Bug #59668
opencache-flush-evict-all finished with errors - (2) No such file or directory
0%
Description
I can't remove the writeback cache pool.
Problems :
- Documentation seems outdated : procedure to remote the cache mention to use the `ceph osd tier cache-mode cephfs_data_cache proxy` mode, but this option does not exist. So I used the readproxy instead.
- The cache-flush-evict-all reports many errors (check below).
- The cache pool is not empty (46194 objects remains in the cache pool)
root@pve1:~# ceph osd tier cache-mode cephfs_data_cache readproxy root@pve1:~# rados -p cephfs_data_cache cache-flush-evict-all ... failed to flush /1000065c15a.00000000: (2) No such file or directory 1000082419d.00000001 failed to flush /1000082419d.00000001: (2) No such file or directory 100008437ce.00000000 failed to flush /100008437ce.00000000: (2) No such file or directory 1000065b25c.00000000 failed to flush /1000065b25c.00000000: (2) No such file or directory cache-flush-evict-all finished with errors root@pve1:~# rados -p cephfs_data_cache ls | wc -l 46194 root@pve1:~# rados -p cephfs_data_cache cache-flush-evict-all 2>&1 | grep 'failed to flush' | grep -v 'No such file or directory' | wc -l 0
At this stage, I don't know if it's safe to set the cache-mode to none and remove it.
How should I fix the issue?
Thanks
Updated by Pivert Dubuisson 12 months ago
Well, I removed the cache, and it broke the filesystem.
All files opened by applications and recently updated are corrupt.
Also, all snapshots are corrupt (hourly snapshots from scheduler)
What should have been done to prevent this disaster ?