X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Ftests%2Freplay-dual.sh;h=6db74a28d107b0fc8012db28573850d6c9446e4d;hp=3c84d8dfc960da0fd59289caa461f3153829de60;hb=68fb53ad4bb2dbc2104d19613546ecf2534d97d5;hpb=2dc9c16e770415d56839e1996015fec5fab93f29

diff --git a/lustre/tests/replay-dual.sh b/lustre/tests/replay-dual.sh
index 3c84d8d..b92df6b 100755
--- a/lustre/tests/replay-dual.sh
+++ b/lustre/tests/replay-dual.sh
@@ -1,81 +1,118 @@
-#!/bin/sh
+#!/bin/bash
 
 set -e
 
-LUSTRE=${LUSTRE:-`dirname $0`/..}
-. $LUSTRE/tests/test-framework.sh
+PTLDEBUG=${PTLDEBUG:--1}
+MOUNT_2=${MOUNT_2:-"yes"}
 
+LUSTRE=${LUSTRE:-$(cd $(dirname $0)/..; echo $PWD)}
+. $LUSTRE/tests/test-framework.sh
 init_test_env $@
+init_logging
 
-. ${CONFIG:=$LUSTRE/tests/cfg/local.sh}
+remote_mds_nodsh && skip "remote MDS with nodsh" && exit 0
 
-gen_config() {
-    rm -f $XMLCONFIG
-    add_mds mds --dev $MDSDEV --size $MDSSIZE
-    if [ ! -z "$mdsfailover_HOST" ]; then
-	 add_mdsfailover mds --dev $MDSDEV --size $MDSSIZE
-    fi
-    
-    add_lov lov1 mds --stripe_sz $STRIPE_BYTES \
-	--stripe_cnt $STRIPES_PER_OBJ --stripe_pattern 0
-    add_ost ost --lov lov1 --dev $OSTDEV --size $OSTSIZE
-    add_ost ost2 --lov lov1 --dev ${OSTDEV}-2 --size $OSTSIZE
-    add_client client mds --lov lov1 --path $MOUNT
-}
+ALWAYS_EXCEPT="$REPLAY_DUAL_EXCEPT "
+# bug number for skipped test:  LU-2012 LU-8333 LU-7372
+ALWAYS_EXCEPT+="                14b     21b     26 "
+# UPDATE THE COMMENT ABOVE WITH BUG NUMBERS WHEN CHANGING ALWAYS_EXCEPT!
 
+[[ "$mds1_FSTYPE" == zfs ]] &&
+# bug number for skipped test:        LU-2230
+	ALWAYS_EXCEPT="$ALWAYS_EXCEPT 21b"
 
+#                                   7  (min)"
+[ "$SLOW" = "no" ] && EXCEPT_SLOW="21b "
 
 build_test_filter
+check_and_setup_lustre
 
-cleanup() {
-    # make sure we are using the primary MDS, so the config log will
-    # be able to clean up properly.
-    activemds=`facet_active mds`
-    if [ $activemds != "mds" ]; then
-        fail mds
-    fi
+MOUNTED=$(mounted_lustre_filesystems)
+if ! $(echo $MOUNTED' ' | grep -w -q $MOUNT2' '); then
+	zconf_mount $HOSTNAME $MOUNT2
+	MOUNTED2=yes
+fi
 
-    umount $MOUNT2
-    umount $MOUNT
-    rmmod llite
-    stop mds ${FORCE}
-    stop ost2 ${FORCE}
-    stop ost ${FORCE}  --dump cleanup-dual.log
-}
+assert_DIR
+rm -rf $DIR/[df][0-9]*
 
-if [ "$ONLY" == "cleanup" ]; then
-    sysctl -w portals.debug=0
-    cleanup
-    exit
-fi
+[ "$DAEMONFILE" ] && $LCTL debug_daemon start $DAEMONFILE $DAEMONSIZE
 
-gen_config
-start ost --reformat $OSTLCONFARGS 
-PINGER=`cat /proc/fs/lustre/pinger`
+# if there is no CLIENT1 defined, some tests can be ran on localhost
+CLIENT1=${CLIENT1:-$HOSTNAME}
+# if CLIENT2 doesn't exist then use CLIENT1 instead
+# All tests should use CLIENT2 with MOUNT2 only therefore it will work if
+# $CLIENT2 == CLIENT1
+# Exception is the test which need two separate nodes
+CLIENT2=${CLIENT2:-$CLIENT1}
 
-if [ "$PINGER" != "on" ]; then
-    echo "ERROR: Lustre must be built with --enable-pinger for replay-dual"
-    stop mds
-    exit 1
+# LU-482 Avert LVM and VM inability to flush caches in pre .33 kernels
+if [ $LINUX_VERSION_CODE -lt $(version_code 2.6.33) ]; then
+	sync
+	do_facet $SINGLEMDS "sync; sleep 10; sync; sleep 10; sync"
 fi
 
-start ost2 --reformat $OSTLCONFARGS 
-[ "$DAEMONFILE" ] && $LCTL debug_daemon start $DAEMONFILE $DAEMONSIZE
-start mds $MDSLCONFARGS --reformat
-grep " $MOUNT " /proc/mounts || zconf_mount `hostname` $MOUNT
-grep " $MOUNT2 " /proc/mounts || zconf_mount `hostname` $MOUNT2
+LU482_FAILED=$(mktemp -u $TMP/$TESTSUITE.lu482.XXXXXX)
+test_0a() {
+	echo "Check file is LU482_FAILED=$LU482_FAILED"
+	touch $MOUNT2/$tfile-A # force sync FLD/SEQ update before barrier
+	replay_barrier $SINGLEMDS
+#define OBD_FAIL_PTLRPC_FINISH_REPLAY | OBD_FAIL_ONCE
+	touch $MOUNT2/$tfile
+	createmany -o $MOUNT1/$tfile- 50
+	$LCTL set_param fail_loc=0x80000514
+	facet_failover $SINGLEMDS
+	[ -f "$LU482_FAILED" ] && skip "LU-482 failure" && return 0
+	client_up || return 1
+	umount -f $MOUNT2
+	client_up || return 1
+	zconf_mount `hostname` $MOUNT2 || error "mount2 fais"
+	unlinkmany $MOUNT1/$tfile- 50 || return 2
+	rm $MOUNT2/$tfile || return 3
+	rm $MOUNT2/$tfile-A || return 4
+}
+run_test 0a "expired recovery with lost client"
 
-echo $TIMEOUT > /proc/sys/lustre/timeout
-echo $UPCALL > /proc/sys/lustre/upcall
+if [ -f "$LU482_FAILED" ]; then
+	log "Found check file $LU482_FAILED, aborting test script"
+	rm -vf "$LU482_FAILED"
+	complete $SECONDS
+	do_nodes $CLIENTS umount -f $MOUNT2 || true
+	do_nodes $CLIENTS umount -f $MOUNT || true
+	# copied from stopall, but avoid the MDS recovery
+    for num in `seq $OSTCOUNT`; do
+        stop ost$num -f
+        rm -f $TMP/ost${num}active
+    done
+    if ! combined_mgs_mds ; then
+        stop mgs
+    fi
 
-[ "$DAEMONFILE" ] && $LCTL debug_daemon start $DAEMONFILE $DAEMONSIZE
+	exit_status
+fi
+
+test_0b() {
+    replay_barrier $SINGLEMDS
+    touch $MOUNT2/$tfile
+    touch $MOUNT1/$tfile-2
+    umount $MOUNT2
+    facet_failover $SINGLEMDS
+    umount -f $MOUNT1
+    zconf_mount `hostname` $MOUNT1 || error "mount1 fais"
+    zconf_mount `hostname` $MOUNT2 || error "mount2 fais"
+    # it is uncertain if file-2 exists or not, remove it if it does
+    checkstat $MOUNT1/$tfile-2 && rm $MOUNT1/$tfile-2
+    checkstat $MOUNT2/$tfile && return 2
+    return 0
+}
+run_test 0b "lost client during waiting for next transno"
 
 test_1() {
     touch $MOUNT1/a
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     touch $MOUNT2/b
 
-    fail mds
+    fail $SINGLEMDS
     checkstat $MOUNT2/a || return 1
     checkstat $MOUNT1/b || return 2
     rm $MOUNT2/a $MOUNT1/b
@@ -88,24 +125,23 @@ run_test 1 "|X| simple create"
 
 
 test_2() {
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     mkdir $MOUNT1/adir
 
-    fail mds
+    fail $SINGLEMDS
     checkstat $MOUNT2/adir || return 1
     rmdir $MOUNT2/adir
     checkstat $MOUNT2/adir && return 2
     return 0
 }
-
 run_test 2 "|X| mkdir adir"
 
 test_3() {
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     mkdir $MOUNT1/adir
     mkdir $MOUNT2/adir/bdir
 
-    fail mds
+    fail $SINGLEMDS
     checkstat $MOUNT2/adir      || return 1
     checkstat $MOUNT1/adir/bdir || return 2
     rmdir $MOUNT2/adir/bdir $MOUNT1/adir
@@ -113,16 +149,15 @@ test_3() {
     checkstat $MOUNT2/adir/bdir && return 4
     return 0
 }
-
 run_test 3 "|X| mkdir adir, mkdir adir/bdir "
 
 test_4() {
     mkdir $MOUNT1/adir
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     mkdir $MOUNT1/adir  && return 1
     mkdir $MOUNT2/adir/bdir
 
-    fail mds
+    fail $SINGLEMDS
     checkstat $MOUNT2/adir      || return 2
     checkstat $MOUNT1/adir/bdir || return 3
 
@@ -131,23 +166,20 @@ test_4() {
     checkstat $MOUNT2/adir/bdir && return 5
     return 0
 }
-
 run_test 4 "|X| mkdir adir (-EEXIST), mkdir adir/bdir "
 
 
 test_5() {
     # multiclient version of replay_single.sh/test_8
     mcreate $MOUNT1/a
-    multiop $MOUNT2/a o_tSc &
+    multiop_bg_pause $MOUNT2/a o_tSc || return  1
     pid=$!
-    # give multiop a chance to open
-    sleep 1 
     rm -f $MOUNT1/a
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     kill -USR1 $pid
     wait $pid || return 1
 
-    fail mds
+    fail $SINGLEMDS
     [ -e $MOUNT2/a ] && return 2
     return 0
 }
@@ -156,26 +188,901 @@ run_test 5 "open, unlink |X| close"
 
 test_6() {
     mcreate $MOUNT1/a
-    multiop $MOUNT2/a o_c &
+    multiop_bg_pause $MOUNT2/a o_c || return 1
     pid1=$!
-    multiop $MOUNT1/a o_c &
+    multiop_bg_pause $MOUNT1/a o_c || return 1
     pid2=$!
-    # give multiop a chance to open
-    sleep 1 
     rm -f $MOUNT1/a
-    replay_barrier mds
+    replay_barrier $SINGLEMDS
     kill -USR1 $pid1
     wait $pid1 || return 1
 
-    fail mds
+    fail $SINGLEMDS
     kill -USR1 $pid2
     wait $pid2 || return 1
     [ -e $MOUNT2/a ] && return 2
     return 0
 }
-run_test 6 "open1, open2, unlink |X| close1 [fail mds] close2"
+run_test 6 "open1, open2, unlink |X| close1 [fail $SINGLEMDS] close2"
 
-if [ "$ONLY" != "setup" ]; then
-	equals_msg test complete, cleaning up
-	cleanup
-fi
+test_8() {
+    replay_barrier $SINGLEMDS
+    drop_reint_reply "mcreate $MOUNT1/$tfile"    || return 1
+    fail $SINGLEMDS
+    checkstat $MOUNT2/$tfile || return 2
+    rm $MOUNT1/$tfile || return 3
+
+    return 0
+}
+run_test 8 "replay of resent request"
+
+test_9() {
+    replay_barrier $SINGLEMDS
+    mcreate $MOUNT1/$tfile-1
+    mcreate $MOUNT2/$tfile-2
+    # drop first reint reply
+    do_facet $SINGLEMDS lctl set_param fail_loc=0x80000119
+    fail $SINGLEMDS
+    do_facet $SINGLEMDS lctl set_param fail_loc=0
+
+    rm $MOUNT1/$tfile-[1,2] || return 1
+
+    return 0
+}
+run_test 9 "resending a replayed create"
+
+test_10() {
+    mcreate $MOUNT1/$tfile-1
+    replay_barrier $SINGLEMDS
+    munlink $MOUNT1/$tfile-1
+    mcreate $MOUNT2/$tfile-2
+    # drop first reint reply
+    do_facet $SINGLEMDS lctl set_param fail_loc=0x80000119
+    fail $SINGLEMDS
+    do_facet $SINGLEMDS lctl set_param fail_loc=0
+
+    checkstat $MOUNT1/$tfile-1 && return 1
+    checkstat $MOUNT1/$tfile-2 || return 2
+    rm $MOUNT1/$tfile-2
+
+    return 0
+}
+run_test 10 "resending a replayed unlink"
+
+test_11() {
+	replay_barrier $SINGLEMDS
+	mcreate $DIR1/$tfile-1
+	mcreate $DIR2/$tfile-2
+	mcreate $DIR1/$tfile-3
+	mcreate $DIR2/$tfile-4
+	mcreate $DIR1/$tfile-5
+	# drop all reint replies for a while
+	do_facet $SINGLEMDS $LCTL set_param fail_loc=0x0119
+	# note that with this fail_loc set, facet_failover df will fail
+	facet_failover $SINGLEMDS
+
+	local clients=${CLIENTS:-$HOSTNAME}
+	wait_clients_import_state "$clients" $SINGLEMDS FULL
+
+	do_facet $SINGLEMDS $LCTL set_param fail_loc=0
+
+	rm $DIR1/$tfile-[1-5] || return 1
+
+	return 0
+}
+run_test 11 "both clients timeout during replay"
+
+test_12() {
+    replay_barrier $SINGLEMDS
+
+    multiop_bg_pause $DIR/$tfile mo_c || return 1
+    MULTIPID=$!
+
+#define OBD_FAIL_LDLM_ENQUEUE_NET			0x302
+    do_facet $SINGLEMDS lctl set_param fail_loc=0x80000302
+    facet_failover $SINGLEMDS
+    do_facet $SINGLEMDS lctl set_param fail_loc=0
+    clients_up || return 1
+
+    ls $DIR/$tfile
+    kill -USR1 $MULTIPID || return 3
+    wait $MULTIPID || return 4
+    $CHECKSTAT -t file $DIR/$tfile || return 2
+    rm $DIR/$tfile
+
+    return 0
+}
+run_test 12 "open resend timeout"
+
+test_13() {
+    multiop_bg_pause $DIR/$tfile mo_c || return 1
+    MULTIPID=$!
+
+    replay_barrier $SINGLEMDS
+
+    kill -USR1 $MULTIPID || return 3
+    wait $MULTIPID || return 4
+
+    # drop close
+    do_facet $SINGLEMDS lctl set_param fail_loc=0x80000115
+    facet_failover $SINGLEMDS
+    do_facet $SINGLEMDS lctl set_param fail_loc=0
+    clients_up || return 1
+
+    ls $DIR/$tfile
+    $CHECKSTAT -t file $DIR/$tfile || return 2
+    rm $DIR/$tfile
+
+    return 0
+}
+run_test 13 "close resend timeout"
+
+# test 14a removed after 18143 because it shouldn't fail anymore and do the same
+# as test_15a
+
+test_14b() {
+	wait_mds_ost_sync
+	wait_delete_completed
+
+	local beforeused=$(df -P $DIR | tail -1 | awk '{ print $3 }')
+
+	mkdir -p $MOUNT1/$tdir
+	$LFS setstripe -i 0 $MOUNT1/$tdir
+	replay_barrier $SINGLEMDS
+	createmany -o $MOUNT1/$tdir/$tfile- 5
+
+	$LFS setstripe -i 0 $MOUNT2/$tfile-2
+	dd if=/dev/zero of=$MOUNT2/$tfile-2 bs=1M count=5
+	createmany -o $MOUNT1/$tdir/$tfile-3- 5
+	umount $MOUNT2
+
+	fail $SINGLEMDS
+	wait_recovery_complete $SINGLEMDS || error "MDS recovery not done"
+
+	# first set of files should have been replayed
+	unlinkmany $MOUNT1/$tdir/$tfile- 5 || error "first unlinks failed"
+	unlinkmany $MOUNT1/$tdir/$tfile-3- 5 || error "second unlinks failed"
+
+	zconf_mount $HOSTNAME $MOUNT2 || error "mount $MOUNT2 failed"
+	[ -f $MOUNT2/$tfile-2 ] && error "$MOUNT2/$tfile-2 exists!"
+
+	wait_mds_ost_sync || error "wait_mds_ost_sync failed"
+	wait_delete_completed || error "wait_delete_complete failed"
+
+	local afterused=$(df -P $DIR | tail -1 | awk '{ print $3 }')
+	log "before $beforeused, after $afterused"
+	# leave some margin for some files/dirs to be modified (OI, llog, etc)
+	[ $afterused -le $((beforeused + $(fs_log_size))) ] ||
+		error "after $afterused > before $beforeused"
+}
+run_test 14b "delete ost orphans if gap occured in objids due to VBR"
+
+test_15a() { # was test_15
+    replay_barrier $SINGLEMDS
+    createmany -o $MOUNT1/$tfile- 25
+    createmany -o $MOUNT2/$tfile-2- 1
+    umount $MOUNT2
+
+    fail $SINGLEMDS
+
+    unlinkmany $MOUNT1/$tfile- 25 || return 2
+    [ -e $MOUNT1/$tfile-2-0 ] && error "$tfile-2-0 exists"
+
+    zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
+    return 0
+}
+run_test 15a "timeout waiting for lost client during replay, 1 client completes"
+
+test_15c() {
+    replay_barrier $SINGLEMDS
+    for ((i = 0; i < 2000; i++)); do
+        echo "data" > "$MOUNT2/${tfile}-$i" || error "create ${tfile}-$i failed"
+    done
+    umount $MOUNT2
+
+    fail $SINGLEMDS
+
+    zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
+    return 0
+}
+run_test 15c "remove multiple OST orphans"
+
+test_16() {
+    replay_barrier $SINGLEMDS
+    createmany -o $MOUNT1/$tfile- 25
+    createmany -o $MOUNT2/$tfile-2- 1
+    umount $MOUNT2
+
+    facet_failover $SINGLEMDS
+    sleep $TIMEOUT
+    fail $SINGLEMDS
+
+    unlinkmany $MOUNT1/$tfile- 25 || return 2
+
+    zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
+    return 0
+
+}
+run_test 16 "fail MDS during recovery (3571)"
+
+test_17() {
+    remote_ost_nodsh && skip "remote OST with nodsh" && return 0
+
+    createmany -o $MOUNT1/$tfile- 25
+    createmany -o $MOUNT2/$tfile-2- 1
+
+    # Make sure the disconnect is lost
+    replay_barrier ost1
+    umount $MOUNT2
+
+    facet_failover ost1
+    sleep $TIMEOUT
+    fail ost1
+
+    unlinkmany $MOUNT1/$tfile- 25 || return 2
+
+    zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
+    return 0
+
+}
+run_test 17 "fail OST during recovery (3571)"
+
+# cleanup with blocked enqueue fails until timer elapses (MDS busy), wait for it
+export NOW=0
+
+test_18() { # bug 3822 - evicting client with enqueued lock
+	#set -vx
+	local DLMTRACE=$(do_facet $SINGLEMDS lctl get_param debug)
+	do_facet $SINGLEMDS lctl set_param debug=+dlmtrace
+	mkdir -p $MOUNT1/$tdir || error "mkdir $MOUNT1/$tdir failed"
+	touch $MOUNT1/$tdir/${tfile}0 || error "touch file failed"
+	statmany -s $MOUNT1/$tdir/$tfile 1 500 &
+	OPENPID=$!
+	NOW=$SECONDS
+	#define OBD_FAIL_LDLM_ENQUEUE_BLOCKED    0x30b
+	do_facet $SINGLEMDS lctl set_param fail_loc=0x8000030b  # hold enqueue
+	sleep 1
+	#define OBD_FAIL_LDLM_BL_CALLBACK_NET			0x305
+	do_facet client lctl set_param ldlm.namespaces.*.early_lock_cancel=0
+	do_facet client lctl set_param fail_loc=0x80000305  # drop cb, evict
+	cancel_lru_locks mdc
+	sleep 0.1 # wait to ensure first client is one that will be evicted
+	openfile -f O_RDONLY $MOUNT2/$tdir/$tfile
+	wait $OPENPID
+	do_facet client lctl set_param ldlm.namespaces.*.early_lock_cancel=1
+	do_facet $SINGLEMDS lctl debug_kernel |
+		grep "not entering recovery" && error "client not evicted"
+	do_facet client "lctl set_param fail_loc=0"
+	do_facet $SINGLEMDS "lctl set_param fail_loc=0"
+}
+run_test 18 "ldlm_handle_enqueue succeeds on evicted export (3822)"
+
+test_19() { # Bug 10991 - resend of open request does not fail assertion.
+    replay_barrier $SINGLEMDS
+    drop_mdt_ldlm_reply "createmany -o $DIR/$tfile 1" || return 1
+    fail $SINGLEMDS
+    checkstat $DIR2/${tfile}0 || return 2
+    rm $DIR/${tfile}0 || return 3
+
+    return 0
+}
+run_test 19 "resend of open request"
+
+test_20() { #16389
+	local before=$SECONDS
+	replay_barrier $SINGLEMDS
+	touch $DIR1/$tfile.a
+	touch $DIR2/$tfile.b
+	umount $DIR2
+	fail $SINGLEMDS
+	rm $DIR1/$tfile.a
+	zconf_mount $HOSTNAME $DIR2 || error "mount $DIR2 fail"
+	local tier1=$((SECONDS - before))
+
+	before=$SECONDS
+	replay_barrier $SINGLEMDS
+	touch $DIR1/$tfile.a
+	touch $DIR2/$tfile.b
+	umount $DIR2
+	fail $SINGLEMDS
+	rm $DIR1/$tfile.a
+	zconf_mount $HOSTNAME $DIR2 || error "mount $DIR2 fail"
+	local tier2=$((SECONDS - before))
+
+	# timeout is more than 1.5x original timeout
+	((tier2 < tier1 * 6 / 4)) ||
+		error "recovery time $tier2 >= 1.5x original time $tier1"
+}
+run_test 20 "recovery time is not increasing"
+
+# commit on sharing tests
+test_21a() {
+    local param_file=$TMP/$tfile-params
+
+	save_lustre_params $SINGLEMDS "mdt.*.commit_on_sharing" > $param_file
+    do_facet $SINGLEMDS lctl set_param mdt.*.commit_on_sharing=1
+    touch  $MOUNT1/$tfile-1
+    mv  $MOUNT2/$tfile-1 $MOUNT2/$tfile-2
+    mv  $MOUNT1/$tfile-2 $MOUNT1/$tfile-3
+    replay_barrier_nosync $SINGLEMDS
+    umount $MOUNT2
+
+    facet_failover $SINGLEMDS
+
+    # all renames are replayed
+    unlink  $MOUNT1/$tfile-3 || return 2
+
+    zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
+
+    do_facet $SINGLEMDS lctl set_param mdt.*.commit_on_sharing=0
+    rm -rf $MOUNT1/$tfile-*
+    restore_lustre_params < $param_file
+    rm -f $param_file
+    return 0
+}
+run_test 21a "commit on sharing"
+
+test_21b_sub () {
+	local mds=$1
+	do_node $CLIENT1 rm -f $MOUNT1/$tfile-*
+
+	do_facet $mds sync
+	do_node $CLIENT1 touch $MOUNT1/$tfile-1
+	do_node $CLIENT2 mv $MOUNT1/$tfile-1 $MOUNT1/$tfile-2
+	do_node $CLIENT1 mv $MOUNT1/$tfile-2 $MOUNT1/$tfile-3
+
+	replay_barrier_nosync $mds
+	shutdown_client $CLIENT2 $MOUNT1
+
+	facet_failover $mds
+
+	# were renames replayed?
+	local rc=0
+	echo UNLINK $MOUNT1/$tfile-3
+	do_node $CLIENT1 unlink  $MOUNT1/$tfile-3 ||
+		{ echo "unlink $tfile-3 fail!" && rc=1; }
+
+	boot_node $CLIENT2
+	zconf_mount_clients $CLIENT2 $MOUNT1 ||
+		error "mount $CLIENT2 $MOUNT1 fail"
+
+	return $rc
+}
+
+test_21b() {
+	[ -z "$CLIENTS" ] && skip "Need two or more clients" && return
+	[ $CLIENTCOUNT -lt 2 ] &&
+		{ skip "Need 2+ clients, have $CLIENTCOUNT" && return; }
+
+	if [ "$FAILURE_MODE" = "HARD" ] && mixed_mdt_devs; then
+		skip "Several MDTs on one MDS with FAILURE_MODE=$FAILURE_MODE"
+		return 0
+	fi
+
+	zconf_umount_clients $CLIENTS $MOUNT2
+	zconf_mount_clients $CLIENTS $MOUNT1
+
+	local param_file=$TMP/$tfile-params
+
+	local mdtidx=$($LFS getstripe -m $MOUNT1)
+	local facet=mds$((mdtidx + 1))
+
+	save_lustre_params $facet "mdt.*.commit_on_sharing" > $param_file
+
+	# COS enabled
+	local COS=1
+	do_facet $facet lctl set_param mdt.*.commit_on_sharing=$COS
+
+	test_21b_sub $facet || error "Not all renames are replayed. COS=$COS"
+
+	# there is still a window when transactions may be written to disk
+	# before the mds device is set R/O. To avoid such a rare test failure,
+	# the check is repeated several times.
+	COS=0
+	local n_attempts=1
+	while true; do
+		# COS disabled (should fail)
+		do_facet $facet lctl set_param mdt.*.commit_on_sharing=$COS
+
+		test_21b_sub $facet || break
+		n_attempts=$((n_attempts + 1))
+		[ $n_attempts -gt 3 ] &&
+			error "can't check if COS works: rename replied w/o COS"
+	done
+	zconf_mount_clients $CLIENTS $MOUNT2
+	restore_lustre_params < $param_file
+	rm -f $param_file
+	return 0
+}
+run_test 21b "commit on sharing, two clients"
+
+checkstat_22() {
+	checkstat $MOUNT1/$remote_dir || return 1
+	checkstat $MOUNT1/$remote_dir/dir || return 2
+	checkstat $MOUNT1/$remote_dir/$tfile-1 || return 3
+	checkstat $MOUNT1/$remote_dir/dir/$tfile-1 || return 4
+	return 0
+}
+
+create_remote_dir_files_22() {
+	do_node $CLIENT2 mkdir ${MOUNT2}/$remote_dir/dir || return 1
+	do_node $CLIENT1 createmany -o $MOUNT1/$remote_dir/dir/$tfile- 2 ||
+							    return 2
+	do_node $CLIENT2 createmany -o $MOUNT2/$remote_dir/$tfile- 2 ||
+							    return 3
+	return 0
+}
+
+test_22a () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	([ $FAILURE_MODE == "HARD" ] &&
+		[ "$(facet_host mds1)" == "$(facet_host mds2)" ]) &&
+		skip "MDTs needs to be on diff hosts for HARD fail mode" &&
+		return 0
+
+	local MDTIDX=1
+	local remote_dir=${tdir}/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+
+	# OBD_FAIL_MDS_REINT_NET_REP       0x119
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0x119
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+
+	fail mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "lfs mkdir failed"
+
+	replay_barrier mds$MDTIDX
+	create_remote_dir_files_22 || error "Remote creation failed $?"
+	fail mds$MDTIDX
+
+	checkstat_22 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 22a "c1 lfs mkdir -i 1 dir1, M1 drop reply & fail, c2 mkdir dir1/dir"
+
+test_22b () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	local MDTIDX=1
+	local remote_dir=$tdir/remote_dir
+
+	# OBD_FAIL_MDS_REINT_NET_REP       0x119
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0x119
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+
+	fail mds${MDTIDX},mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "lfs mkdir failed"
+
+	replay_barrier mds$MDTIDX
+	create_remote_dir_files_22 || error "Remote creation failed $?"
+	fail mds${MDTIDX}
+
+	checkstat_22 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 22b "c1 lfs mkdir -i 1 d1, M1 drop reply & fail M0/M1, c2 mkdir d1/dir"
+
+test_22c () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	([ $FAILURE_MODE == "HARD" ] &&
+		[ "$(facet_host mds1)" == "$(facet_host mds2)" ]) &&
+		skip "MDTs needs to be on diff hosts for HARD fail mode" &&
+		return 0
+	local MDTIDX=1
+	local remote_dir=${tdir}/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+
+	# OBD_FAIL_OUT_UPDATE_NET_REP    0x1701
+	do_facet mds$MDTIDX lctl set_param fail_loc=0x1701
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+	do_facet mds$MDTIDX lctl set_param fail_loc=0
+
+	fail mds$MDTIDX
+	wait $CLIENT_PID || error "lfs mkdir failed"
+
+	replay_barrier mds$MDTIDX
+	create_remote_dir_files_22 || error "Remote creation failed $?"
+	fail mds$MDTIDX
+
+	checkstat_22 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 22c "c1 lfs mkdir -i 1 d1, M1 drop update & fail M1, c2 mkdir d1/dir"
+
+test_22d () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	local MDTIDX=1
+	local remote_dir=${tdir}/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+
+	# OBD_FAIL_OUT_UPDATE_NET_REP    0x1701
+	do_facet mds$MDTIDX lctl set_param fail_loc=0x1701
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+	do_facet mds$MDTIDX lctl set_param fail_loc=0
+
+	fail mds${MDTIDX},mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "lfs mkdir failed"
+
+	replay_barrier mds$MDTIDX
+	create_remote_dir_files_22 || error "Remote creation failed $?"
+	fail mds$MDTIDX
+
+	checkstat_22 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 22d "c1 lfs mkdir -i 1 d1, M1 drop update & fail M0/M1,c2 mkdir d1/dir"
+
+checkstat_23() {
+	checkstat $MOUNT1/$remote_dir || return 1
+	checkstat $MOUNT1/$remote_dir/$tfile-1 || return 2
+	return 0
+}
+
+create_remote_dir_files_23() {
+	do_node $CLIENT2 mkdir ${MOUNT2}/$remote_dir || return 1
+	do_node $CLIENT2 createmany -o $MOUNT2/$remote_dir/$tfile- 2 || return 2
+	return 0
+}
+
+test_23a () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	([ $FAILURE_MODE == "HARD" ] &&
+		[ "$(facet_host mds1)" == "$(facet_host mds2)" ]) &&
+		skip "MDTs needs to be on diff hosts for HARD fail mode" &&
+		return 0
+	local MDTIDX=1
+	local remote_dir=$tdir/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir ||
+			error "lfs mkdir failed"
+	# OBD_FAIL_MDS_REINT_NET_REP       0x119
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0x119
+	do_node $CLIENT1 rmdir $MOUNT1/$remote_dir &
+	local CLIENT_PID=$!
+	sleep 1
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0
+
+	fail mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "rmdir remote dir failed"
+
+	replay_barrier mds${MDTIDX}
+	create_remote_dir_files_23 || error "Remote creation failed $?"
+	fail mds${MDTIDX}
+
+	checkstat_23 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 23a "c1 rmdir d1, M1 drop reply and fail, client2 mkdir d1"
+
+test_23b () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	local MDTIDX=1
+	local remote_dir=$tdir/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir ||
+			error "lfs mkdir failed"
+
+	# OBD_FAIL_MDS_REINT_NET_REP       0x119
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0x119
+	do_node $CLIENT1 rmdir $MOUNT1/$remote_dir &
+	local CLIENT_PID=$!
+	sleep 1
+	do_facet mds$((MDTIDX + 1)) lctl set_param fail_loc=0
+
+	fail mds${MDTIDX},mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "rmdir remote dir failed"
+
+	replay_barrier mds${MDTIDX}
+	create_remote_dir_files_23 || error "Remote creation failed $?"
+	fail mds${MDTIDX}
+
+	checkstat_23 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || error "rmdir remote_dir failed"
+	return 0
+}
+run_test 23b "c1 rmdir d1, M1 drop reply and fail M0/M1, c2 mkdir d1"
+
+test_23c () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+
+	([ $FAILURE_MODE == "HARD" ] &&
+		[ "$(facet_host mds1)" == "$(facet_host mds2)" ]) &&
+		skip "MDTs needs to be on diff hosts for HARD fail mode" &&
+		return 0
+	local MDTIDX=1
+	local remote_dir=$tdir/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir ||
+			error "lfs mkdir failed"
+
+	# OBD_FAIL_OUT_UPDATE_NET_REP    0x1701
+	do_facet mds${MDTIDX} lctl set_param fail_loc=0x1701
+	do_node $CLIENT1 rmdir $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+	do_facet mds${MDTIDX} lctl set_param fail_loc=0
+
+	fail mds${MDTIDX}
+	wait $CLIENT_PID || error "rmdir remote dir failed"
+
+	replay_barrier mds${MDTIDX}
+	create_remote_dir_files_23 || error "Remote creation failed $?"
+	fail mds${MDTIDX}
+
+	checkstat_23 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || return 6
+	return 0
+}
+run_test 23c "c1 rmdir d1, M0 drop update reply and fail M0, c2 mkdir d1"
+
+test_23d () {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	local MDTIDX=1
+	local remote_dir=$tdir/remote_dir
+
+	do_node $CLIENT1 mkdir -p $MOUNT1/${tdir}
+	do_node $CLIENT1 $LFS mkdir -i $MDTIDX $MOUNT1/$remote_dir ||
+			error "lfs mkdir failed"
+
+	# OBD_FAIL_UPDATE_OBJ_NET    0x1701
+	do_facet mds${MDTIDX} lctl set_param fail_loc=0x1701
+	do_node $CLIENT1 rmdir $MOUNT1/$remote_dir &
+	CLIENT_PID=$!
+	sleep 1
+	do_facet mds${MDTIDX} lctl set_param fail_loc=0
+
+	fail mds${MDTIDX},mds$((MDTIDX + 1))
+	wait $CLIENT_PID || error "rmdir remote dir failed"
+
+	replay_barrier mds${MDTIDX}
+	create_remote_dir_files_23 || error "Remote creation failed $?"
+	fail mds${MDTIDX}
+
+	checkstat_23 || error "check stat failed $?"
+
+	rm -rf $MOUNT1/$tdir || return 6
+	return 0
+}
+run_test 23d "c1 rmdir d1, M0 drop update reply and fail M0/M1, c2 mkdir d1"
+
+test_24 () {
+	[[ "$MDS1_VERSION" -gt $(version_code 2.5.2) ]] ||
+		skip "Need MDS version newer than 2.5.2"
+
+	touch $MOUNT/$tfile
+	stat $MOUNT/$tfile >&/dev/null
+# OBD_FAIL_MDS_REINT_NET_REP
+	do_facet $SINGLEMDS $LCTL set_param fail_loc=0x119
+	$TRUNCATE $MOUNT/$tfile 100 &
+	PID=$!
+	sleep 1
+	do_facet $SINGLEMDS lctl set_param fail_loc=0
+	# sync to release rep-ack lock quickly
+	do_nodes $(comma_list $(mdts_nodes)) \
+	    "lctl set_param -n osd*.*MDT*.force_sync 1"
+	rm $MOUNT2/$tfile
+	wait
+}
+run_test 24 "reconstruct on non-existing object"
+
+# end commit on sharing tests 
+
+test_25() {
+	cancel_lru_locks osc
+
+	$LFS setstripe -i 0 -c 1 $DIR/$tfile
+
+	# get lock for the 1st client
+	dd if=/dev/zero of=$DIR/$tfile count=1 >/dev/null ||
+		error "failed to write data"
+
+	# get waiting locks for the 2nd client
+	drop_ldlm_cancel "multiop $DIR2/$tfile Ow512" &
+	sleep 1
+
+	# failover, replay and resend replayed waiting locks
+	if [ "$OST1_VERSION" -ge $(version_code 2.6.90) ]; then
+		#define OBD_FAIL_LDLM_SRV_CP_AST      0x325
+		do_facet ost1 lctl set_param fail_loc=0x80000325
+	else
+		#define OBD_FAIL_OST_LDLM_REPLY_NET	0x213
+		do_facet ost1 lctl set_param fail_loc=0x80000213
+	fi
+
+	fail ost1
+
+	# multiop does not finish because CP AST is skipped;
+	# it is ok to kill it in the test, because CP AST is already re-sent
+	# and it does not hung forever in real life
+	killall multiop
+	wait
+}
+run_test 25 "replay|resend"
+
+cleanup_26() {
+	trap 0
+	kill -9 $tar_26_pid
+	kill -9 $dbench_26_pid
+	killall -9 dbench
+}
+
+test_26() {
+	local clients=${CLIENTS:-$HOSTNAME}
+
+	zconf_mount_clients $clients $MOUNT
+
+	local duration=600
+	[ "$SLOW" = "no" ] && duration=200
+	# set duration to 900 because it takes some time to boot node
+	[ "$FAILURE_MODE" = HARD ] && duration=900
+
+	local start_ts=$SECONDS
+	local rc=0
+
+	trap cleanup_26	EXIT
+	(
+		local tar_dir=$DIR/$tdir/run_tar
+		while true; do
+			test_mkdir -p -c$MDSCOUNT $tar_dir || break
+			if [ $MDSCOUNT -ge 2 ]; then
+				$LFS setdirstripe -D -c$MDSCOUNT $tar_dir ||
+					error "set default dirstripe failed"
+			fi
+			cd $tar_dir || break
+			tar cf - /etc | tar xf - || error "tar failed"
+			cd $DIR/$tdir || break
+			rm -rf $tar_dir || break
+		done
+	)&
+	tar_26_pid=$!
+	echo "Started tar $tar_26_pid"
+
+	(
+		local dbench_dir=$DIR2/$tdir/run_dbench
+		while true; do
+			test_mkdir -p -c$MDSCOUNT $dbench_dir || break
+			if [ $MDSCOUNT -ge 2 ]; then
+				$LFS setdirstripe -D -c$MDSCOUNT $dbench_dir ||
+					error "set default dirstripe failed"
+			fi
+			cd $dbench_dir || break
+			rundbench 1 -D $dbench_dir -t 100 &>/dev/null || break
+			cd $DIR/$tdir || break
+			rm -rf $dbench_dir || break
+		done
+	)&
+	dbench_26_pid=$!
+	echo "Started dbench $dbench_26_pid"
+
+	local num_failovers=0
+	local fail_index=1
+	while [ $((SECONDS - start_ts)) -lt $duration ]; do
+		kill -0 $tar_26_pid || error "tar $tar_26_pid missing"
+		kill -0 $dbench_26_pid || error "dbench $dbench_26_pid missing"
+		sleep 2
+		replay_barrier mds$fail_index
+		sleep 2 # give clients a time to do operations
+		# Increment the number of failovers
+		num_failovers=$((num_failovers + 1))
+		log "$TESTNAME fail mds$fail_index $num_failovers times"
+		fail mds$fail_index
+		if [ $fail_index -ge $MDSCOUNT ]; then
+			fail_index=1
+		else
+			fail_index=$((fail_index + 1))
+		fi
+	done
+	# stop the client loads
+	kill -0 $tar_26_pid || error "tar $tar_26_pid stopped"
+	kill -0 $dbench_26_pid || error "dbench $dbench_26_pid stopped"
+	cleanup_26 || true
+}
+run_test 26 "dbench and tar with mds failover"
+
+test_28() {
+	$LFS setstripe -i 0 -c 1 $DIR2/$tfile
+	dd if=/dev/zero of=$DIR2/$tfile bs=4096 count=1
+
+	#define OBD_FAIL_LDLM_SRV_BL_AST	 0x324
+	do_facet ost1 $LCTL set_param fail_loc=0x80000324
+
+	dd if=/dev/zero of=$DIR/$tfile bs=4096 count=1 &
+	local pid=$!
+	sleep 2
+
+	#define OBD_FAIL_LDLM_GRANT_CHECK        0x32a
+	do_facet ost1 $LCTL set_param fail_loc=0x32a
+
+	fail ost1
+
+	sleep 2
+	cancel_lru_locks OST0000-osc
+	wait $pid || error "dd failed"
+}
+run_test 28 "lock replay should be ordered: waiting after granted"
+
+test_29() {
+	local dir0=$DIR/$tdir/d0
+	local dir1=$DIR/$tdir/d1
+
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return 0
+	[ $CLIENTCOUNT -lt 2 ] && skip "needs >= 2 clients" && return 0
+	[ "$CLIENT1" == "$CLIENT2" ] &&
+		skip "clients must be on different nodes" && return 0
+
+	mkdir -p $DIR/$tdir
+	$LFS mkdir -i0 $dir0
+	$LFS mkdir -i1 $dir1
+	sync
+
+	replay_barrier mds2
+	# create a remote dir, drop reply
+	#define OBD_FAIL_PTLRPC_ROUND_XID 0x530
+	$LCTL set_param fail_loc=0x530 fail_val=36
+	#define OBD_FAIL_MDS_REINT_MULTI_NET_REP 0x15a
+	do_facet mds2 $LCTL set_param fail_loc=0x8000015a
+	echo make remote dir d0 for $dir0
+	$LFS mkdir -i1 -c1 $dir0/d3 &
+	sleep 1
+
+	echo make local dir d1 for $dir1
+	do_node $CLIENT2 $LCTL set_param fail_loc=0x530 fail_val=36
+	do_node $CLIENT2 mkdir $dir1/d4
+
+	fail mds2
+}
+run_test 29 "replay vs update with the same xid"
+
+test_30() {
+	$LFS setstripe -E 1m -L mdt -E -1 $DIR/$tfile
+	#first write to have no problems with grants
+	dd if=/dev/zero of=$DIR/$tfile bs=4k count=10 ||
+		error "dd on client failed"
+	dd if=/dev/zero of=$DIR/$tfile bs=4k count=10 seek=10 ||
+		error "dd on client failed"
+
+	#define OBD_FAIL_LDLM_REPLAY_PAUSE	 0x32e
+	lctl set_param fail_loc=0x32e fail_val=4
+	dd of=/dev/null if=$DIR2/$tfile &
+	local pid=$!
+	sleep 1
+
+	fail $SINGLEMDS
+
+	wait $pid || error "dd on client failed"
+}
+run_test 30 "layout lock replay is not blocked on IO"
+
+complete $SECONDS
+SLEEP=$((SECONDS - $NOW))
+[ $SLEEP -lt $TIMEOUT ] && sleep $SLEEP
+[ "$MOUNTED2" = yes ] && zconf_umount $HOSTNAME $MOUNT2 || true
+check_and_cleanup_lustre
+exit_status