Just had a chat in #xfs on freenode and it might be that a
previous xfs_growfs under an older kernel didn't zero enough.
When I recall it correctly the XFS filesystem in question was
resized in the past under then stable Debian Wheezy (might even
have been resized under Debian Squeeze).
Chat for reference:
< bfoster> strange, seems like it is running a verifier on the
secondary sb's
< bfoster> e.g., commit 1375cb65e, but that fix seems to go way back
< bfoster> did growfs run and fail prior to the run with these
reported errors?
< thermoman> bfoster: never had problems before
< thermoman> but the FS might have been created with an older
xfsprogs and it has been definately be resized before
< bfoster> or hmm, perhaps its remnants from a previous growfs
< bfoster> ok, I think that makes more sense, you went from 109 ags
to 160 some odd ags
< bfoster> and grow/repair is complaining about ags below that
109 threshold
< bfoster> which suggests the previous grows didn't zero enough,
or whatever that old problem was
< bfoster> I think the kernel you're on now should "do the right thing"
Might be related:
http://xfs.org/index.php/XFS_FAQ#Q:_I.27m_getting_.22Internal_error_xfs_sb_read_verify.22_errors_when_I_try_to_run_xfs_growfs_under_kernels_v3.10_through_v3.12
