All of my Btrfs file systems, including new ones, have errors according to lowmem mode, and no errors reported at all for original mode. So which is correct? If lowmem mode is correct, then there are obviously kernel bugs that are causing problems right away, even on minutes old file systems.
I can't tell from the output whether these are serious errors or minor errors either, because the output from check is totally incomprehensible. Further the usage summary at the end, extent bytes, and fs bytes, etc. are different, sometimes by an order of magnitude, between lowmem and original. Errors like this: ERROR: block group[46200258560 1073741824] used 1073741824 but extent items used 1144422400 ERROR: block group[85928706048 1073741824] used 1073741824 but extent items used 0 ERROR: block group[178270502912 1073741824] used 1073741824 but extent items used 1083211776 ERROR: block group[360806612992 1073741824] used 1073479680 but extent items used 1074769920 ERROR: extent[366498091008, 134217728] referencer count mismatch (root: 818, owner: 73782, offset: 134217728) wanted: 4, have: 26 One file system has over 100 lines (exactly the same thing, no difference) ERROR: data extent[16913485824 7577600] backref lost Another file system, 15 minutes old with two mounts in its whole lifetime, and only written with kernel 4.10-rc3 has over 30 lines of varying numbers: ERROR: root 257 EXTENT DATA[150134 11317248] prealloc shouln't have datasum That file system should have no preallocated extents (It's a clean installation of Fedora Rawhide, using only rsync) Again, zero errors with original mode; but all file systems of all ages have errors with lowmem. This is kindof annoying to say the least, it's like we're adding dice to the check and repair situation on Btrfs which is already incredibly more complicated and unhelpful than any other file system. btrfs-progs 4.9. kernels vary from 4.4 to 4.10-rc4, but one fs is only a month old having used 4.7 and 4.8 kernels; and another one just 4.10-rc3 as I said. -- Chris Murphy -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
