[PATCH 06/17] xfs: run xfs_repair at the end of each test

[Eryu Guan]

On Thu, Jul 21, 2016 at 04:46:54PM -0700, Darrick J. Wong wrote:
> Run xfs_repair twice at the end of each test -- once to rebuild
> the btree indices, and again with -n to check the rebuild work.

Seems like it's two more xfs_repair, three in total :)

> 
> Signed-off-by: Darrick J. Wong <darrick.wong at oracle.com>
> ---
>  README    |    4 ++++
>  common/rc |   30 ++++++++++++++++++++++++++++++
>  2 files changed, 34 insertions(+)
> 
> 
> diff --git a/README b/README
> index 2647e12..4509cc1 100644
> --- a/README
> +++ b/README
> @@ -80,6 +80,10 @@ Preparing system for tests (IRIX and Linux):
>                 added to the end of fsstresss and fsx invocations, respectively,
>                 in case you wish to exclude certain operational modes from these
>                 tests.
> +             - set TEST_XFS_REPAIR_REBUILD=1 to have _check_xfs_filesystem
> +               run xfs_repair -n to check the filesystem; xfs_repair to rebuild
> +               metadata indexes; and xfs_repair -n (a third time) to check the
> +               results of the rebuilding.
>  
>          - or add a case to the switch in common/config assigning
>            these variables based on the hostname of your test
> diff --git a/common/rc b/common/rc
> index 7c79bf8..3b45578 100644
> --- a/common/rc
> +++ b/common/rc
> @@ -2428,6 +2428,36 @@ _check_xfs_filesystem()
>          ok=0
>      fi
>  
> +    if [ -n "$TEST_XFS_REPAIR_REBUILD" ]; then
> +        $XFS_REPAIR_PROG -n $extra_options $extra_log_options $extra_rt_options $device >$tmp.repair 2>&1
> +        if [ $? -ne 0 ]
> +        then
> +            echo "_check_xfs_filesystem: filesystem on $device is inconsistent (r) (see $seqres.full)"
> +
> +            echo "_check_xfs_filesystem: filesystem on $device is inconsistent" >>$seqres.full
> +            echo "*** xfs_repair -n output ***"	>>$seqres.full
> +            cat $tmp.repair | _fix_malloc		>>$seqres.full
> +            echo "*** end xfs_repair output"	>>$seqres.full
> +
> +            ok=0
> +        fi
> +        rm -f $tmp.fs_check $tmp.logprint $tmp.repair
> +
> +        $XFS_REPAIR_PROG $extra_options $extra_log_options $extra_rt_options $device >$tmp.repair 2>&1
> +        if [ $? -ne 0 ]
> +        then
> +            echo "_check_xfs_filesystem: filesystem on $device is inconsistent (r) (see $seqres.full)"
> +
> +            echo "_check_xfs_filesystem: filesystem on $device is inconsistent" >>$seqres.full
> +            echo "*** xfs_repair -n output ***"	>>$seqres.full
> +            cat $tmp.repair | _fix_malloc		>>$seqres.full
> +            echo "*** end xfs_repair output"	>>$seqres.full
> +
> +            ok=0
> +        fi
> +        rm -f $tmp.fs_check $tmp.logprint $tmp.repair
> +    fi
> +

I think we can move this hunk after the original xfs_repair and swap the
xfs_repair order, i.e.

xfs_repair -n # this is the original repair check
if <check_rebuild>; then
	xfs_repair # do rebuild without -n
	xfs_repair -n # check rebuild result
fi

This seems more clear to me and we can mark which repair is failing more
easily (right now the three xfs_repairs are all marked as "(r)", it's
hard to say which one is failing), e.g.

xfs_repair -n # marked as "(r)", means "repair"
if <check_rebuild>; then
	xfs_repair # mark as "(rb)", means "rebuild"?
	xfs_repair -n # mark as "(rr)", means "repair rebuild"?
fi

Thanks,
Eryu

>      $XFS_REPAIR_PROG -n $extra_options $extra_log_options $extra_rt_options $device >$tmp.repair 2>&1
>      if [ $? -ne 0 ]
>      then
>