HAMMER filesystem check / CRC verification

Tomohiro Kusumi kusumi.tomohiro at gmail.com
Sat Feb 20 09:23:15 PST 2016


(resending as I forgot to send to users@ again..)

You might be seeing what I've mentioned in the previous reply.
If that's the case it's probably not a bug, and you see this in older
versions too.

> Note that there seems to be a bug in hammer where a valid filesystem
would still show ^B in hammer show output. I usually run some test scripts
(that contain there 3 commands) before I commit something, but I see ^B in
hammer show every once in a while. It could be that the way hammer show
tests the filesystem is missing something and results in showing errors
that aren't really errors. This occurred at least back in DragonFly 4.0 era
in 2014, and it still does on 4.4. It has been on my todo list for a long
time.

If you have redirected that hammer show output to a file, you can
copy-paste that part (where it showed ^B), or put the whole output to
somewhere else.
(If you haven't you may still be able to reproduce it as long as that
particular btree nodes have not been changed (e.g. splitted) by whatever
filesystem operations you have on your system)

btw, as you may know, hammer show output prints the name of directory
entries (i.e. file name), so be careful if it's not appropriate to send
that.


2016-02-21 1:06 GMT+09:00 PeerCorps Trust Fund <ipc at peercorpstrust.org>:

> So after a recent run of hammer show, I get the following:
>
> HAMMER zone statistics
>         zone #  blocks       items              used[B]             used[%]
>         zone 0  0            0                  0                   0
>         zone 1  0            0                  0                   0
>         zone 2  0            0                  0                   0
>         zone 3  0            0                  0                   0
>         zone 4  0            0                  0                   0
>         zone 5  0            0                  0                   0
>         zone 6  0            0                  0                   0
>         zone 7  0            0                  0                   0
>         zone 8  201          383556             1571045376          93.1757
>         zone 9  2            50127              4809376             28.6661
>         zone 10 122556       16105131           1027866722304       99.9798
>         zone 11 1            24                 119312              1.42231
>         zone 12 0            0                  0                   0
>         zone 13 0            0                  0                   0
>         zone 14 0            0                  0                   0
>         zone 15 0            0                  0                   0
>
> ----------------------------------------------------------------------
>         total   122760       16538838           1029442696368       99.9667
> 54 bad nodes
> 1895 bad elms
>
> Does this output mean that I should be restoring the above disk from
> backup or something else is going on here?
>
> Mike
>
>
> On 01/22/2016 05:57 PM, Tomohiro Kusumi wrote:
> > If you are to redirect the hammer show output to a file, it's better to
> > write to a file on different filesystem (e.g. ufs or another hammer).
> >
> > It's possible that hammer show never ends because what hammer show does
> is
> > run through all ondisk btree nodes. Writing a file is adding nodes into
> the
> > btree of that filesystem.
> >
> >
> > 2016-01-23 0:44 GMT+09:00 Tomohiro Kusumi <kusumi.tomohiro at gmail.com>:
> >
> >> (resending as i forgot to send to users at ...)
> >>
> >> # hammer -v -f ${DEV} blockmap
> >> # hammer -v -f ${DEV} checkmap
> >> # hammer -v -f ${DEV} show
> >>
> >> These three do some validation from different point of view.
> >> If the last two lines of each output says 0 error, then no error was
> found
> >> at least by what these 3 commands test for.
> >>
> >> Note that there seems to be a bug in hammer where a valid filesystem
> would
> >> still show ^B in hammer show output. I usually run some test scripts
> (that
> >> contain there 3 commands) before I commit something, but I see ^B in
> hammer
> >> show every once in a while. It could be that the way hammer show tests
> the
> >> filesystem is missing something and results in showing errors that
> aren't
> >> really errors. This occurred at least back in DragonFly 4.0 era in 2014,
> >> and it still does on 4.4. It has been on my todo list for a long time.
> >>
> >>
> >> 2016-01-22 23:59 GMT+09:00 PeerCorps Trust Fund <ipc at peercorpstrust.org
> >:
> >>
> >>> Hi All,
> >>>
> >>> In ZFS, the command "zpool status" gives a succinct overview of the
> >>> status of mounted file systems. In particular, whether there might be
> >>> checksum errors on files or other data structures.
> >>>
> >>>         NAME        STATE     READ WRITE CKSUM
> >>>         tank        ONLINE       0     0     0
> >>>           ada0p2    ONLINE       0     0     0
> >>>
> >>> In HAMMER the "show" command provides an extremely detailed output of
> all
> >>> data structures and CRC verifications.
> >>>
> >>> I've used the command:
> >>>
> >>> hammer -f /dev/serno/disk-serial-number show |grep B------
> >>>
> >>> On a mounted disk and the output was clean (nothing) after running for
> >>> some time with noted disk activity. Does this mean that all CRCs are
> valid?
> >>> Are there other commands in addition to "show" that allow for
> validation
> >>> and examination of data structures on HAMMER volumes?
> >>>
> >>> --
> >>> Mike
> >>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dragonflybsd.org/pipermail/users/attachments/20160221/2cdd41e8/attachment-0003.htm>


More information about the Users mailing list