X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Ftests%2Fsanity.sh;h=4d50dcfab8877826132194e9dcfb059c9f72282c;hp=c1cb61642eef1a8776a8bb21595a78129fcdc024;hb=777b04a093;hpb=cc10922919325c212ae98a69d63328c0efbd4f83

diff --git a/lustre/tests/sanity.sh b/lustre/tests/sanity.sh
old mode 100644
new mode 100755
index c1cb616..4d50dcf
--- a/lustre/tests/sanity.sh
+++ b/lustre/tests/sanity.sh
@@ -31,8 +31,8 @@ GRANT_CHECK_LIST=${GRANT_CHECK_LIST:-""}
 export PARALLEL=${PARALLEL:-"no"}
 
 TRACE=${TRACE:-""}
-LUSTRE_TESTS_API_DIR=${LUSTRE_TESTS_API_DIR:-${LUSTRE}/tests/clientapi}
 LUSTRE=${LUSTRE:-$(dirname $0)/..}
+LUSTRE_TESTS_API_DIR=${LUSTRE_TESTS_API_DIR:-${LUSTRE}/tests/clientapi}
 . $LUSTRE/tests/test-framework.sh
 init_test_env $@
 
@@ -41,8 +41,8 @@ init_logging
 ALWAYS_EXCEPT="$SANITY_EXCEPT "
 # bug number for skipped test: LU-9693 LU-6493 LU-9693
 ALWAYS_EXCEPT+="               42a     42b     42c "
-# bug number:    LU-8411 LU-9054
-ALWAYS_EXCEPT+=" 407     312 "
+# bug number:    LU-8411 LU-9054 LU-13314
+ALWAYS_EXCEPT+=" 407     312	56ob"
 
 if $SHARED_KEY; then
 	# bug number:    LU-9795 LU-9795 LU-9795 LU-9795
@@ -64,12 +64,18 @@ if [[ $(uname -m) = aarch64 ]]; then
 fi
 
 # skip nfs tests on kernels >= 4.14.0 until they are fixed
-if [ $LINUX_VERSION_CODE -ge $(version_code 4.14.0) ];then
+if [ $LINUX_VERSION_CODE -ge $(version_code 4.14.0) ]; then
 	# bug number:	LU-12661
 	ALWAYS_EXCEPT+=" 817"
 fi
+# skip cgroup tests on RHEL8.1 kernels until they are fixed
+if (( $LINUX_VERSION_CODE >= $(version_code 4.18.0) &&
+      $LINUX_VERSION_CODE <  $(version_code 5.4.0) )); then
+	# bug number:	LU-13063
+	ALWAYS_EXCEPT+=" 411"
+fi
 
-#                                  5          12          (min)"
+#                                  5          12     8   12  (min)"
 [ "$SLOW" = "no" ] && EXCEPT_SLOW="27m 64b 68 71 115 135 136 300o"
 
 if [ "$mds1_FSTYPE" = "zfs" ]; then
@@ -550,11 +556,6 @@ test_17g() {
 	[ $MDS1_VERSION -le $(version_code 2.3.55) ] &&
 		TESTS="4094 4095"
 
-	# skip long symlink name for rhel6.5.
-	# rhel6.5 has a limit (PATH_MAX - sizeof(struct filename))
-	grep -q '6.5' /etc/redhat-release &>/dev/null &&
-		TESTS="59 60 61 4062 4063"
-
 	for i in $TESTS; do
 		local SYMNAME=$(str_repeat 'x' $i)
 		ln -s $SYMNAME $DIR/$tdir/f$i || error "failed $i-char symlink"
@@ -2641,7 +2642,7 @@ test_27I() {
 run_test 27I "check that root dir striping does not break parent dir one"
 
 test_27J() {
-	[[ $(lustre_version_code $SINGLEMDS) -le $(version_code 2.12.51) ]] &&
+	[[ $MDS1_VERSION -le $(version_code 2.12.51) ]] &&
 		skip "Need MDS version newer than 2.12.51"
 
 	test_mkdir $DIR/$tdir
@@ -2736,7 +2737,7 @@ test_27J() {
 run_test 27J "basic ops on file with foreign LOV"
 
 test_27K() {
-	[[ $(lustre_version_code $SINGLEMDS) -le $(version_code 2.12.49) ]] &&
+	[[ $MDS1_VERSION -le $(version_code 2.12.49) ]] &&
 		skip "Need MDS version newer than 2.12.49"
 
 	test_mkdir $DIR/$tdir
@@ -3748,6 +3749,46 @@ test_33g() {
 }
 run_test 33g "nonroot user create already existing root created file"
 
+test_33h() {
+	[ $MDSCOUNT -lt 2 ] && skip_env "needs >= 2 MDTs"
+	[ $MDS1_VERSION -lt $(version_code 2.13.50) ] &&
+		skip "Need MDS version at least 2.13.50"
+
+	test_mkdir -c $MDSCOUNT -H crush $DIR/$tdir ||
+		error "mkdir $tdir failed"
+	touch $DIR/$tdir/$tfile || error "touch $tfile failed"
+
+	local index=$($LFS getstripe -m $DIR/$tdir/$tfile)
+	local index2
+
+	for fname in $DIR/$tdir/$tfile.bak \
+		     $DIR/$tdir/$tfile.SAV \
+		     $DIR/$tdir/$tfile.orig \
+		     $DIR/$tdir/$tfile~; do
+		touch $fname  || error "touch $fname failed"
+		index2=$($LFS getstripe -m $fname)
+		[ $index -eq $index2 ] ||
+			error "$fname MDT index mismatch $index != $index2"
+	done
+
+	local failed=0
+	for i in {1..50}; do
+		for fname in $(mktemp -u $DIR/$tdir/.$tfile.XXXXXX) \
+			     $(mktemp $DIR/$tdir/$tfile.XXXXXXXX); do
+			touch $fname  || error "touch $fname failed"
+			index2=$($LFS getstripe -m $fname)
+			if [[ $index != $index2 ]]; then
+				failed=$((failed + 1))
+				echo "$fname MDT index mismatch $index != $index2"
+			fi
+		done
+	done
+	echo "$failed MDT index mismatches"
+	(( failed < 4 )) || error "MDT index mismatch $failed times"
+
+}
+run_test 33h "temp file is located on the same MDT as target"
+
 TEST_34_SIZE=${TEST_34_SIZE:-2000000000000}
 test_34a() {
 	rm -f $DIR/f34
@@ -6437,7 +6478,9 @@ test_56wb() {
 	echo "done."
 
 	echo -n "Removing test file from pool '$pool'..."
-	$LFS migrate $file1 &> /dev/null ||
+	# "lfs migrate $file" won't remove the file from the pool
+	# until some striping information is changed.
+	$LFS migrate -c 1 $file1 &> /dev/null ||
 		error "cannot remove from pool"
 	[ "$($LFS getstripe -p $file1)" ] &&
 		error "pool still set"
@@ -6459,19 +6502,23 @@ run_test 56wb "check lfs_migrate pool support"
 
 test_56wc() {
 	local file1="$DIR/$tdir/file1"
+	local parent_ssize
+	local parent_scount
+	local cur_ssize
+	local cur_scount
+	local orig_ssize
 
 	echo -n "Creating test dir..."
 	test_mkdir $DIR/$tdir &> /dev/null || error "cannot create dir"
-	local def_stripe_size=$($LFS getstripe -S $DIR/$tdir 2>/dev/null)
 	$LFS setstripe -S 1M -c 1 "$DIR/$tdir" &> /dev/null ||
-		error "cannot set stripe"
+		error "cannot set stripe by '-S 1M -c 1'"
 	echo "done"
 
 	echo -n "Setting initial stripe for test file..."
 	$LFS setstripe -S 512K -c 1 "$file1" &> /dev/null ||
 		error "cannot set stripe"
-	[ $($LFS getstripe -S "$file1") -eq 524288 ] ||
-		error "stripe size not set"
+	cur_ssize=$($LFS getstripe -S "$file1")
+	[ $cur_ssize -eq 524288 ] || error "setstripe -S $cur_ssize != 524288"
 	echo "done."
 
 	# File currently set to -S 512K -c 1
@@ -6488,8 +6535,8 @@ test_56wc() {
 	echo -n "Verifying -S option is passed through to lfs migrate..."
 	$LFS_MIGRATE -y -S 1M "$file1" &> /dev/null ||
 		error "migration failed"
-	[ $($LFS getstripe -S "$file1") -eq 1048576 ] ||
-		error "file was not restriped"
+	cur_ssize=$($LFS getstripe -S "$file1")
+	[ $cur_ssize -eq 1048576 ] || error "migrate -S $cur_ssize != 1048576"
 	echo "done."
 
 	# File currently set to -S 1M -c 1
@@ -6500,8 +6547,9 @@ test_56wc() {
 		error "long option without argument not supported"
 	$LFS_MIGRATE -y --stripe-size 512K "$file1" &> /dev/null ||
 		error "long option with argument not supported"
-	[ $($LFS getstripe -S "$file1") -eq 524288 ] ||
-		error "file not restriped with --stripe-size option"
+	cur_ssize=$($LFS getstripe -S "$file1")
+	[ $cur_ssize -eq 524288 ] ||
+		error "migrate --stripe-size $cur_ssize != 524288"
 	echo "done."
 
 	# File currently set to -S 512K -c 1
@@ -6510,8 +6558,8 @@ test_56wc() {
 		echo -n "Verifying explicit stripe count can be set..."
 		$LFS_MIGRATE -y -c 2 "$file1" &> /dev/null ||
 			error "migrate failed"
-		[ $($LFS getstripe -c "$file1") -eq 2 ] ||
-			error "file not restriped to explicit count"
+		cur_scount=$($LFS getstripe -c "$file1")
+		[ $cur_scount -eq 2 ] || error "migrate -c $cur_scount != 2"
 		echo "done."
 	fi
 
@@ -6520,17 +6568,21 @@ test_56wc() {
 	# Ensure parent striping is used if -R is set, and no stripe
 	# count or size is specified
 	echo -n "Setting stripe for parent directory..."
-	$LFS setstripe -S 1M -c 1 "$DIR/$tdir" &> /dev/null ||
-		error "cannot set stripe"
+	$LFS setstripe -S 2M -c 1 "$DIR/$tdir" &> /dev/null ||
+		error "cannot set stripe '-S 2M -c 1'"
 	echo "done."
 
 	echo -n "Verifying restripe option uses parent stripe settings..."
+	parent_ssize=$($LFS getstripe -S $DIR/$tdir 2>/dev/null)
+	parent_scount=$($LFS getstripe -c $DIR/$tdir 2>/dev/null)
 	$LFS_MIGRATE -y -R "$file1" &> /dev/null ||
 		error "migrate failed"
-	[ $($LFS getstripe -S "$file1") -eq $def_stripe_size ] ||
-		error "file not restriped to parent settings"
-	[ $($LFS getstripe -c "$file1") -eq 1 ] ||
-		error "file not restriped to parent settings"
+	cur_ssize=$($LFS getstripe -S "$file1")
+	[ $cur_ssize -eq $parent_ssize ] ||
+		error "migrate -R stripe_size $cur_ssize != $parent_ssize"
+	cur_scount=$($LFS getstripe -c "$file1")
+	[ $cur_scount -eq $parent_scount ] ||
+		error "migrate -R stripe_count $cur_scount != $parent_scount"
 	echo "done."
 
 	# File currently set to -S 1M -c 1
@@ -6538,13 +6590,14 @@ test_56wc() {
 	# Ensure striping is preserved if -R is not set, and no stripe
 	# count or size is specified
 	echo -n "Verifying striping size preserved when not specified..."
-	local orig_stripe_size=$($LFS getstripe -S "$file1" 2>/dev/null)
+	orig_ssize=$($LFS getstripe -S "$file1" 2>/dev/null)
 	$LFS setstripe -S 2M -c 1 "$DIR/$tdir" &> /dev/null ||
 		error "cannot set stripe on parent directory"
 	$LFS_MIGRATE -y "$file1" &> /dev/null ||
 		error "migrate failed"
-	[ $($LFS getstripe -S "$file1") -eq $orig_stripe_size ] ||
-		error "file was restriped"
+	cur_ssize=$($LFS getstripe -S "$file1")
+	[ $cur_ssize -eq $orig_ssize ] ||
+		error "migrate by default $cur_ssize != $orig_ssize"
 	echo "done."
 
 	# Ensure file name properly detected when final option has no argument
@@ -6832,6 +6885,75 @@ test_56xc() {
 }
 run_test 56xc "lfs migration autostripe"
 
+test_56xd() {
+	[[ $OSTCOUNT -lt 2 ]] && skip_env "needs >= 2 OSTs"
+
+	local dir=$DIR/$tdir
+	local f_mgrt=$dir/$tfile.mgrt
+	local f_yaml=$dir/$tfile.yaml
+	local f_copy=$dir/$tfile.copy
+	local layout_yaml="-E 1M -S 512K -c 1 -E -1 -S 1M -c 2 -i 0"
+	local layout_copy="-c 2 -S 2M -i 1"
+	local yamlfile=$dir/yamlfile
+	local layout_before;
+	local layout_after;
+
+	test_mkdir "$dir" || error "cannot create dir $dir"
+	$LFS setstripe $layout_yaml $f_yaml ||
+		error "cannot setstripe $f_yaml with layout $layout_yaml"
+	$LFS getstripe --yaml $f_yaml > $yamlfile
+	$LFS setstripe $layout_copy $f_copy ||
+		error "cannot setstripe $f_copy with layout $layout_copy"
+	touch $f_mgrt
+	dd if=/dev/zero of=$f_mgrt bs=1M count=4
+
+	# 1. test option --yaml
+	$LFS_MIGRATE -y --yaml $yamlfile $f_mgrt ||
+		error "cannot migrate $f_mgrt with --yaml $yamlfile"
+	layout_before=$(get_layout_param $f_yaml)
+	layout_after=$(get_layout_param $f_mgrt)
+	[ "$layout_after" == "$layout_before" ] ||
+		error "lfs_migrate --yaml: $layout_after != $layout_before"
+
+	# 2. test option --copy
+	$LFS_MIGRATE -y --copy $f_copy $f_mgrt ||
+		error "cannot migrate $f_mgrt with --copy $f_copy"
+	layout_before=$(get_layout_param $f_copy)
+	layout_after=$(get_layout_param $f_mgrt)
+	[ "$layout_after" == "$layout_before" ] ||
+		error "lfs_migrate --copy: $layout_after != $layout_before"
+}
+run_test 56xd "check lfs_migrate --yaml and --copy support"
+
+test_56xe() {
+	[[ $OSTCOUNT -lt 2 ]] && skip_env "needs >= 2 OSTs"
+
+	local dir=$DIR/$tdir
+	local f_comp=$dir/$tfile
+	local layout="-E 1M -S 512K -c 1 -E -1 -S 1M -c 2 -i 0"
+	local layout_before=""
+	local layout_after=""
+
+	test_mkdir "$dir" || error "cannot create dir $dir"
+	$LFS setstripe $layout $f_comp ||
+		error "cannot setstripe $f_comp with layout $layout"
+	layout_before=$(get_layout_param $f_comp)
+	dd if=/dev/zero of=$f_comp bs=1M count=4
+
+	# 1. migrate a comp layout file by lfs_migrate
+	$LFS_MIGRATE -y $f_comp || error "cannot migrate $f_comp by lfs_migrate"
+	layout_after=$(get_layout_param $f_comp)
+	[ "$layout_before" == "$layout_after" ] ||
+		error "lfs_migrate: $layout_before != $layout_after"
+
+	# 2. migrate a comp layout file by lfs migrate
+	$LFS migrate $f_comp || error "cannot migrate $f_comp by lfs migrate"
+	layout_after=$(get_layout_param $f_comp)
+	[ "$layout_before" == "$layout_after" ] ||
+		error "lfs migrate: $layout_before != $layout_after"
+}
+run_test 56xe "migrate a composite layout file"
+
 test_56y() {
 	[ $MDS1_VERSION -lt $(version_code 2.4.53) ] &&
 		skip "No HSM $(lustre_build_version $SINGLEMDS) MDS < 2.4.53"
@@ -7491,8 +7613,8 @@ run_test 63b "async write errors should be returned to fsync ==="
 test_64a () {
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
 
-	df $DIR
-	lctl get_param -n osc.*[oO][sS][cC][_-]*.cur* | grep "[0-9]"
+	lfs df $DIR
+	lctl get_param osc.*[oO][sS][cC][_-]*.cur* | grep "=[1-9]"
 }
 run_test 64a "verify filter grant calculations (in kernel) ====="
 
@@ -7838,7 +7960,7 @@ run_test 65m "normal user can't set filesystem default stripe"
 
 test_65n() {
 	[ -n "$FILESET" ] && skip "Not functional for FILESET set"
-	[[ $(lustre_version_code $SINGLEMDS) -ge $(version_code 2.12.50) ]] ||
+	[[ $MDS1_VERSION -ge $(version_code 2.12.50) ]] ||
 		skip "Need MDS version at least 2.12.50"
 	[[ $PARALLEL != "yes" ]] || skip "skip parallel run"
 
@@ -8155,28 +8277,27 @@ num_inodes() {
 test_76() { # Now for bug 20433, added originally in bug 1443
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
 
-	local CPUS=$(getconf _NPROCESSORS_ONLN 2>/dev/null)
-
 	cancel_lru_locks osc
-	BEFORE_INODES=$(num_inodes)
-	echo "before inodes: $BEFORE_INODES"
-	local COUNT=1000
-	[ "$SLOW" = "no" ] && COUNT=100
-	for i in $(seq $COUNT); do
+	local cpus=$(getconf _NPROCESSORS_ONLN 2>/dev/null)
+	local before=$(num_inodes)
+	local count=$((512 * cpus))
+	[ "$SLOW" = "no" ] && count=$((64 * cpus))
+
+	echo "before inodes: $before"
+	for i in $(seq $count); do
 		touch $DIR/$tfile
 		rm -f $DIR/$tfile
 	done
 	cancel_lru_locks osc
-	AFTER_INODES=$(num_inodes)
-	echo "after inodes: $AFTER_INODES"
-	local wait=0
-	while [[ $((AFTER_INODES-1*${CPUS:-1})) -gt $BEFORE_INODES ]]; do
-		sleep 2
-		AFTER_INODES=$(num_inodes)
-		wait=$((wait+2))
-		echo "wait $wait seconds inodes: $AFTER_INODES"
-		if [ $wait -gt 30 ]; then
-			error "inode slab grew from $BEFORE_INODES to $AFTER_INODES"
+	local after=$(num_inodes)
+	echo "after inodes: $after"
+	while (( after > before + 8 * ${cpus:-1} )); do
+		sleep 1
+		after=$(num_inodes)
+		wait=$((wait + 1))
+		(( wait % 5 == 0 )) && echo "wait $wait seconds inodes: $after"
+		if (( wait > 30 )); then
+			error "inode slab grew from $before to $after"
 		fi
 	done
 }
@@ -8931,11 +9052,6 @@ test_101c() {
 }
 run_test 101c "check stripe_size aligned read-ahead ================="
 
-set_read_ahead() {
-	$LCTL get_param -n llite.*.max_read_ahead_mb | head -n 1
-	$LCTL set_param -n llite.*.max_read_ahead_mb $1 > /dev/null 2>&1
-}
-
 test_101d() {
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
 
@@ -8955,7 +9071,10 @@ test_101d() {
 	cancel_lru_locks osc
 
 	echo Disable read-ahead
-	local old_READAHEAD=$(set_read_ahead 0)
+	local old_RA=$($LCTL get_param -n llite.*.max_read_ahead_mb | head -n 1)
+	$LCTL set_param -n llite.*.max_read_ahead_mb=0
+	stack_trap "$LCTL set_param -n llite.*.max_read_ahead_mb $old_RA" EXIT
+	$LCTL get_param -n llite.*.max_read_ahead_mb
 
 	echo Reading the test file $file with read-ahead disabled
 	local raOFF=$(do_and_time "dd if=$file of=/dev/null bs=1M count=$sz_MB")
@@ -8963,7 +9082,7 @@ test_101d() {
 	echo Cancel LRU locks on lustre client to flush the client cache
 	cancel_lru_locks osc
 	echo Enable read-ahead with ${ra_MB}MB
-	set_read_ahead $ra_MB
+	$LCTL set_param -n llite.*.max_read_ahead_mb=$ra_MB
 
 	echo Reading the test file $file with read-ahead enabled
 	local raON=$(do_and_time "dd if=$file of=/dev/null bs=1M count=$sz_MB")
@@ -8971,7 +9090,6 @@ test_101d() {
 	echo "read-ahead disabled time read $raOFF"
 	echo "read-ahead enabled  time read $raON"
 
-	set_read_ahead $old_READAHEAD
 	rm -f $file
 	wait_delete_completed
 
@@ -9155,6 +9273,30 @@ test_101h() {
 }
 run_test 101h "Readahead should cover current read window"
 
+test_101i() {
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=10 ||
+		error "dd 10M file failed"
+
+	local max_per_file_mb=$($LCTL get_param -n \
+		llite.*.max_read_ahead_per_file_mb 2>/dev/null)
+	cancel_lru_locks osc
+	stack_trap "$LCTL set_param llite.*.max_read_ahead_per_file_mb=$max_per_file_mb"
+	$LCTL set_param llite.*.max_read_ahead_per_file_mb=1 ||
+		error "set max_read_ahead_per_file_mb to 1 failed"
+
+	echo "Reset readahead stats"
+	$LCTL set_param llite.*.read_ahead_stats=0
+
+	dd if=$DIR/$tfile of=/dev/null bs=2M
+
+	$LCTL get_param llite.*.read_ahead_stats
+	local miss=$($LCTL get_param -n llite.*.read_ahead_stats |
+		     awk '/misses/ { print $2 }')
+	[ $miss -eq 5 ] || error "expected misses 5 but got $miss"
+	rm -f $DIR/$tfile
+}
+run_test 101i "allow current readahead to exceed reservation"
+
 setup_test102() {
 	test_mkdir $DIR/$tdir
 	chown $RUNAS_ID $DIR/$tdir
@@ -11162,6 +11304,27 @@ test_123b () { # statahead(bug 15027)
 }
 run_test 123b "not panic with network error in statahead enqueue (bug 15027)"
 
+test_123c() {
+	[[ $MDSCOUNT -lt 2 ]] && skip_env "needs >= 2 MDTs"
+
+	test_mkdir -i 0 -c 1 $DIR/$tdir.0
+	test_mkdir -i 1 -c 1 $DIR/$tdir.1
+	touch $DIR/$tdir.1/{1..3}
+	mv $DIR/$tdir.1/{1..3} $DIR/$tdir.0
+
+	remount_client $MOUNT
+
+	$MULTIOP $DIR/$tdir.0 Q
+
+	# let statahead to complete
+	ls -l $DIR/$tdir.0 > /dev/null
+
+	testid=$(echo $TESTNAME | tr '_' ' ')
+	dmesg | tac | sed "/$testid/,$ d" | grep "Can not initialize inode" &&
+		error "statahead warning" || true
+}
+run_test 123c "Can not initialize inode warning on DNE statahead"
+
 test_124a() {
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
 	$LCTL get_param -n mdc.*.connect_flags | grep -q lru_resize ||
@@ -12472,33 +12635,22 @@ test_133g() {
 	remote_mds_nodsh && skip "remote MDS with nodsh"
 	remote_ost_nodsh && skip "remote OST with nodsh"
 
-	# eventually, this can also be replaced with "lctl get_param -R",
-	# but not until that option is always available on the server
 	local facet
 	for facet in mds1 ost1; do
-		[ $(lustre_version_code $facet) -le $(version_code 2.5.54) ] &&
-			skip_noexit "Too old lustre on $facet"
-		local facet_proc_dirs=$(do_facet $facet \
-					\\\ls -d $proc_regexp 2>/dev/null)
-		echo "${facet}_proc_dirs='$facet_proc_dirs'"
-		[ -z "$facet_proc_dirs" ] && error "no proc_dirs on $facet"
-		do_facet $facet find $facet_proc_dirs \
-			! -name req_history \
-			-exec cat '{}' \\\; &> /dev/null
-
-		do_facet $facet find $facet_proc_dirs \
-			! -name req_history \
-			-type f \
-			-exec cat '{}' \\\; &> /dev/null ||
-				error "proc file read failed"
-
-		do_facet $facet find $facet_proc_dirs \
-			-ignore_readdir_race \
-			-type f \
-			-not -name force_lbug \
-			-not -name changelog_mask \
-			-exec badarea_io '{}' \\\; ||
-				error_133 "$facet find $facet_proc_dirs failed"
+		local facet_ver=$(lustre_version_code $facet)
+		if [ $facet_ver -ge $(version_code 2.7.65) ]; then
+			do_facet $facet "$LCTL get_param -R '*'" &> /dev/null
+		else
+			log "$facet: too old lustre for get_param -R"
+		fi
+		if [ $facet_ver -ge $(version_code 2.5.54) ]; then
+			do_facet $facet "$LCTL list_param -R '*' | grep '=' |
+				tr -d= | egrep -v 'force_lbug|changelog_mask' |
+				xargs badarea_io" ||
+					error_133 "$facet badarea_io failed"
+		else
+			skip_noexit "$facet: too old lustre for get_param -R"
+		fi
 	done
 
 	# remount the FS in case writes/reads /proc break the FS
@@ -12651,7 +12803,7 @@ test_136() {
 	local fname
 
 	mkdir -p $DIR/$tdir || error "failed to create $DIR/$tdir"
-	$SETSTRIPE -c 1 -i 0 $DIR/$tdir || error "failed to set striping"
+	$LFS setstripe -c 1 -i 0 $DIR/$tdir || error "failed to set striping"
 	#set only one record at plain llog
 #define OBD_FAIL_CATALOG_FULL_CHECK                0x131a
 	do_facet $SINGLEMDS $LCTL set_param fail_loc=0x131a fail_val=1
@@ -13864,15 +14016,20 @@ test_160f() {
 	local i
 
 	# generate some changelog records to accumulate on each MDT
-	test_mkdir -c $MDSCOUNT $DIR/$tdir || error "test_mkdir $tdir failed"
+	# use fnv1a because created files should be evenly distributed
+	test_mkdir -c $MDSCOUNT -H fnv_1a_64 $DIR/$tdir ||
+		error "test_mkdir $tdir failed"
+	log "$(date +%s): creating first files"
 	createmany -m $DIR/$tdir/$tfile $((MDSCOUNT * 2)) ||
 		error "create $DIR/$tdir/$tfile failed"
 
 	# check changelogs have been generated
+	local start=$SECONDS
+	local idle_time=$((MDSCOUNT * 5 + 5))
 	local nbcl=$(changelog_dump | wc -l)
 	[[ $nbcl -eq 0 ]] && error "no changelogs found"
 
-	for param in "changelog_max_idle_time=10" \
+	for param in "changelog_max_idle_time=$idle_time" \
 		     "changelog_gc=1" \
 		     "changelog_min_gc_interval=2" \
 		     "changelog_min_free_cat_entries=3"; do
@@ -13884,8 +14041,11 @@ test_160f() {
 		do_nodes $mdts $LCTL set_param mdd.*.$param
 	done
 
-	# force cl_user2 to be idle (1st part)
-	sleep 9
+	# force cl_user2 to be idle (1st part), but also cancel the
+	# cl_user1 records so that it is not evicted later in the test.
+	local sleep1=$((idle_time / 2))
+	echo "$(date +%s): sleep1 $sleep1/${idle_time}s"
+	sleep $sleep1
 
 	# simulate changelog catalog almost full
 	#define OBD_FAIL_CAT_FREE_RECORDS	0x1313
@@ -13921,13 +14081,16 @@ test_160f() {
 			      "$user_rec1, but is $user_rec2"
 	done
 
-	# force cl_user2 to be idle (2nd part) and to reach
-	# changelog_max_idle_time
-	sleep 2
+	# force cl_user2 idle (2nd part) to just exceed changelog_max_idle_time
+	local sleep2=$((idle_time - (SECONDS - start) + 1))
+	echo "$(date +%s): sleep2 $sleep2/${idle_time}s"
+	sleep $sleep2
 
-	# generate one more changelog to trigger fail_loc
-	createmany -m $DIR/$tdir/${tfile}bis $((MDSCOUNT * 2)) ||
-		error "create $DIR/$tdir/${tfile}bis failed"
+	# Generate one more changelog to trigger GC at fail_loc for cl_user2.
+	# cl_user1 should be OK because it recently processed records.
+	echo "$(date +%s): creating $((MDSCOUNT * 2)) files"
+	createmany -m $DIR/$tdir/${tfile}b $((MDSCOUNT * 2)) ||
+		error "create $DIR/$tdir/${tfile}b failed"
 
 	# ensure gc thread is done
 	for i in $(mdts_nodes); do
@@ -13981,7 +14144,9 @@ test_160g() {
 	local i
 
 	# generate some changelog records to accumulate on each MDT
-	test_mkdir -c $MDSCOUNT $DIR/$tdir || error "mkdir $tdir failed"
+	# use fnv1a because created files should be evenly distributed
+	test_mkdir -c $MDSCOUNT -H fnv_1a_64 $DIR/$tdir ||
+		error "mkdir $tdir failed"
 	createmany -m $DIR/$tdir/$tfile $((MDSCOUNT * 2)) ||
 		error "create $DIR/$tdir/$tfile failed"
 
@@ -14095,7 +14260,9 @@ test_160h() {
 	local i
 
 	# generate some changelog records to accumulate on each MDT
-	test_mkdir -c $MDSCOUNT $DIR/$tdir || error "test_mkdir $tdir failed"
+	# use fnv1a because created files should be evenly distributed
+	test_mkdir -c $MDSCOUNT -H fnv_1a_64 $DIR/$tdir ||
+		error "test_mkdir $tdir failed"
 	createmany -m $DIR/$tdir/$tfile $((MDSCOUNT * 2)) ||
 		error "create $DIR/$tdir/$tfile failed"
 
@@ -14244,7 +14411,9 @@ test_160i() {
 	changelog_register || error "first changelog_register failed"
 
 	# generate some changelog records to accumulate on each MDT
-	test_mkdir -c $MDSCOUNT $DIR/$tdir || error "mkdir $tdir failed"
+	# use fnv1a because created files should be evenly distributed
+	test_mkdir -c $MDSCOUNT -H fnv_1a_64 $DIR/$tdir ||
+		error "mkdir $tdir failed"
 	createmany -m $DIR/$tdir/$tfile $((MDSCOUNT * 2)) ||
 		error "create $DIR/$tdir/$tfile failed"
 
@@ -14303,20 +14472,27 @@ test_160j() {
 		skip "Need MDS version at least 2.12.56"
 
 	mount_client $MOUNT2 || error "mount_client on $MOUNT2 failed"
+	stack_trap "umount $MOUNT2" EXIT
 
 	changelog_register || error "first changelog_register failed"
+	stack_trap "changelog_deregister" EXIT
 
 	# generate some changelog
-	test_mkdir -c $MDSCOUNT $DIR/$tdir || error "mkdir $tdir failed"
+	# use fnv1a because created files should be evenly distributed
+	test_mkdir -c $MDSCOUNT -H fnv_1a_64 $DIR/$tdir ||
+		error "mkdir $tdir failed"
 	createmany -m $DIR/$tdir/${tfile}bis $((MDSCOUNT * 2)) ||
 		error "create $DIR/$tdir/${tfile}bis failed"
 
 	# open the changelog device
 	exec 3>/dev/changelog-$FSNAME-MDT0000
+	stack_trap "exec 3>&-" EXIT
 	exec 4</dev/changelog-$FSNAME-MDT0000
+	stack_trap "exec 4<&-" EXIT
 
 	# umount the first lustre mount
 	umount $MOUNT
+	stack_trap "mount_client $MOUNT" EXIT
 
 	# read changelog
 	cat <&4 >/dev/null || error "read changelog failed"
@@ -14327,16 +14503,6 @@ test_160j() {
 		error "User $cl_user not found in changelog_users"
 
 	printf 'clear:'$cl_user':0' >&3
-
-	# close
-	exec 3>&-
-	exec 4<&-
-
-	# cleanup
-	changelog_deregister || error "changelog_deregister failed"
-
-	umount $MOUNT2
-	mount_client $MOUNT || error "mount_client on $MOUNT failed"
 }
 run_test 160j "client can be umounted  while its chanangelog is being used"
 
@@ -15530,7 +15696,7 @@ jobstats_set() {
 		"$FSNAME.sys.jobid_var" $new_jobenv
 }
 
-test_205() { # Job stats
+test_205a() { # Job stats
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
 	[[ $MDS1_VERSION -ge $(version_code 2.7.1) ]] ||
 		skip "Need MDS version with at least 2.7.1"
@@ -15632,7 +15798,18 @@ test_205() { # Job stats
 
 	verify_jobstats "touch $DIR/$tfile" $SINGLEMDS
 }
-run_test 205 "Verify job stats"
+run_test 205a "Verify job stats"
+
+# LU-13117
+test_205b() {
+	$LCTL set_param jobid_var=USER jobid_name="%e.%u"
+	env -i USERTESTJOBSTATS=foolish touch $DIR/$tfile.1
+	do_facet $SINGLEMDS $LCTL get_param mdt.*.job_stats |
+		grep job_id: | grep foolish &&
+			error "Unexpected jobid found"
+	true
+}
+run_test 205b "Verify job stats jobid parsing"
 
 # LU-1480, LU-1773 and LU-1657
 test_206() {
@@ -16169,8 +16346,7 @@ test_224c() { # LU-6441
 	save_writethrough $p
 	set_cache writethrough on
 
-	local pages_per_rpc=$($LCTL get_param \
-				osc.*.max_pages_per_rpc)
+	local pages_per_rpc=$($LCTL get_param osc.*.max_pages_per_rpc)
 	local at_max=$($LCTL get_param -n at_max)
 	local timeout=$($LCTL get_param -n timeout)
 	local test_at="at_max"
@@ -16579,6 +16755,27 @@ test_230b() {
 	ln -s $migrate_dir/$tfile $migrate_dir/${tfile}_ln
 	ln -s $other_dir/$tfile $migrate_dir/${tfile}_ln_other
 
+	local len
+	local lnktgt
+
+	# inline symlink
+	for len in 58 59 60; do
+		lnktgt=$(str_repeat 'l' $len)
+		touch $migrate_dir/$lnktgt
+		ln -s $lnktgt $migrate_dir/${len}char_ln
+	done
+
+	# PATH_MAX
+	for len in 4094 4095; do
+		lnktgt=$(str_repeat 'l' $len)
+		ln -s $lnktgt $migrate_dir/${len}char_ln
+	done
+
+	# NAME_MAX
+	for len in 254 255; do
+		touch $migrate_dir/$(str_repeat 'l' $len)
+	done
+
 	$LFS migrate -m $MDTIDX $migrate_dir ||
 		error "fails on migrating remote dir to MDT1"
 
@@ -16586,7 +16783,8 @@ test_230b() {
 	for ((i = 0; i < 10; i++)); do
 		for file in $(find $migrate_dir/dir_${i}); do
 			mdt_index=$($LFS getstripe -m $file)
-			[ $mdt_index == $MDTIDX ] ||
+			# broken symlink getstripe will fail
+			[ $mdt_index -ne $MDTIDX ] && stat -L $file &&
 				error "$file is not on MDT${MDTIDX}"
 		done
 	done
@@ -16650,7 +16848,7 @@ test_230b() {
 	echo "migrate back to MDT0, checking.."
 	for file in $(find $migrate_dir); do
 		mdt_index=$($LFS getstripe -m $file)
-		[ $mdt_index == $MDTIDX ] ||
+		[ $mdt_index -ne $MDTIDX ] && stat -L $file &&
 			error "$file is not on MDT${MDTIDX}"
 	done
 
@@ -19363,6 +19561,25 @@ test_278() {
 }
 run_test 278 "Race starting MDS between MDTs stop/start"
 
+test_280() {
+	[ $MGS_VERSION -lt $(version_code 2.13.52) ] &&
+		skip "Need MGS version at least 2.13.52"
+	[ $PARALLEL == "yes" ] && skip "skip parallel run"
+	combined_mgs_mds || skip "needs combined MGS/MDT"
+
+	umount_client $MOUNT
+#define OBD_FAIL_MDS_LLOG_UMOUNT_RACE   0x15e
+	do_facet mgs $LCTL set_param fail_loc=0x8000015e fail_val=0
+
+	mount_client $MOUNT &
+	sleep 1
+	stop mgs || error "stop mgs failed"
+	#for a race mgs would crash
+	start mgs $(mgsdevname) $MGS_MOUNT_OPTS || error "start mgs failed"
+	mount_client $MOUNT || error "mount client failed"
+}
+run_test 280 "Race between MGS umount and client llog processing"
+
 cleanup_test_300() {
 	trap 0
 	umask $SAVE_UMASK
@@ -19516,12 +19733,22 @@ test_300d() {
 	#local striped directory
 	$LFS setdirstripe -i 0 -c 2 -H all_char $DIR/$tdir/striped_dir ||
 		error "set striped dir error"
+        #look at the directories for debug purposes
+	ls -l $DIR/$tdir
+	$LFS getdirstripe $DIR/$tdir
+	ls -l $DIR/$tdir/striped_dir
+	$LFS getdirstripe $DIR/$tdir/striped_dir
 	createmany -o $DIR/$tdir/striped_dir/f 10 ||
 		error "create 10 files failed"
 
 	#remote striped directory
 	$LFS setdirstripe -i 1 -c 2 $DIR/$tdir/remote_striped_dir ||
 		error "set striped dir error"
+        #look at the directories for debug purposes
+	ls -l $DIR/$tdir
+	$LFS getdirstripe $DIR/$tdir
+	ls -l $DIR/$tdir/remote_striped_dir
+	$LFS getdirstripe $DIR/$tdir/remote_striped_dir
 	createmany -o $DIR/$tdir/remote_striped_dir/f 10 ||
 		error "create 10 files failed"
 
@@ -20055,7 +20282,7 @@ test_300q() {
 run_test 300q "create remote directory under orphan directory"
 
 test_300r() {
-	[ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.7.55) ] &&
+	[ $MDS1_VERSION -lt $(version_code 2.7.55) ] &&
 		skip "Need MDS version at least 2.7.55" && return
 	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return
 
@@ -20370,7 +20597,7 @@ test_316() {
 	chown nobody $DIR/$tdir/d
 	touch $DIR/$tdir/d/file
 
-	$LFS mv -M1 $DIR/$tdir/d || error "lfs mv failed"
+	$LFS mv -m1 $DIR/$tdir/d || error "lfs mv failed"
 }
 run_test 316 "lfs mv"
 
@@ -20452,9 +20679,8 @@ test_318() {
 			   llite.*.max_read_ahead_async_active 2>/dev/null)
 	[ $max_active -ne 256 ] && error "expected 256 but got $max_active"
 
-	# currently reset to 0 is unsupported, leave it 512 for now.
-	$LCTL set_param llite.*.max_read_ahead_async_active=0 &&
-		error "set max_read_ahead_async_active should fail"
+	$LCTL set_param llite.*.max_read_ahead_async_active=0 ||
+		error "set max_read_ahead_async_active should succeed"
 
 	$LCTL set_param llite.*.max_read_ahead_async_active=512
 	max_active=$($LCTL get_param -n \
@@ -20513,6 +20739,105 @@ test_319() {
 }
 run_test 319 "lost lease lock on migrate error"
 
+test_398a() { # LU-4198
+	$LFS setstripe -c 1 -i 0 $DIR/$tfile
+	$LCTL set_param ldlm.namespaces.*.lru_size=clear
+
+	# request a new lock on client
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=1
+
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=1 oflag=direct conv=notrunc
+	local lock_count=$($LCTL get_param -n \
+			   ldlm.namespaces.*-OST0000-osc-ffff*.lru_size)
+	[[ $lock_count -eq 0 ]] || error "lock should be cancelled by direct IO"
+
+	$LCTL set_param ldlm.namespaces.*-OST0000-osc-ffff*.lru_size=clear
+
+	# no lock cached, should use lockless IO and not enqueue new lock
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=1 oflag=direct conv=notrunc
+	lock_count=$($LCTL get_param -n \
+		     ldlm.namespaces.*-OST0000-osc-ffff*.lru_size)
+	[[ $lock_count -eq 0 ]] || error "no lock should be held by direct IO"
+}
+run_test 398a "direct IO should cancel lock otherwise lockless"
+
+test_398b() { # LU-4198
+	which fio || skip_env "no fio installed"
+	$LFS setstripe -c -1 $DIR/$tfile
+
+	local size=12
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=$size
+
+	local njobs=4
+	echo "mix direct rw ${size}M to OST0 by fio with $njobs jobs..."
+	fio --name=rand-rw --rw=randrw --bs=$PAGE_SIZE --direct=1 \
+		--numjobs=$njobs --fallocate=none \
+		--iodepth=16 --allow_file_create=0 --size=$((size/njobs))M \
+		--filename=$DIR/$tfile &
+	bg_pid=$!
+
+	echo "mix buffer rw ${size}M to OST0 by fio with $njobs jobs..."
+	fio --name=rand-rw --rw=randrw --bs=$PAGE_SIZE \
+		--numjobs=$njobs --fallocate=none \
+		--iodepth=16 --allow_file_create=0 --size=$((size/njobs))M \
+		--filename=$DIR/$tfile || true
+	wait $bg_pid
+
+	rm -rf $DIR/$tfile
+}
+run_test 398b "DIO and buffer IO race"
+
+test_398c() { # LU-4198
+	which fio || skip_env "no fio installed"
+
+	saved_debug=$($LCTL get_param -n debug)
+	$LCTL set_param debug=0
+
+	local size=$(lctl get_param -n osc.$FSNAME-OST0000*.kbytesavail | head -1)
+	((size /= 1024)) # by megabytes
+	((size /= 2)) # write half of the OST at most
+	[ $size -gt 40 ] && size=40 #reduce test time anyway
+
+	$LFS setstripe -c 1 $DIR/$tfile
+
+	# it seems like ldiskfs reserves more space than necessary if the
+	# writing blocks are not mapped, so it extends the file firstly
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=$size && sync
+	cancel_lru_locks osc
+
+	# clear and verify rpc_stats later
+	$LCTL set_param osc.${FSNAME}-OST0000-osc-ffff*.rpc_stats=clear
+
+	local njobs=4
+	echo "writing ${size}M to OST0 by fio with $njobs jobs..."
+	fio --name=rand-write --rw=randwrite --bs=$PAGE_SIZE --direct=1 \
+		--numjobs=$njobs --fallocate=none --ioengine=libaio \
+		--iodepth=16 --allow_file_create=0 --size=$((size/njobs))M \
+		--filename=$DIR/$tfile
+	[ $? -eq 0 ] || error "fio write error"
+
+	[ $($LCTL get_param -n \
+	 ldlm.namespaces.${FSNAME}-OST0000-osc-ffff*.lock_count) -eq 0 ] ||
+		error "Locks were requested while doing AIO"
+
+	# get the percentage of 1-page I/O
+	pct=$($LCTL get_param osc.${FSNAME}-OST0000-osc-ffff*.rpc_stats |
+		grep -A 1 'pages per rpc' | grep -v 'pages per rpc' |
+		awk '{print $7}')
+	[ $pct -le 50 ] || error "$pct% of I/O are 1-page"
+
+	echo "mix rw ${size}M to OST0 by fio with $njobs jobs..."
+	fio --name=rand-rw --rw=randrw --bs=$PAGE_SIZE --direct=1 \
+		--numjobs=$njobs --fallocate=none --ioengine=libaio \
+		--iodepth=16 --allow_file_create=0 --size=$((size/njobs))M \
+		--filename=$DIR/$tfile
+	[ $? -eq 0 ] || error "fio mixed read write error"
+
+	rm -rf $DIR/$tfile
+	$LCTL set_param debug="$saved_debug"
+}
+run_test 398c "run fio to test AIO"
+
 test_fake_rw() {
 	local read_write=$1
 	if [ "$read_write" = "write" ]; then
@@ -20598,6 +20923,10 @@ test_400a() { # LU-1606, was conf-sanity test_74
 	local prefix=/usr/include/lustre
 	local prog
 
+	# Oleg removes c files in his test rig so test if any c files exist
+	[ -z "$(ls -A $LUSTRE_TESTS_API_DIR)" ] && \
+		skip_env "Needed c test files are missing"
+
 	if ! [[ -d $prefix ]]; then
 		# Assume we're running in tree and fixup the include path.
 		extra_flags+=" -I$LUSTRE/../lnet/include/uapi -I$LUSTRE/include/uapi -I$LUSTRE/include"
@@ -20605,7 +20934,7 @@ test_400a() { # LU-1606, was conf-sanity test_74
 	fi
 
 	for prog in $LUSTRE_TESTS_API_DIR/*.c; do
-		$CC -Wall -Werror $extra_flags -o $out $prog -llustreapi ||
+		$CC -Wall -Werror -std=c99 $extra_flags -o $out $prog -llustreapi ||
 			error "client api broken"
 	done
 	rm -f $out
@@ -20637,7 +20966,7 @@ test_400b() { # LU-1606, LU-5011
 			continue # lustre_ioctl.h is internal header
 		fi
 
-		$CC -Wall -Werror -include $header -c -x c /dev/null -o $out ||
+		$CC -Wall -Werror -std=c99 -include $header -c -x c /dev/null -o $out ||
 			error "cannot compile '$header'"
 	done
 	rm -f $out
@@ -21026,7 +21355,7 @@ run_test 411 "Slab allocation error with cgroup does not LBUG"
 
 test_412() {
 	[ $MDSCOUNT -lt 2 ] && skip_env "needs >= 2 MDTs"
-	if [ $(lustre_version_code mds1) -lt $(version_code 2.10.55) ]; then
+	if [ $MDS1_VERSION -lt $(version_code 2.10.55) ]; then
 		skip "Need server version at least 2.10.55"
 	fi
 
@@ -21294,7 +21623,7 @@ run_test 414 "simulate ENOMEM in ptlrpc_register_bulk()"
 
 test_415() {
 	[ $PARALLEL == "yes" ] && skip "skip parallel run"
-	[ $(lustre_version_code mds1) -lt $(version_code 2.11.52) ] &&
+	[ $MDS1_VERSION -lt $(version_code 2.11.52) ] &&
 		skip "Need server version at least 2.11.52"
 
 	# LU-11102
@@ -21336,7 +21665,7 @@ test_415() {
 run_test 415 "lock revoke is not missing"
 
 test_416() {
-	[ $(lustre_version_code mds1) -lt $(version_code 2.11.55) ] &&
+	[ $MDS1_VERSION -lt $(version_code 2.11.55) ] &&
 		skip "Need server version at least 2.11.55"
 
 	# define OBD_FAIL_OSD_TXN_START    0x19a
@@ -21790,8 +22119,29 @@ test_422() {
 }
 run_test 422 "kill a process with RPC in progress"
 
+stat_test() {
+    df -h $MOUNT &
+    df -h $MOUNT &
+    df -h $MOUNT &
+    df -h $MOUNT &
+    df -h $MOUNT &
+    df -h $MOUNT &
+}
+
+test_423() {
+    local _stats
+    # ensure statfs cache is expired
+    sleep 2;
+
+    _stats=$(stat_test | grep $MOUNT | sort -u | wc -l)
+    [[ ${_stats} -ne 1 ]] && error "statfs wrong"
+
+    return 0
+}
+run_test 423 "statfs should return a right data"
+
 prep_801() {
-	[[ $(lustre_version_code mds1) -lt $(version_code 2.9.55) ]] ||
+	[[ $MDS1_VERSION -lt $(version_code 2.9.55) ]] ||
 	[[ $OST1_VERSION -lt $(version_code 2.9.55) ]] &&
 		skip "Need server version at least 2.9.55"
 
@@ -22022,7 +22372,7 @@ cleanup_802a() {
 
 test_802a() {
 	[[ $mds1_FSTYPE = zfs ]] || skip "ZFS specific test"
-	[[ $(lustre_version_code mds1) -lt $(version_code 2.9.55) ]] ||
+	[[ $MDS1_VERSION -lt $(version_code 2.9.55) ]] ||
 	[[ $OST1_VERSION -lt $(version_code 2.9.55) ]] &&
 		skip "Need server version at least 2.9.55"
 
@@ -22541,7 +22891,7 @@ test_810() {
 run_test 810 "partial page writes on ZFS (LU-11663)"
 
 test_811() {
-	[ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.11.56) ] &&
+	[ $MDS1_VERSION -lt $(version_code 2.11.56) ] &&
 		skip "Need MDS version at least 2.11.56"
 
 	#define OBD_FAIL_MDS_ORPHAN_DELETE	0x165
@@ -22551,8 +22901,7 @@ test_811() {
 	stop mds1
 	start mds1 $(mdsdevname 1) $MDS_MOUNT_OPTS
 
-	sleep 5
-	[[ $(do_facet mds1 pgrep orph_.*-MDD | wc -l) -eq 0 ]] ||
+	wait_update_facet mds1 "pgrep orph_.*-MDD | wc -l" "0" ||
 		error "MDD orphan cleanup thread not quit"
 }
 run_test 811 "orphan name stub can be cleaned up in startup"
@@ -22872,6 +23221,42 @@ test_819b() {
 }
 run_test 819b "too big niobuf in write"
 
+
+function test_820_start_ost() {
+	sleep 5
+
+	for num in $(seq $OSTCOUNT); do
+		start ost$num $(ostdevname $num) $OST_MOUNT_OPTS
+	done
+}
+
+test_820() {
+	[[ $MDSCOUNT -lt 2 ]] && skip_env "needs >= 2 MDTs"
+
+	mkdir $DIR/$tdir
+	umount_client $MOUNT || error "umount failed"
+	for num in $(seq $OSTCOUNT); do
+		stop ost$num
+	done
+
+	# mount client with no active OSTs
+	# so that the client can't initialize max LOV EA size
+	# from OSC notifications
+	mount_client $MOUNT || error "mount failed"
+	# delay OST starting to keep this 0 max EA size for a while
+	test_820_start_ost &
+
+	# create a directory on MDS2
+	test_mkdir -i 1 -c1 $DIR/$tdir/mds2 ||
+		error "Failed to create directory"
+	# open intent should update default EA size
+	# see mdc_update_max_ea_from_body()
+	# notice this is the very first RPC to MDS2
+	cp /etc/services $DIR/$tdir/mds2 ||
+		error "Failed to copy files to mds$n"
+}
+run_test 820 "update max EA from open intent"
+
 #
 # tests that do cleanup/setup should be run at the end
 #
@@ -22915,6 +23300,18 @@ test_901() {
 }
 run_test 901 "don't leak a mgc lock on client umount"
 
+# LU-13377
+test_902() {
+	[ $CLIENT_VERSION -lt $(version_code 2.13.52) ] &&
+		skip "client does not have LU-13377 fix"
+	#define OBD_FAIL_LLITE_SHORT_COMMIT 0x1415
+	$LCTL set_param fail_loc=0x1415
+	dd if=/dev/zero of=$DIR/$tfile bs=1M count=1
+	cancel_lru_locks osc
+	rm -f $DIR/$tfile
+}
+run_test 902 "test short write doesn't hang lustre"
+
 complete $SECONDS
 [ -f $EXT2_DEV ] && rm $EXT2_DEV || true
 check_and_cleanup_lustre