Hi Leszek, Sorry it took me so long to follow-up on this bug.
Leszek Dubiel <[email protected]> writes: >> I made further tests. >> >> This system runs since 2018-02, so maybe it is spoiled from some bugs in >> debian stretch (ver 9). For i386 (i686), maybe! If I remember correctly there are still 32bit bugs. Amd64/intel64 has been good since linux-4.4, and I have two heavily used systems that have been fine since then. But before for that? Ouf... Leszek Dubiel <[email protected]> writes: > Another test with "--repair" make BUG_ON: > > time btrfs check --mode=lowmem --progress --repair /dev/sda1 > enabling repair mode > WARNING: low-memory mode repair support is only partial > Opening filesystem to check... > Checking filesystem on /dev/sda1 > UUID: 666a7089-d716-44ff-8081-56b969b58eff > [1/7] checking root items (0:04:08 elapsed, > 10279825 items checked) > Fixed 0 roots. > ERROR: extent[1198869807104 16384] backref lost (owner: 2, level: 0) > root 2s checked) > ERROR: fail to allocate new chunk No space left on device Do you remember if you ever balanced metadata? > Try to exclude all metadata blcoks and extents, it may be slowd, 81140 > items checked) > Added an extent item [1198869807104 16384] (2:16:04 elapsed, 81140 > items checked) > Added one tree block ref start 1198869807104 root 26:05 elapsed, 81140 > items checked) > ERROR: extent[81379328 16384] backref lost (owner: 2, level: 0) root 2 > items checked) > Added an extent item [81379328 16384] > transaction.c:168: btrfs_commit_transaction: BUG_ON `ret` triggered, > value -17 > btrfs(+0x3c0ad)[0x5320ad] > btrfs(btrfs_commit_transaction+0x68)[0x5324bb] > btrfs(+0x78c24)[0x56ec24] > btrfs(check_chunks_and_extents_lowmem+0x198)[0x572060] > btrfs(cmd_check+0x16e2)[0x554da5] > btrfs(main+0x22c)[0x502edc] > /lib/i386-linux-gnu/libc.so.6(__libc_start_main+0xf1)[0xb7badb41] > btrfs(_start+0x31)[0x502f21] > Przerwane > > real 140m42,553s > user 6m14,394s > sys 2m21,156s I guess the upside is it failed faster this time? (2h20min vs 3days) The lowmem mode was practically brand new back then, and I wonder how useful it is to keep this bug open. Do you remember if you contacted upstream linux-btrfs? Often they'll expand he scope of btrfs-check to accommodate new cases. Cheers! Nicholas
signature.asc
Description: PGP signature
