X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Ftests%2Fsanity-hsm.sh;h=3fb1e30f74f5defb994896fd8d7d8ac3bda1a543;hb=6dc66068546824a7a0da23fab64fd4934f6be095;hp=1aa09b9c7ca003f6cf9156dbbfdfbd4ad020feed;hpb=03e785e6967b2a487b22946c18074fe4438cb5b3;p=fs%2Flustre-release.git

diff --git a/lustre/tests/sanity-hsm.sh b/lustre/tests/sanity-hsm.sh
old mode 100644
new mode 100755
index 1aa09b9..3fb1e30
--- a/lustre/tests/sanity-hsm.sh
+++ b/lustre/tests/sanity-hsm.sh
@@ -11,10 +11,14 @@ SRCDIR=$(dirname $0)
 export PATH=$PWD/$SRCDIR:$SRCDIR:$PWD/$SRCDIR/utils:$PATH:/sbin:/usr/sbin
 
 ONLY=${ONLY:-"$*"}
-# bug number for skipped test:    3815
+# bug number for skipped test:    LU-3815
 ALWAYS_EXCEPT="$SANITY_HSM_EXCEPT 34 35 36"
-# bug number for skipped test:4178         4176
+# bug number for skipped test:LU-5474
+ALWAYS_EXCEPT="$ALWAYS_EXCEPT 90"
+# bug number for skipped test:LU-4178      LU-4176
 ALWAYS_EXCEPT="$ALWAYS_EXCEPT 200 221 223b 31a"
+# bug number for skipped test:LU-3852
+ALWAYS_EXCEPT="$ALWAYS_EXCEPT 251"
 # UPDATE THE COMMENT ABOVE WITH BUG NUMBERS WHEN CHANGING ALWAYS_EXCEPT!
 
 LUSTRE=${LUSTRE:-$(cd $(dirname $0)/..; echo $PWD)}
@@ -26,7 +30,6 @@ init_logging
 
 MULTIOP=${MULTIOP:-multiop}
 OPENFILE=${OPENFILE:-openfile}
-MCREATE=${MCREATE:-mcreate}
 MOUNT_2=${MOUNT_2:-"yes"}
 FAIL_ON_ERROR=false
 
@@ -90,6 +93,9 @@ init_agt_vars() {
 
 	export HSMTOOL=${HSMTOOL:-"lhsmtool_posix"}
 	export HSMTOOL_VERBOSE=${HSMTOOL_VERBOSE:-""}
+	export HSMTOOL_UPDATE_INTERVAL=${HSMTOOL_UPDATE_INTERVAL:=""}
+	export HSMTOOL_EVENT_FIFO=${HSMTOOL_EVENT_FIFO:=""}
+	export HSMTOOL_TESTDIR
 	export HSMTOOL_BASE=$(basename "$HSMTOOL" | cut -f1 -d" ")
 	HSM_ARCHIVE=$(copytool_device $SINGLEAGT)
 	HSM_ARCHIVE_NUMBER=2
@@ -100,6 +106,9 @@ init_agt_vars() {
 
 	# archive is purged at copytool setup
 	HSM_ARCHIVE_PURGE=true
+
+	# Don't allow copytool error upon start/setup
+	HSMTOOL_NOERROR=false
 }
 
 # Get the backend root path for the given agent facet.
@@ -112,6 +121,7 @@ copytool_device() {
 
 # Stop copytool and unregister an existing changelog user.
 cleanup() {
+	copytool_monitor_cleanup
 	copytool_cleanup
 	changelog_cleanup
 	cdt_set_sanity_policy
@@ -124,7 +134,7 @@ get_mdt_devices() {
 		local idx=$(($mdtno - 1))
 		MDT[$idx]=$($LCTL get_param -n \
 			mdc.$FSNAME-MDT000${idx}-mdc-*.mds_server_uuid |
-			awk '{gsub(/_UUID/,""); print $1}' | head -1)
+			awk '{gsub(/_UUID/,""); print $1}' | head -n1)
 	done
 }
 
@@ -140,6 +150,63 @@ search_and_kill_copytool() {
 	do_nodesv $agents "killall -q $HSMTOOL_BASE" || true
 }
 
+copytool_monitor_setup() {
+	local facet=${1:-$SINGLEAGT}
+	local agent=$(facet_active_host $facet)
+
+	local cmd="mktemp --tmpdir=/tmp -d ${TESTSUITE}.${TESTNAME}.XXXX"
+	local test_dir=$(do_node $agent "$cmd") ||
+		error "Failed to create tempdir on $agent"
+	export HSMTOOL_MONITOR_DIR=$test_dir
+
+	# Create the fifo and a monitor (cat dies when copytool dies)
+	do_node $agent "mkfifo -m 0644 $test_dir/fifo" ||
+		error "failed to create copytool fifo on $agent"
+	cmd="cat $test_dir/fifo > $test_dir/events &"
+	cmd+=" echo \\\$! > $test_dir/monitor_pid"
+
+	if [[ $PDSH == *Rmrsh* ]]; then
+		# This is required for pdsh -Rmrsh and its handling of remote
+		# shells.
+		# Regular ssh and pdsh -Rssh work fine without this
+		# backgrounded subshell nonsense.
+		(do_node $agent "$cmd") &
+		export HSMTOOL_MONITOR_PDSH=$!
+
+		# Slightly racy, but just making a best-effort to catch obvious
+		# problems.
+		sleep 1
+		ps -p $HSMTOOL_MONITOR_PDSH >&- ||
+			error "Failed to start copytool monitor on $agent"
+	else
+		do_node $agent "$cmd"
+		if [ $? != 0 ]; then
+			error "Failed to start copytool monitor on $agent"
+		fi
+	fi
+}
+
+copytool_monitor_cleanup() {
+	local facet=${1:-$SINGLEAGT}
+	local agent=$(facet_active_host $facet)
+
+	if [ -n "$HSMTOOL_MONITOR_DIR" ]; then
+		# Should die when the copytool dies, but just in case.
+		local cmd="kill \\\$(cat $HSMTOOL_MONITOR_DIR/monitor_pid)"
+		cmd+=" 2>/dev/null || true"
+		do_node $agent "$cmd"
+		do_node $agent "rm -fr $HSMTOOL_MONITOR_DIR"
+		export HSMTOOL_MONITOR_DIR=
+	fi
+
+	# The pdsh should die on its own when the monitor dies. Just
+	# in case, though, try to clean up to avoid any cruft.
+	if [ -n "$HSMTOOL_MONITOR_PDSH" ]; then
+		kill $HSMTOOL_MONITOR_PDSH 2>/dev/null
+		export HSMTOOL_MONITOR_PDSH=
+	fi
+}
+
 copytool_setup() {
 	local facet=${1:-$SINGLEAGT}
 	local lustre_mntpnt=${2:-$MOUNT}
@@ -164,6 +231,10 @@ copytool_setup() {
 	# independent of hardware
 	local cmd="$HSMTOOL $HSMTOOL_VERBOSE --daemon --hsm-root $hsm_root"
 	[[ -z "$arc_id" ]] || cmd+=" --archive $arc_id"
+	[[ -z "$HSMTOOL_UPDATE_INTERVAL" ]] ||
+		cmd+=" --update-interval $HSMTOOL_UPDATE_INTERVAL"
+	[[ -z "$HSMTOOL_EVENT_FIFO" ]] ||
+		cmd+=" --event-fifo $HSMTOOL_EVENT_FIFO"
 	cmd+=" --bandwidth 1 $lustre_mntpnt"
 
 	# Redirect the standard output and error to a log file which
@@ -172,18 +243,36 @@ copytool_setup() {
 	[[ -z "$TESTNAME" ]] || prefix=$prefix.$TESTNAME
 	local copytool_log=$prefix.copytool${arc_id}_log.$agent.log
 
-	do_facet $facet "$cmd < /dev/null > $copytool_log 2>&1" ||
-		error "start copytool $facet on $agent failed"
+	do_facet $facet "$cmd < /dev/null > $copytool_log 2>&1"
+	if [[ $? !=  0 ]]; then
+		[[ $HSMTOOL_NOERROR == true ]] ||
+			error "start copytool $facet on $agent failed"
+		echo "start copytool $facet on $agent failed"
+	fi
+
 	trap cleanup EXIT
 }
 
+get_copytool_event_log() {
+	local facet=${1:-$SINGLEAGT}
+	local agent=$(facet_active_host $facet)
+
+	[ -z "$HSMTOOL_MONITOR_DIR" ] &&
+		error "Can't get event log: No monitor directory!"
+
+	do_node $agent "cat $HSMTOOL_MONITOR_DIR/events" ||
+		error "Could not collect event log from $agent"
+}
+
 copytool_cleanup() {
 	trap - EXIT
-	local agents=${1:-$(facet_active_host $SINGLEAGT)}
+	local facet=$SINGLEAGT
+	local agents=${1:-$(facet_active_host $facet)}
 	local mdtno
 	local idx
 	local oldstate
 	local mdt_hsmctrl
+	local hsm_root=$(copytool_device $facet)
 
 	do_nodesv $agents "pkill -INT -x $HSMTOOL_BASE" || return 0
 	sleep 1
@@ -208,6 +297,7 @@ copytool_cleanup() {
 			"$oldstate" 20 ||
 			error "mds${mdtno} cdt state is not $oldstate"
 	done
+	do_facet $facet "rm -rf $hsm_root"
 }
 
 copytool_suspend() {
@@ -493,7 +583,7 @@ cleanup_large_files() {
 
 make_large_for_striping() {
 	local file2=${1/$DIR/$DIR2}
-	local sz=$($LCTL get_param -n lov.*-clilov-*.stripesize | head -1)
+	local sz=$($LCTL get_param -n lov.*-clilov-*.stripesize | head -n1)
 
 	cleanup_large_files
 
@@ -580,8 +670,10 @@ get_request_count() {
 
 wait_all_done() {
 	local timeout=$1
+	local fid=$2
 
 	local cmd="$LCTL get_param -n $HSM_PARAM.actions"
+	[[ -n $fid ]] && cmd+=" | grep '$fid'"
 	cmd+=" | egrep 'WAITING|STARTED'"
 
 	wait_result $SINGLEMDS "$cmd" "" $timeout ||
@@ -593,6 +685,18 @@ wait_for_grace_delay() {
 	sleep $val
 }
 
+parse_json_event() {
+	local raw_event=$1
+
+	# python2.6 in EL6 includes an internal json module
+	local json_parser='import json; import fileinput;'
+	json_parser+=' print "\n".join(["local %s=\"%s\"" % tuple for tuple in '
+	json_parser+='json.loads([line for line in '
+	json_parser+='fileinput.input()][0]).items()])'
+
+	echo $raw_event | python -c "$json_parser"
+}
+
 # populate MDT device array
 get_mdt_devices
 
@@ -891,7 +995,7 @@ test_10d() {
 }
 run_test 10d "Archive a file on the default archive id"
 
-test_11() {
+test_11a() {
 	mkdir -p $DIR/$tdir
 	copy2archive /etc/hosts $tdir/$tfile
 	local f=$DIR/$tdir/$tfile
@@ -915,7 +1019,31 @@ test_11() {
 	local AFILE=$(do_facet $SINGLEAGT ls $HSM_ARCHIVE'/*/*/*/*/*/*/'$fid) ||
 		error "fid $fid not in archive $HSM_ARCHIVE"
 }
-run_test 11 "Import a file"
+run_test 11a "Import a file"
+
+test_11b() {
+	# test needs a running copytool
+	copytool_setup
+
+	mkdir -p $DIR/$tdir
+	local f=$DIR/$tdir/$tfile
+	local fid=$(copy_file /etc/hosts $f)
+	$LFS hsm_archive -a $HSM_ARCHIVE_NUMBER $f ||
+		error "hsm_archive failed"
+	wait_request_state $fid ARCHIVE SUCCEED
+
+	local FILE_HASH=$(md5sum $f)
+	rm -f $f
+
+	import_file $fid $f
+
+	echo "$FILE_HASH" | md5sum -c
+
+	[[ $? -eq 0 ]] || error "Restored file differs"
+
+	copytool_cleanup
+}
+run_test 11b "Import a deleted file using its FID"
 
 test_12a() {
 	# test needs a running copytool
@@ -926,16 +1054,16 @@ test_12a() {
 
 	local f=$DIR/$tdir/$tfile
 	import_file $tdir/$tfile $f
-	local f=$DIR2/$tdir/$tfile
+	local f2=$DIR2/$tdir/$tfile
 	echo "Verifying released state: "
-	check_hsm_flags $f "0x0000000d"
+	check_hsm_flags $f2 "0x0000000d"
 
-	local fid=$(path2fid $f)
-	$LFS hsm_restore $f
+	local fid=$(path2fid $f2)
+	$LFS hsm_restore $f2
 	wait_request_state $fid RESTORE SUCCEED
 
 	echo "Verifying file state: "
-	check_hsm_flags $f "0x00000009"
+	check_hsm_flags $f2 "0x00000009"
 
 	do_facet $SINGLEAGT diff -q $HSM_ARCHIVE/$tdir/$tfile $f
 
@@ -1155,6 +1283,57 @@ test_12n() {
 }
 run_test 12n "Import/implicit restore/release"
 
+test_12o() {
+	# test needs a running copytool
+	copytool_setup
+
+	mkdir -p $DIR/$tdir
+	local f=$DIR/$tdir/$tfile
+	local fid=$(copy_file /etc/hosts $f)
+
+	$LFS hsm_archive --archive $HSM_ARCHIVE_NUMBER $f
+	wait_request_state $fid ARCHIVE SUCCEED
+	$LFS hsm_release $f || error "release of $f failed"
+
+#define OBD_FAIL_MDS_HSM_SWAP_LAYOUTS		0x152
+	do_facet $SINGLEMDS lctl set_param fail_loc=0x152
+
+	# set no retry action mode
+	cdt_set_no_retry
+
+	diff -q /etc/hosts $f
+	local st=$?
+
+	# we check we had a restore failure
+	wait_request_state $fid RESTORE FAILED
+
+	[[ $st -eq 0 ]] && error "Restore must fail"
+
+	# remove no retry action mode
+	cdt_clear_no_retry
+
+	# check file is still released
+	check_hsm_flags $f "0x0000000d"
+
+	# retry w/o failure injection
+	do_facet $SINGLEMDS lctl set_param fail_loc=0
+
+	# to be sure previous RESTORE result is gone
+	cdt_purge
+	wait_for_grace_delay
+
+	diff -q /etc/hosts $f
+	st=$?
+
+	# we check we had a restore done
+	wait_request_state $fid RESTORE SUCCEED
+
+	[[ $st -eq 0 ]] || error "Restored file differs"
+
+	copytool_cleanup
+}
+run_test 12o "Layout-swap failure during Restore leaves file released"
+
 test_13() {
 	# test needs a running copytool
 	copytool_setup
@@ -2084,6 +2263,15 @@ test_33() {
 	wait_request_state $fid ARCHIVE SUCCEED
 	$LFS hsm_release $f
 
+	# to be sure wait_all_done will not be mislead by previous tests
+	# and ops.
+	cdt_purge
+	wait_for_grace_delay
+	# Also raise grace_delay significantly so the Canceled
+	# Restore action will stay enough long avail.
+	local old_grace=$(get_hsm_param grace_delay)
+	set_hsm_param grace_delay 100
+
 	md5sum $f >/dev/null &
 	local pid=$!
 	wait_request_state $fid RESTORE STARTED
@@ -2096,8 +2284,29 @@ test_33() {
 
 	$LFS hsm_cancel $f
 
-	wait_request_state $fid RESTORE CANCELED
-	wait_request_state $fid CANCEL SUCCEED
+	# instead of waiting+checking both Restore and Cancel ops
+	# sequentially, wait for both to be finished and then check
+	# each results.
+	wait_all_done 100 $fid
+	local rstate=$(get_request_state $fid RESTORE)
+	local cstate=$(get_request_state $fid CANCEL)
+
+	# restore orig grace_delay.
+	set_hsm_param grace_delay $old_grace
+
+	if [[ "$rstate" == "CANCELED" ]] ; then
+		[[ "$cstate" == "SUCCEED" ]] ||
+			error "Restore state is CANCELED and Cancel state " \
+			       "is not SUCCEED but $cstate"
+		echo "Restore state is CANCELED, Cancel state is SUCCEED"
+	elif [[ "$rstate" == "SUCCEED" ]] ; then
+		[[ "$cstate" == "FAILED" ]] ||
+			error "Restore state is SUCCEED and Cancel state " \
+				"is not FAILED but $cstate"
+		echo "Restore state is SUCCEED, Cancel state is FAILED"
+	else
+		error "Restore state is $rstate and Cancel state is $cstate"
+	fi
 
 	[ -z $killed ] ||
 		error "Cannot kill process waiting for restore ($killed)"
@@ -2484,10 +2693,295 @@ test_58() {
 }
 run_test 58 "Truncate a released file will trigger restore"
 
-test_90() {
-	file_count=57
+test_60() {
+	# This test validates the fix for LU-4512. Ensure that the -u
+	# option changes the progress reporting interval from the default
+	# (30 seconds) to the user-specified interval.
+	local interval=5
+	local progress_timeout=$((interval * 3))
+
+	# test needs a new running copytool
+	copytool_cleanup
+	HSMTOOL_UPDATE_INTERVAL=$interval copytool_setup
+
+	mkdir -p $DIR/$tdir
+	local f=$DIR/$tdir/$tfile
+	local fid=$(make_large_for_progress $f)
+
+	local start_at=$(date +%s)
+	$LFS hsm_archive --archive $HSM_ARCHIVE_NUMBER $f ||
+		error "could not archive file"
+
+	local mdtidx=0
+	local mdt=${MDT_PREFIX}${mdtidx}
+	local mds=mds$((mdtidx + 1))
+
+	local cmd="$LCTL get_param -n ${mdt}.hsm.active_requests"
+	cmd+=" | awk '/'$fid'.*action=ARCHIVE/ {print \\\$12}' | cut -f2 -d="
+
+	local RESULT
+	local WAIT=0
+	local sleep=1
+
+	echo -n "Expecting a progress update within $progress_timeout seconds... "
+	while [ true ]; do
+		RESULT=$(do_node $(facet_active_host $mds) "$cmd")
+		if [ $RESULT -gt 0 ]; then
+			echo "$RESULT bytes copied in $WAIT seconds."
+			break
+		elif [ $WAIT -ge $progress_timeout ]; then
+			error "Timed out waiting for progress update!"
+			break
+		fi
+		WAIT=$((WAIT + sleep))
+		sleep $sleep
+	done
+
+	local finish_at=$(date +%s)
+	local elapsed=$((finish_at - start_at))
+
+	# Ensure that the progress update occurred within the expected window.
+	if [ $elapsed -lt $interval ]; then
+		error "Expected progress update after at least $interval seconds"
+	fi
+
+	cdt_clear_no_retry
+	copytool_cleanup
+}
+run_test 60 "Changing progress update interval from default"
+
+test_70() {
+	# test needs a new running copytool
+	copytool_cleanup
+	copytool_monitor_setup
+	HSMTOOL_EVENT_FIFO=$HSMTOOL_MONITOR_DIR/fifo copytool_setup
+
+	# Just start and stop the copytool to generate events.
+	cdt_clear_no_retry
+
+	# Wait for the copytool to register.
+	wait_update --verbose $(facet_active_host mds1) \
+		"$LCTL get_param -n ${MDT_PREFIX}0.hsm.agents | grep -o ^uuid" \
+		uuid 100 ||
+		error "copytool failed to register with MDT0000"
+
+	copytool_cleanup
+
+	local REGISTER_EVENT
+	local UNREGISTER_EVENT
+	while read event; do
+		local parsed=$(parse_json_event "$event")
+		if [ -z "$parsed" ]; then
+			error "Copytool sent malformed event: $event"
+		fi
+		eval $parsed
+
+		if [ $event_type == "REGISTER" ]; then
+			REGISTER_EVENT=$event
+		elif [ $event_type == "UNREGISTER" ]; then
+			UNREGISTER_EVENT=$event
+		fi
+	done < <(echo $"$(get_copytool_event_log)")
+
+	if [ -z "$REGISTER_EVENT" ]; then
+		error "Copytool failed to send register event to FIFO"
+	fi
+
+	if [ -z "$UNREGISTER_EVENT" ]; then
+		error "Copytool failed to send unregister event to FIFO"
+	fi
+
+	copytool_monitor_cleanup
+	echo "Register/Unregister events look OK."
+}
+run_test 70 "Copytool logs JSON register/unregister events to FIFO"
+
+test_71() {
+	# Bump progress interval for livelier events.
+	local interval=5
+
+	# test needs a new running copytool
+	copytool_cleanup
+	copytool_monitor_setup
+	HSMTOOL_UPDATE_INTERVAL=$interval \
+	HSMTOOL_EVENT_FIFO=$HSMTOOL_MONITOR_DIR/fifo copytool_setup
+
 	mkdir -p $DIR/$tdir
 	local f=$DIR/$tdir/$tfile
+	local fid=$(make_large_for_progress $f)
+
+	$LFS hsm_archive --archive $HSM_ARCHIVE_NUMBER $f ||
+		error "could not archive file"
+	wait_request_state $fid ARCHIVE SUCCEED
+
+	local expected_fields="event_time data_fid source_fid"
+	expected_fields+=" total_bytes current_bytes"
+
+	local START_EVENT
+	local FINISH_EVENT
+	while read event; do
+		# Make sure we're not getting anything from previous events.
+		for field in $expected_fields; do
+			unset $field
+		done
+
+		local parsed=$(parse_json_event "$event")
+		if [ -z "$parsed" ]; then
+			error "Copytool sent malformed event: $event"
+		fi
+		eval $parsed
+
+		if [ $event_type == "ARCHIVE_START" ]; then
+			START_EVENT=$event
+			continue
+		elif [ $event_type == "ARCHIVE_FINISH" ]; then
+			FINISH_EVENT=$event
+			continue
+		elif [ $event_type != "ARCHIVE_RUNNING" ]; then
+			continue
+		fi
+
+		# Do some simple checking of the progress update events.
+		for expected_field in $expected_fields; do
+			if [ -z ${!expected_field+x} ]; then
+				error "Missing $expected_field field in event"
+			fi
+		done
+
+		if [ $total_bytes -eq 0 ]; then
+			error "Expected total_bytes to be > 0"
+		fi
+
+		# These should be identical throughout an archive
+		# operation.
+		if [ $source_fid != $data_fid ]; then
+			error "Expected source_fid to equal data_fid"
+		fi
+	done < <(echo $"$(get_copytool_event_log)")
+
+	if [ -z "$START_EVENT" ]; then
+		error "Copytool failed to send archive start event to FIFO"
+	fi
+
+	if [ -z "$FINISH_EVENT" ]; then
+		error "Copytool failed to send archive finish event to FIFO"
+	fi
+
+	echo "Archive events look OK."
+
+	cdt_clear_no_retry
+	copytool_cleanup
+	copytool_monitor_cleanup
+}
+run_test 71 "Copytool logs JSON archive events to FIFO"
+
+test_72() {
+	# Bump progress interval for livelier events.
+	local interval=5
+
+	# test needs a new running copytool
+	copytool_cleanup
+	copytool_monitor_setup
+	HSMTOOL_UPDATE_INTERVAL=$interval \
+	HSMTOOL_EVENT_FIFO=$HSMTOOL_MONITOR_DIR/fifo copytool_setup
+	local test_file=$HSMTOOL_MONITOR_DIR/file
+
+	local cmd="dd if=/dev/urandom of=$test_file count=16 bs=1000000 "
+	cmd+="conv=fsync"
+	do_facet $SINGLEAGT "$cmd" ||
+		error "cannot create $test_file on $SINGLEAGT"
+	copy2archive $test_file $tdir/$tfile
+
+	mkdir -p $DIR/$tdir
+	local f=$DIR/$tdir/$tfile
+	import_file $tdir/$tfile $f
+	f=$DIR2/$tdir/$tfile
+	echo "Verifying released state: "
+	check_hsm_flags $f "0x0000000d"
+
+	local fid=$(path2fid $f)
+	$LFS hsm_restore $f
+	wait_request_state $fid RESTORE SUCCEED
+
+	local expected_fields="event_time data_fid source_fid"
+	expected_fields+=" total_bytes current_bytes"
+
+	local START_EVENT
+	local FINISH_EVENT
+	while read event; do
+		# Make sure we're not getting anything from previous events.
+		for field in $expected_fields; do
+			unset $field
+		done
+
+		local parsed=$(parse_json_event "$event")
+		if [ -z "$parsed" ]; then
+			error "Copytool sent malformed event: $event"
+		fi
+		eval $parsed
+
+		if [ $event_type == "RESTORE_START" ]; then
+			START_EVENT=$event
+			if [ $source_fid != $data_fid ]; then
+				error "source_fid should == data_fid at start"
+			fi
+			continue
+		elif [ $event_type == "RESTORE_FINISH" ]; then
+			FINISH_EVENT=$event
+			if [ $source_fid != $data_fid ]; then
+				error "source_fid should == data_fid at finish"
+			fi
+			continue
+		elif [ $event_type != "RESTORE_RUNNING" ]; then
+			continue
+		fi
+
+		# Do some simple checking of the progress update events.
+		for expected_field in $expected_fields; do
+			if [ -z ${!expected_field+x} ]; then
+				error "Missing $expected_field field in event"
+			fi
+		done
+
+		if [ $total_bytes -eq 0 ]; then
+			error "Expected total_bytes to be > 0"
+		fi
+
+		# When a restore starts out, the data fid is the same as the
+		# source fid. After the restore has gotten going, we learn
+		# the new data fid. Once the restore has finished, the source
+		# fid is set to the new data fid.
+		#
+		# We test this because some monitoring software may depend on
+		# this behavior. If it changes, then the consumers of these
+		# events may need to be modified.
+		if [ $source_fid == $data_fid ]; then
+			error "source_fid should != data_fid during restore"
+		fi
+	done < <(echo $"$(get_copytool_event_log)")
+
+	if [ -z "$START_EVENT" ]; then
+		error "Copytool failed to send restore start event to FIFO"
+	fi
+
+	if [ -z "$FINISH_EVENT" ]; then
+		error "Copytool failed to send restore finish event to FIFO"
+	fi
+
+	echo "Restore events look OK."
+
+	cdt_clear_no_retry
+	copytool_cleanup
+	copytool_monitor_cleanup
+
+	rm -rf $test_dir
+}
+run_test 72 "Copytool logs JSON restore events to FIFO"
+
+test_90() {
+	file_count=51 # Max number of files constrained by LNET message size
+	mkdir $DIR/$tdir || error "mkdir $DIR/$tdir failed"
+	local f=$DIR/$tdir/$tfile
 	local FILELIST=/tmp/filelist.txt
 	local i=""
 
@@ -3527,9 +4021,9 @@ test_402() {
 	copytool_cleanup
 
 	# deactivate all mdc on agent1
-	mdc_change_state $SINGLEAGT "MDT000." "deactivate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT000." "deactivate"
 
-	copytool_setup $SINGLEAGT
+	HSMTOOL_NOERROR=true copytool_setup $SINGLEAGT
 
 	check_agent_unregistered "uuid" # match any agent
 
@@ -3537,7 +4031,7 @@ test_402() {
 	search_copytools $agent && error "Copytool start should have failed"
 
 	# reactivate MDCs
-	mdc_change_state $SINGLEAGT "MDT000." "activate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT000." "activate"
 }
 run_test 402 "Copytool start fails if all MDTs are inactive"
 
@@ -3551,7 +4045,7 @@ test_403() {
 	local uuid=$(do_rpc_nodes $agent get_client_uuid | cut -d' ' -f2)
 
 	# deactivate all mdc for MDT0001
-	mdc_change_state $SINGLEAGT "MDT0001" "deactivate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT0001" "deactivate"
 
 	copytool_setup
 	# check the agent is registered on MDT0000, and not on MDT0001
@@ -3562,7 +4056,7 @@ test_403() {
 	search_copytools $agent || error "No running copytools on $agent"
 
 	# reactivate all mdc for MDT0001
-	mdc_change_state $SINGLEAGT "MDT0001" "activate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT0001" "activate"
 
 	# make sure the copytool is now registered to all MDTs
 	check_agent_registered $uuid
@@ -3586,7 +4080,7 @@ test_404() {
 	local fid1=$(make_small $dir_mdt0/$tfile)
 
 	# deactivate all mdc for MDT0001
-	mdc_change_state $SINGLEAGT "MDT0001" "deactivate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT0001" "deactivate"
 
 	# send an HSM request for files in MDT0000
 	$LFS hsm_archive $dir_mdt0/$tfile || error "lfs hsm_archive"
@@ -3596,7 +4090,7 @@ test_404() {
 		echo "archive successful on mdt0"
 
 	# reactivate all mdc for MDT0001
-	mdc_change_state $SINGLEAGT "MDT0001" "activate"
+	mdc_change_state $SINGLEAGT "$FSNAME-MDT0001" "activate"
 
 	copytool_cleanup
 	# clean test files and directories
@@ -3604,6 +4098,56 @@ test_404() {
 }
 run_test 404 "Inactive MDT does not block requests for active MDTs"
 
+test_405() {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return
+
+	copytool_setup
+
+	mkdir -p $DIR/$tdir
+
+	local striped_dir=$DIR/$tdir/striped_dir
+
+	# create striped dir on all of MDTs
+	$LFS mkdir -i 0 -c $MDSCOUNT $striped_dir || error "lfs mkdir"
+
+	local fid1=$(make_small_sync $striped_dir/${tfile}_0)
+	local fid2=$(make_small_sync $striped_dir/${tfile}_1)
+	local fid3=$(make_small_sync $striped_dir/${tfile}_2)
+	local fid4=$(make_small_sync $striped_dir/${tfile}_3)
+
+	local idx1=$($LFS getstripe -M $striped_dir/${tfile}_0)
+	local idx2=$($LFS getstripe -M $striped_dir/${tfile}_1)
+	local idx3=$($LFS getstripe -M $striped_dir/${tfile}_2)
+	local idx4=$($LFS getstripe -M $striped_dir/${tfile}_3)
+
+	# check that compound requests are shunt to the rights MDTs
+	$LFS hsm_archive $striped_dir/${tfile}_0 $striped_dir/${tfile}_1  \
+			 $striped_dir/${tfile}_2 $striped_dir/${tfile}_3 ||
+		error "lfs hsm_archive"
+
+	wait_request_state $fid1 ARCHIVE SUCCEED $idx1 &&
+		echo "archive successful on $fid1"
+	wait_request_state $fid2 ARCHIVE SUCCEED $idx2 &&
+		echo "archive successful on $fid2"
+	wait_request_state $fid3 ARCHIVE SUCCEED $idx3 &&
+		echo "archive successful on $fid3"
+	wait_request_state $fid4 ARCHIVE SUCCEED $idx4 &&
+		echo "archive successful on $fid4"
+
+	$LFS hsm_release $striped_dir/${tfile}_0 || error "lfs hsm_release 1"
+	$LFS hsm_release $striped_dir/${tfile}_1 || error "lfs hsm_release 2"
+	$LFS hsm_release $striped_dir/${tfile}_2 || error "lfs hsm_release 3"
+	$LFS hsm_release $striped_dir/${tfile}_3 || error "lfs hsm_release 4"
+
+	cat $striped_dir/${tfile}_0 > /dev/null || error "cat ${tfile}_0 failed"
+	cat $striped_dir/${tfile}_1 > /dev/null || error "cat ${tfile}_1 failed"
+	cat $striped_dir/${tfile}_2 > /dev/null || error "cat ${tfile}_2 failed"
+	cat $striped_dir/${tfile}_3 > /dev/null || error "cat ${tfile}_3 failed"
+
+	copytool_cleanup
+}
+run_test 405 "archive and release under striped directory"
+
 copytool_cleanup
 
 complete $SECONDS