[Bugs] [Bug 1449329] When either killing or restarting a brick with performance.stat-prefetch on , stat sometimes returns a bad st_size value.

Fri May 12 10:34:00 UTC 2017

https://bugzilla.redhat.com/show_bug.cgi?id=1449329

miklos.fokin at appeartv.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
              Flags|needinfo?(miklos.fokin at appe |
                   |artv.com)                   |


--- Comment #7 from miklos.fokin at appeartv.com ---
Hello, sorry for the late response this time as well.
I put the logging in place and tried it out.
I am pasting the client logs together with the reproduction output.

[2017-05-12 08:47:52.845882] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 44400000 bytes from brick:1
[2017-05-12 08:47:52.859213] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 44800000 bytes from brick:1
[2017-05-12 08:47:52.876972] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 45200000 bytes from brick:1
[2017-05-12 08:47:52.893153] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 45600000 bytes from brick:1
[2017-05-12 08:47:52.907866] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 46000000 bytes from brick:1
[2017-05-12 08:47:52.924883] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 46400000 bytes from brick:1
[2017-05-12 08:47:52.940711] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 46800000 bytes from brick:1
[2017-05-12 08:47:52.957370] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 47200000 bytes from brick:1
[2017-05-12 08:47:52.971701] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 47600000 bytes from brick:1
[2017-05-12 08:47:52.993772] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 48000000 bytes from brick:1
[2017-05-12 08:47:53.007742] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 48400000 bytes from brick:1
[2017-05-12 08:47:53.025668] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 48800000 bytes from brick:1
[2017-05-12 08:47:53.038570] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 49200000 bytes from brick:1
[2017-05-12 08:47:53.080572] I [afr-common.c:3439:afr_fsync_cbk]
0-volume1-replicate-0: Unwinding with a size of 49600000 bytes from brick:1

fstat fields:
in_fstat.st_dev: 47
in_fstat.st_ino: 12217325060665640715
in_fstat.st_mode: 33261
in_fstat.st_nlink: 1
in_fstat.st_uid: 0
in_fstat.st_gid: 0
in_fstat.st_rdev: 0
in_fstat.st_size: 44800000
in_fstat.st_blksize: 131072
in_fstat.st_blocks: 23808
in_fstat.st_atime: 1494578873
in_fstat.st_mtime: 1494578873
in_fstat.st_ctime: 1494578873
2017-05-12 10:47:53.083996: fstat returned a size of 44800000, bytes_written:
49600000

>From this I would be guessing that this problem is elsewhere.
1-2 days ago I did check my own logs that I added with the diff I put up here a
while ago, and it seemed like this issue is caused when an invalidate call
arrives at md-cache, but I haven't yet had time to look at it in more depth, as
I'm trying to test other things in the meantime (it might have its roots in afr
as well, it is just that I only had logs from md-cache for invalidate).
I'm pasting a part of my own logs from another reproduction just in case it
might help figuring out what goes wrong.


2017-05-12 12:23:58.122391: mdc_invalidate
2017-05-12 12:23:58.122450: iatt_set_validate
2017-05-12 12:23:58.122464: mdc fields:
md_nlink: 1
md_uid: 0
md_gid: 0
md_rdev: 0
md_size: 28800000
md_blocks: 56250
md_atime: 1494584636
md_mtime: 1494584638
md_ctime 1494584638

2017-05-12 12:23:58.122473: prebuf:
2017-05-12 12:23:58.122480: mdc_from_iatt
2017-05-12 12:23:58.122491: iatt fields:
ia_dev: 2050
ia_ino: 11843813978708742420
ia_nlink: 1
ia_uid: 0
ia_gid: 0
ia_rdev: 0
ia_size: 0
ia_blksize: 4096
ia_blocks: 0
ia_atime: 1494584638
ia_mtime: 1494584638
ia_ctime 1494584638

2017-05-12 12:23:58.122510: mdc_fstat
2017-05-12 12:23:58.122522: after_local_check
2017-05-12 12:23:58.122544: mdc_to_iatt
2017-05-12 12:23:58.122556: mdc fields:
md_nlink: 1
md_uid: 0
md_gid: 0
md_rdev: 0
md_size: 0
md_blocks: 0
md_atime: 1494584638
md_mtime: 1494584638
md_ctime 1494584638


fstat fields:
in_fstat.st_dev: 47
in_fstat.st_ino: 11843813978708742420
in_fstat.st_mode: 33261
in_fstat.st_nlink: 1
in_fstat.st_uid: 0
in_fstat.st_gid: 0
in_fstat.st_rdev: 0
in_fstat.st_size: 0
in_fstat.st_blksize: 131072
in_fstat.st_blocks: 0
in_fstat.st_atime: 1494584638
in_fstat.st_mtime: 1494584638
in_fstat.st_ctime: 1494584638
2017-05-12 12:23:58.122649: fstat returned a size of 0, bytes_written: 28800000

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=tio8f2v8XH&a=cc_unsubscribe