[...]
3540 trunc from 0x789289 to 0xd8b1
3542 write 0x8571af thru 0x85e2af (0x7101 bytes)
3543 trunc from 0x85e2b0 to 0x7dc986
3550 read 0x3b37bd thru 0x3b50a8 (0x18ec bytes)
3551 mapread 0x686907 thru 0x693eb9 (0xd5b3 bytes)
READ BAD DATA: offset = 0x686907, size = 0xd5b3, fname = 112.2
OFFSET GOOD BAD RANGE
0x686907 0x0000 0x198f 0x00000
operation# (mod 256) for the bad data may be 143
0x686908 0x0000 0x8fec 0x00001
operation# (mod 256) for the bad data may be 143
In theory that range should be truncated and zeroed in the first
"trunc" from 0x789289 to 0xd8b1 above. From the failure logs, it is
the same issue with before in generic/057 test case. But locally I
s/057/075/
couldn't reproduce it after I fixing the bug in ceph fsize_support
branch.
BTW, could you reproduce this every time ? Or just occasionally ?
[...]
