just about 1 month ago
Testing a different fix, the above would have reintroduced a crash on OSD startup while reading the pg log.
Updated by Samuel Just about 1 month ago
- Status changed from Fix Under Review to In Progress
Updated by Matan Breizman 13 days ago
- Status changed from In Progress to Resolved
- Pull request ID set to 56534