scrub_prep set MDT failloc to 0x198. It prevents the inserting
of osd_obj to OI, and brokes the logic of osd_fid_lookup. The
creating and searching quota slv file could happen during this
failloc. osd_fid_lookup() returned osd object with oo_inode equal
to zero and a bit later, osd_index_try got BUG cause it tried to
access to &LDISKFS_I(inode)->i_es_lock.
The patch adds OST object creation for quota synchronization before
setting failloc.
Test-Parameters: trivial testlist=sanity-scrub
Signed-off-by: Alexander Boyko <c17825@cray.com>
Cray-bug-id: LUS-6051
Change-Id: I8a37c476c2c95b235b0f0ecc7ec1ab0066ae3883
Reviewed-on: https://review.whamcloud.com/33750
Tested-by: Jenkins
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: James Nunez <jnunez@whamcloud.com>
Reviewed-by: Andriy Skulysh <c17819@cray.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
error "Failed to copy files to mds$n"
mkdir -p $DIR/$tdir/mds$n/d_$tfile ||
error "mkdir failed on mds$n"
- createmany -m $DIR/$tdir/mds$n/d_$tfile/f 2 > \
+ touch $DIR/$tdir/mds$n/d_$tfile/f1 > \
/dev/null || error "create failed on mds$n"
+ dd if=/dev/zero of=$DIR/$tdir/mds$n/d_$tfile/f2 bs=1M count=1 ||
+ error "write failed on mds$n"
if [[ $nfiles -gt 0 ]]; then
createmany -m $DIR/$tdir/mds$n/$tfile $nfiles > \
/dev/null || error "createmany failed on mds$n"