5.9-rc1 kernel scrub exits 0, but progs reports it was aborted

5.9-rc1 kernel scrub exits 0, but progs reports it was aborted

[Chris Murphy]

There are multiple problems with scrub with this combination:

5.9.0-0.rc1.1.fc34.x86_64+debug
btrfs-progs-5.7-5.fc33.x86_64

The simplest one:

[101728.012329] BTRFS info (device loop1): scrub: started on devid 1
[101728.832029] BTRFS info (device loop1): scrub: finished on devid 1
with status: 0
[root@fmac libvimages]# btrfs scrub status /media
UUID:             c0cd2b08-c550-4a03-9960-2e4865b97d64
Scrub started:    Wed Aug 19 17:20:52 2020
Status:           aborted
Duration:         0:00:00
Total to scrub:   3.85GiB
Rate:             0.00B/s
Error summary:    no errors found

Consistently it's reported as aborted.


When problems are found in an earlier scrub, also with exit 0
completion, 'btrfs scrub status' reports aborted but also says no
errors found even though they were found and fixed.

[101533.811248] BTRFS info (device loop1): scrub: started on devid 1
[101534.667318] BTRFS info (device loop1): scrub: finished on devid 1
with status: 0
[101567.638306] repair_io_failure: 546 callbacks suppressed
[101567.638315] BTRFS info (device loop1): read error corrected: ino
256 off 0 (dev /dev/loop2 sector 4076824)
[101567.638340] BTRFS info (device loop1): read error corrected: ino
256 off 8192 (dev /dev/loop2 sector 4076840)
[101567.638387] BTRFS info (device loop1): read error corrected: ino
256 off 12288 (dev /dev/loop2 sector 4076848)
[101567.638396] BTRFS info (device loop1): read error corrected: ino
256 off 20480 (dev /dev/loop2 sector 4076864)
[101567.638422] BTRFS info (device loop1): read error corrected: ino
256 off 24576 (dev /dev/loop2 sector 4076872)
[101567.638438] BTRFS info (device loop1): read error corrected: ino
256 off 28672 (dev /dev/loop2 sector 4076880)
[101567.638620] BTRFS info (device loop1): read error corrected: ino
256 off 32768 (dev /dev/loop2 sector 4076888)
[101567.638628] BTRFS info (device loop1): read error corrected: ino
256 off 36864 (dev /dev/loop2 sector 4076896)
[101567.638636] BTRFS info (device loop1): read error corrected: ino
256 off 40960 (dev /dev/loop2 sector 4076904)
[101567.638640] BTRFS info (device loop1): read error corrected: ino
256 off 45056 (dev /dev/loop2 sector 4076912)
[root@fmac libvimages]# btrfs scrub status /media
UUID:             c0cd2b08-c550-4a03-9960-2e4865b97d64
Scrub started:    Wed Aug 19 17:17:37 2020
Status:           aborted
Duration:         0:00:01
Total to scrub:   3.85GiB
Rate:             1.92GiB/s
Error summary:    no errors found


I don't know what the 'repair_io_failure' is all about. This is a 2x
device raid1 on loop. The loop backing files are on btrfs, nodatacow.



-- 
Chris Murphy