X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Ftests%2Finsanity.sh;h=917c23cf0782ae231af75e502b6f0b44fa1c69e9;hb=0354fa98966eef9874b3fe6818c2c6f1a2433297;hp=7a9b58f40e7901973fd86e4e5fe828af19e50c06;hpb=c5050e412572b00cbe93d8517d2d1f767bebfa92;p=fs%2Flustre-release.git

diff --git a/lustre/tests/insanity.sh b/lustre/tests/insanity.sh
index 7a9b58f..917c23c 100755
--- a/lustre/tests/insanity.sh
+++ b/lustre/tests/insanity.sh
@@ -1,27 +1,41 @@
-#!/bin/sh
-# Test multiple failures, AKA Test 17
+#!/bin/bash
+#
+# Test different failure modes combinations
 
 set -e
 
-LUSTRE=${LUSTRE:-`dirname $0`/..}
+LUSTRE=${LUSTRE:-$(dirname $0)/..}
 . $LUSTRE/tests/test-framework.sh
-
 init_test_env $@
+init_logging
 
-. ${CONFIG:=$LUSTRE/tests/cfg/insanity-local.sh}
-
-ALWAYS_EXCEPT=""
+# bug number for skipped test:
+ALWAYS_EXCEPT="$INSANITY_EXCEPT"
+# UPDATE THE COMMENT ABOVE WITH BUG NUMBERS WHEN CHANGING ALWAYS_EXCEPT!
 
 build_test_filter
 
-assert_env mds_HOST ost1_HOST ost2_HOST client_HOST LIVE_CLIENT 
+if [ "$FAILURE_MODE" = "HARD" ]; then
+	skip_env "$TESTSUITE: is not functional with FAILURE_MODE = HARD, " \
+		"please use recovery-double-scale, bz20407"
+fi
+
+SINGLECLIENT=${SINGLECLIENT:-$HOSTNAME}
+LIVE_CLIENT=${LIVE_CLIENT:-$SINGLECLIENT}
+FAIL_CLIENTS=${FAIL_CLIENTS:-$RCLIENTS}
 
-# This can be a regexp, to allow more clients
-CLIENTS=${CLIENTS:-"`comma_list $LIVE_CLIENT $FAIL_CLIENTS`"}
+assert_env mds_HOST MDSCOUNT
+assert_env ost_HOST OSTCOUNT
+assert_env LIVE_CLIENT FSNAME
 
-CLIENTLIST="$LIVE_CLIENT $FAIL_CLIENTS"
+require_dsh_mds || exit 0
+require_dsh_ost || exit 0
+
+# FAIL_CLIENTS list should not contain the LIVE_CLIENT
+FAIL_CLIENTS=$(echo " $FAIL_CLIENTS " | sed -re "s/\s+$LIVE_CLIENT\s+/ /g")
 
 DIR=${DIR:-$MOUNT}
+TESTDIR=$DIR/d0.$TESTSUITE
 
 #####
 # fail clients round robin
@@ -30,512 +44,708 @@ DIR=${DIR:-$MOUNT}
 FAIL_LIST=($FAIL_CLIENTS)
 FAIL_NUM=${#FAIL_LIST[*]}
 FAIL_NEXT=0
+typeset -i  FAIL_NEXT
 DOWN_NUM=0   # number of nodes currently down
 
-# return next client to fail
-fail_client() {
-    ret=${FAIL_LIST[$FAIL_NEXT]}
-    FAIL_NEXT=$(( (FAIL_NEXT+1) % FAIL_NUM ))
-    echo $ret
+# set next client to fail
+set_fail_client() {
+	FAIL_CLIENT=${FAIL_LIST[$FAIL_NEXT]}
+	FAIL_NEXT=$(( (FAIL_NEXT+1) % FAIL_NUM ))
+	echo "fail $FAIL_CLIENT, next is $FAIL_NEXT"
 }
 
-shutdown_client() {
-    client=$1
-    if [ "$FAILURE_MODE" = HARD ]; then
-       $POWER_DOWN $client
-    elif [ "$FAILURE_MODE" = SOFT ]; then
-       $PDSH $client $LCONF --clenaup --force --nomod $XMLCONFIG
-    fi
+fail_clients() {
+	num=$1
+
+	log "Request fail clients: $num, to fail: $FAIL_NUM, failed: $DOWN_NUM"
+	if [ -z "$num"  ] || [ "$num" -gt $((FAIL_NUM - DOWN_NUM)) ]; then
+		num=$((FAIL_NUM - DOWN_NUM))
+	fi
+
+	if [ -z "$num" ] || [ "$num" -le 0 ]; then
+		log "No clients failed!"
+		return
+	fi
+
+	client_mkdirs
+
+	for i in `seq $num`; do
+		set_fail_client
+		client=$FAIL_CLIENT
+		DOWN_CLIENTS="$DOWN_CLIENTS $client"
+		shutdown_client $client
+	done
+
+	echo "down clients: $DOWN_CLIENTS"
+
+	for client in $DOWN_CLIENTS; do
+		boot_node $client
+	done
+	DOWN_NUM=`echo $DOWN_CLIENTS | wc -w`
+	client_rmdirs
 }
 
-reboot_node() {
-    NODE=$1
-    if [ "$FAILURE_MODE" = HARD ]; then
-       $POWER_UP $NODE
-    fi
+reintegrate_clients() {
+	for client in $DOWN_CLIENTS; do
+		wait_for_host $client
+		echo "Restarting $client"
+		zconf_mount $client $MOUNT || return 1
+	done
+
+	DOWN_CLIENTS=""
+	DOWN_NUM=0
 }
 
-fail_clients() {
-    num=$1
-    if [ -z "$num" -o $num -gt $((FAIL_NUM - DOWN_NUM)) ]; then
-	num=$((FAIL_NUM - DOWN_NUM)) 
-    fi
-    
-    if [ -z "$num"  -o $num -le 0 ]; then
-        return
-    fi
-
-    for i in `seq $num`; do
-       client=`fail_client`
-       DOWN_CLIENTS="$DOWN_CLIENTS $client"
-       client_mkdirs
-       shutdown_client $client
-    done
-
-    for client in $DOWN_CLIENTS; do
-	reboot_node $client
-    done
-    DOWN_NUM=`echo $DOWN_CLIENTS | wc -w`
-    $PDSH $LIVE_CLIENT "cd $MOUNT && rmdir $CLIENTLIST"
+start_ost() {
+	start ost$1 `ostdevname $1` $OST_MOUNT_OPTS
 }
 
-reintegrate_clients() {
-    for client in $DOWN_CLIENTS; do
-	wait_for_host $client
-	$PDSH $client "$LCONF --node client --select mds_svc=`facet_active mds` $CLIENTOPTS $XMLCONFIG"
-    done
-    DOWN_CLIENTS=""
-    DOWN_NUM=0
+start_mdt() {
+	start mds$1 $(mdsdevname $1) $MDS_MOUNT_OPTS
 }
 
-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
+trap exit INT
 
-    add_lov lov1 mds --stripe_sz $STRIPE_BYTES\
-	--stripe_cnt $STRIPES_PER_OBJ --stripe_pattern 0
-    add_ost ost1 --lov lov1 --dev $OSTDEV --size $OSTSIZE
-    add_ost ost2 --lov lov1 --dev ${OSTDEV}-2 --size $OSTSIZE
-    add_client client mds --lov lov1 --path $MOUNT
+client_touch() {
+	file=$1
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		echo $DOWN_CLIENTS | grep -q $c && continue
+		$PDSH $c touch $TESTDIR/${c}_$file || return 1
+	done
 }
 
-setup() {
-    wait_for ost1
-    start ost1 ${REFORMAT} $OSTLCONFARGS 
-    wait_for ost2
-    start ost2 ${REFORMAT} $OSTLCONFARGS 
-    [ "$DAEMONFILE" ] && $LCTL debug_daemon start $DAEMONFILE $DAEMONSIZE
-    wait_for mds
-    start mds $MDSLCONFARGS ${REFORMAT}
-    while ! $PDSH $HOST "ls -ld $LUSTRE"; do sleep 5; done
-    do_node $CLIENTS lconf --node client_facet \
-	--select mds_service=$ACTIVEMDS $XMLCONFIG
+client_rm() {
+	file=$1
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		$PDSH $c rm $TESTDIR/${c}_$file
+	done
 }
 
-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
-    for node in $CLIENTS; do
-	do_node $node lconf ${FORCE} --select mds_svc=${activemds}_facet --cleanup --node client_facet $XMLCONFIG || true
-    done
-
-    stop mds ${FORCE} $MDSLCONFARGS
-    stop ost1 ${FORCE}
-    stop ost2 ${FORCE} --dump cleanup.log
+client_mkdirs() {
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		echo "$c mkdir $TESTDIR/$c"
+		$PDSH $c "mkdir $TESTDIR/$c && ls -l $TESTDIR/$c"
+	done
 }
 
-trap exit INT
-
-client_mkdirs() {
-   $PDSH $CLIENTS "mkdir $MOUNT/\`hostname\`; ls $MOUNT/\`hostname\` > /dev/null"
+client_rmdirs() {
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		echo "rmdir $TESTDIR/$c"
+		$PDSH $LIVE_CLIENT "rmdir $TESTDIR/$c"
+	done
 }
 
 clients_recover_osts() {
     facet=$1
-    $PDSH $CLIENTS "$LCTL "'--device %OSC_`hostname`_OST_'"${facet}_svc_MNT_client recover"
+#    do_node $CLIENTS "$LCTL "'--device %OSC_`hostname`_'"${facet}_svc_MNT_client_facet recover"
 }
 
-if [ "$ONLY" == "cleanup" ]; then
-    cleanup
-    exit
-fi
-
-gen_config
-setup
-
-if [ "$ONLY" == "setup" ]; then
-    exit 0
-fi
+check_and_setup_lustre
 
-# 9 Different Failure Modes Combinations
-echo "Starting Test 17 at `date`"
+rm -rf $TESTDIR
+mkdir -p $TESTDIR
 
 test_0() {
-    echo "Failover MDS"
-    facet_failover mds
-    wait $DFPID || return 1
-
-    echo "Failing OST1"
-    facet_failover ost1
-    wait $DFPID || return 2
-
-    echo "Failing OST2"
-    facet_failover ost2
-    wait $DFPID || return 3
-    return 0
+	for i in $(seq $MDSCOUNT) ; do
+		fail mds$i
+	done
+
+	for i in $(seq $OSTCOUNT) ; do
+		fail ost$i
+	done
+	return 0
 }
 run_test 0 "Fail all nodes, independently"
 
 ############### First Failure Mode ###############
 test_1() {
-echo "Don't do a MDS - MDS Failure Case"
-echo "This makes no sense"
-# FIXME every test makes sense
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+
+	clients_up
+
+	shutdown_facet mds1
+	reboot_facet mds1
+
+	# prepare for MDS failover
+	change_active mds1
+	reboot_facet mds1
+
+	clients_up &
+	DFPID=$!
+	sleep 5
+
+	shutdown_facet mds2
+
+	echo "Reintegrating MDS2"
+	reboot_facet mds2
+	wait_for_facet mds2
+	start_mdt 2 || return 2
+
+	wait_for_facet mds1
+	start_mdt 1 || return $?
+
+	#Check FS
+	wait $DFPID
+	echo "Verify reintegration"
+	clients_up || return 1
 }
 run_test 1 "MDS/MDS failure"
 ###################################################
 
 ############### Second Failure Mode ###############
 test_2() {
-    echo "Verify Lustre filesystem is up and running"
-    client_df
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
 
-    echo "Failing MDS"
-    shutdown_facet mds
-    reboot_facet mds
+	for i in $(seq $MDSCOUNT) ; do
+		shutdown_facet mds$i
+		reboot_facet mds$i
 
-    # prepare for MDS failover
-    change_active mds
-    reboot_facet mds
+		# prepare for MDS failover
+		change_active mds$i
+		reboot_facet mds$i
+	done
 
-    client_df &
-    DFPID=$!
-    sleep 5
+	clients_up &
+	DFPID=$!
+	sleep 5
 
-    echo "Failing OST"
-    shutdown_facet ost1
+	shutdown_facet ost1
 
-    echo "Reintegrating OST"
-    reboot_facet ost1
-    wait_for ost1
-    start ost1
+	echo "Reintegrating OST"
+	reboot_facet ost1
+	wait_for_facet ost1
+	start_ost 1 || return 2
 
-    echo "Failover MDS"
-    wait_for mds
-    start mds
+	for i in $(seq $MDSCOUNT) ; do
+		wait_for_facet mds$i
+		start_mdt $i || return $?
+	done
 
-    #Check FS
-    wait $DFPID
-    clients_recover_osts ost1
-    echo "Verify reintegration"
-    client_df
+	#Check FS
+	wait $DFPID
+	clients_recover_osts ost1
+	echo "Verify reintegration"
+	clients_up || return 1
 
 }
 run_test 2 "Second Failure Mode: MDS/OST `date`"
 ###################################################
 
-
 ############### Third Failure Mode ###############
 test_3() {
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    
-    #MDS Portion
-    facet_failover mds
-    wait $DFPID || echo df failed: $?
-    #Check FS
-
-    echo "Test Lustre stability after MDS failover"
-    client_df
-
-    #CLIENT Portion
-    echo "Failing 2 CLIENTS"
-    fail_clients 2
-    
-    #Check FS
-    echo "Test Lustre stability after CLIENT failure"
-    client_df
-    
-    #Reintegration
-    echo "Reintegrating CLIENTS"
-    reintegrate_clients
-
-    client_df
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	#MDS Portion
+	for i in $(seq $MDSCOUNT) ; do
+		fail mds$i
+	done
+
+	#Check FS
+	echo "Test Lustre stability after MDS failover"
+	clients_up
+
+	#CLIENT Portion
+	echo "Failing 2 CLIENTS"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENT failure"
+	clients_up
+
+	#Reintegration
+	echo "Reintegrating CLIENTS"
+	reintegrate_clients || return 1
+
+	clients_up || return 3
+	sleep 2 # give it a little time for fully recovered before next test
 }
-run_test 3  "Thirdb Failure Mode: MDS/CLIENT `date`"
+run_test 3  "Third Failure Mode: MDS/CLIENT `date`"
 ###################################################
 
 ############### Fourth Failure Mode ###############
 test_4() {
-    echo "Fourth Failure Mode: OST/MDS `date`"
-
-    #OST Portion
-    echo "Failing OST ost1"
-    shutdown_facet ost1
- 
-    #Check FS
-    echo "Test Lustre stability after OST failure"
-    client_df
-
-    #MDS Portion
-    echo "Failing MDS"
-    shutdown_facet mds
-    reboot_facet mds
-
-    # prepare for MDS failover
-    change_active mds
-    reboot_facet mds
-
-    client_df &
-    DFPID=$!
-    sleep 5
-
-    #Reintegration
-    echo "Reintegrating OST"
-    reboot_facet ost1
-    wait_for ost1
-    start ost1
-    
-    echo "Failover MDS"
-    wait_for mds
-    start mds
-    #Check FS
-    
-    wait $DFPID
-    clients_recover_osts ost1
-    echo "Test Lustre stability after MDS failover"
-    client_df
+	echo "Fourth Failure Mode: OST/MDS `date`"
+
+	#OST Portion
+	shutdown_facet ost1
+
+	#Check FS
+	echo "Test Lustre stability after OST failure"
+	clients_up &
+	DFPIDA=$!
+	sleep 5
+
+	for i in $(seq $MDSCOUNT) ; do
+		shutdown_facet mds$i
+		reboot_facet mds$i
+
+		# prepare for MDS failover
+		change_active mds$i
+		reboot_facet mds$i
+	done
+
+	clients_up &
+	DFPIDB=$!
+	sleep 5
+
+	#Reintegration
+	echo "Reintegrating OST"
+	reboot_facet ost1
+	wait_for_facet ost1
+	start_ost 1
+
+	for i in $(seq $MDSCOUNT) ; do
+		wait_for_facet mds$i
+		start_mdt $i || return $?
+	done
+
+	#Check FS
+	wait $DFPIDA
+	wait $DFPIDB
+	clients_recover_osts ost1
+	echo "Test Lustre stability after MDS failover"
+	clients_up || return 1
 }
 run_test 4 "Fourth Failure Mode: OST/MDS `date`"
 ###################################################
 
 ############### Fifth Failure Mode ###############
 test_5() {
-    echo "Fifth Failure Mode: OST/OST `date`"
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    
-    #OST Portion
-    echo "Failing OST"
-    shutdown_facet ost1
-    reboot_facet ost1
-    
-    #Check FS
-    echo "Test Lustre stability after OST failure"
-    client_df
-    
-    #OST Portion
-    echo "Failing OST"
-    shutdown_node ost2
-    reboot_facet ost2
-
-    #Check FS
-    echo "Test Lustre stability after OST failure"
-    client_df
-
-    #Reintegration
-    echo "Reintegrating OSTs"
-    wait_for ost1
-    wait_for ost1
-    start ost1
-    start ost2
-    
-    clients_recover_osts ost1
-    clients_recover_osts ost2
-    client_df
+	[ $OSTCOUNT -lt 2 ] && skip_env "needs >= 2 OSTs"
+
+	echo "Fifth Failure Mode: OST/OST `date`"
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	#OST Portion
+	shutdown_facet ost1
+	reboot_facet ost1
+
+	#Check FS
+	echo "Test Lustre stability after OST failure"
+	clients_up &
+	DFPIDA=$!
+	sleep 5
+
+	#OST Portion
+	shutdown_facet ost2
+	reboot_facet ost2
+
+	#Check FS
+	echo "Test Lustre stability after OST failure"
+	clients_up &
+	DFPIDB=$!
+	sleep 5
+
+	#Reintegration
+	echo "Reintegrating OSTs"
+	wait_for_facet ost1
+	start_ost 1
+	wait_for_facet ost2
+	start_ost 2
+
+	clients_recover_osts ost1
+	clients_recover_osts ost2
+	sleep $TIMEOUT
+
+	wait $DFPIDA
+	wait $DFPIDB
+	clients_up || return 2
 }
 run_test 5 "Fifth Failure Mode: OST/OST `date`"
 ###################################################
 
 ############### Sixth Failure Mode ###############
 test_6() {
-    echo "Sixth Failure Mode: OST/CLIENT `date`"
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-	
-    #OST Portion
-    echo "Failing OST"
-    shutdown_node ost1
-    reboot_facet ost1
-
-    #Check FS
-    echo "Test Lustre stability after OST failure"
-    client_df
-
-    #CLIENT Portion
-    echo "Failing CLIENTs"
-    fail_clients
-    
-    #Check FS
-    echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    
-    #Reintegration
-    echo "Reintegrating OST/CLIENTs"
-    wait_for ost1
-    start ost1
-    reintegrate_clients
-    
-    echo "Verifying mount"
-    client_df
+	echo "Sixth Failure Mode: OST/CLIENT `date`"
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	client_touch testfile || return 2
+
+	#OST Portion
+	shutdown_facet ost1
+	reboot_facet ost1
+
+	#Check FS
+	echo "Test Lustre stability after OST failure"
+	clients_up &
+	DFPIDA=$!
+	echo DFPIDA=$DFPIDA
+	sleep 5
+
+	#CLIENT Portion
+	echo "Failing CLIENTs"
+	fail_clients
+
+	#Check FS
+	echo "Test Lustre stability after CLIENTs failure"
+	clients_up &
+	DFPIDB=$!
+	echo DFPIDB=$DFPIDB
+	sleep 5
+
+	#Reintegration
+	echo "Reintegrating OST/CLIENTs"
+	wait_for_facet ost1
+	start_ost 1
+	reintegrate_clients || return 1
+	sleep 5
+
+	wait_remote_prog "stat -f" $((TIMEOUT * 3 + 20))
+	wait $DFPIDA
+	wait $DFPIDB
+
+	echo "Verifying mount"
+	[ -z "$(mounted_lustre_filesystems)" ] && return 3
+	clients_up
 }
 run_test 6 "Sixth Failure Mode: OST/CLIENT `date`"
 ###################################################
 
-
 ############### Seventh Failure Mode ###############
 test_7() {
-    echo "Seventh Failure Mode: CLIENT/MDS `date`"
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-
-    #CLIENT Portion
-    echo "Part 1: Failing CLIENT"
-    fail_clients 2
-    
-    #Check FS
-    echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-    
-    #Sleep
-    echo "Wait 1 minutes"
-    sleep 60
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-
-    #MDS Portion
-    echo "Failing MDS"
-    facet_failover mds
-
-    #Check FS
-    echo "Test Lustre stability after MDS failover"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-
-    #Reintegration
-    echo "Reintegrating CLIENTs"
-    reintegrate_clients
-    client_df
-    
-    #Sleep
-    echo "wait 1 minutes"
-    sleep 60
+	echo "Seventh Failure Mode: CLIENT/MDS `date`"
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	client_touch testfile  || return 1
+
+	#CLIENT Portion
+	echo "Part 1: Failing CLIENT"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENTs failure"
+	clients_up
+	$PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+	$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
+
+	#Sleep
+	echo "Wait 1 minutes"
+	sleep 60
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && return 2
+
+	clients_up
+	client_rm testfile
+
+	#MDS Portion
+	for i in $(seq $MDSCOUNT) ; do
+		fail mds$i
+	done
+
+	$PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+	$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
+
+	#Reintegration
+	echo "Reintegrating CLIENTs"
+	reintegrate_clients || return 2
+	clients_up
+
+	#Sleep
+	echo "wait 1 minutes"
+	sleep 60
 }
 run_test 7 "Seventh Failure Mode: CLIENT/MDS `date`"
 ###################################################
 
-
 ############### Eighth Failure Mode ###############
 test_8() {
-    echo "Eighth Failure Mode: CLIENT/OST `date`"
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-	
-    #CLIENT Portion
-    echo "Failing CLIENTs"
-    fail_clients 2
-
-    #Check FS
-    echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-
-    #Sleep
-    echo "Wait 1 minutes"
-    sleep 60
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-
-    #OST Portion
-    echo "Failing OST"
-    shutdown_node ost1
-    reboot_facet ost1
-
-    #Check FS
-    echo "Test Lustre stability after OST failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-    
-    #Reintegration
-    echo "Reintegrating CLIENTs/OST"
-    reintegrate_clients
-    start ost1
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/CLIENT_OST_2\`hostname\`_testfile"
-
-    #Sleep
-    echo "Wait 1 minutes"
-    sleep 60
+	echo "Eighth Failure Mode: CLIENT/OST `date`"
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	client_touch testfile
+
+	#CLIENT Portion
+	echo "Failing CLIENTs"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENTs failure"
+	clients_up
+	$PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+	$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
+
+	#Sleep
+	echo "Wait 1 minutes"
+	sleep 60
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	client_touch testfile
+
+	#OST Portion
+	shutdown_facet ost1
+	reboot_facet ost1
+
+	#Check FS
+	echo "Test Lustre stability after OST failure"
+	clients_up &
+	DFPID=$!
+	sleep 5
+	#non-failout hangs forever here
+	#$PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+	#$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
+
+	#Reintegration
+	echo "Reintegrating CLIENTs/OST"
+	reintegrate_clients || return 3
+	wait_for_facet ost1
+	start_ost 1
+	wait $DFPID
+	clients_up || return 1
+	client_touch testfile2 || return 2
+
+	#Sleep
+	echo "Wait 1 minutes"
+	sleep 60
 }
 run_test 8 "Eighth Failure Mode: CLIENT/OST `date`"
 ###################################################
 
-
 ############### Ninth Failure Mode ###############
 test_9() {
-    echo 
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-	
-    #CLIENT Portion
-    echo "Failing CLIENTs"
-    fail_clients 2
-
-    #Check FS
-    echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-
-    #Sleep
-    echo "Wait 1 minutes"
-    sleep 60
-
-    #Create files
-    echo "Verify Lustre filesystem is up and running"
-    client_df
-    $PDSH $CLIENTS "/bin/touch $MOUNT/\`hostname\`_testfile"
-
-    #CLIENT Portion
-    echo "Failing CLIENTs"
-    fail_clients 2
-    
-    #Check FS
-    echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
-
-    #Reintegration
-    echo "Reintegrating  CLIENTs/CLIENTs"
-    reintegrate_clients
-    client_df
-    
-    #Sleep
-    echo "Wait 1 minutes"
-    sleep 60
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	clients_up
+	client_touch testfile || return 1
+
+	#CLIENT Portion
+	echo "Failing CLIENTs"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENTs failure"
+	clients_up
+	$PDSH $LIVE_CLIENT "ls -l $TESTDIR" || return 1
+	$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile" || return 2
+
+	#Sleep
+	echo "Wait 1 minutes"
+	sleep 60
+
+	#Create files
+	echo "Verify Lustre filesystem is up and running"
+	client_up $LIVE_CLIENT || return 3
+	client_touch testfile || return 4
+
+	#CLIENT Portion
+	echo "Failing CLIENTs"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENTs failure"
+	clients_up
+	$PDSH $LIVE_CLIENT "ls -l $TESTDIR" || return 5
+	$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile" || return 6
+
+	#Reintegration
+	echo "Reintegrating  CLIENTs/CLIENTs"
+	reintegrate_clients || return 7
+	clients_up
+
+	#Sleep
+	echo "Wait 1 minutes"
+	sleep 60
 }
 run_test 9 "Ninth Failure Mode: CLIENT/CLIENT `date`"
 ###################################################
 
+############### Tenth Failure Mode ###############
 test_10() {
-    #Run availability after all failures
-    ./availability.sh  21600
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+
+	shutdown_facet mds1
+	reboot_facet mds1
+
+	# prepare for MDS failover
+	change_active mds1
+	reboot_facet mds1
+
+	clients_up &
+	DFPID=$!
+	sleep 5
+
+	shutdown_facet ost1
+
+	echo "Reintegrating OST"
+	reboot_facet ost1
+	wait_for_facet ost1
+	start_ost 1 || return 2
+
+	shutdown_facet mds2
+	reboot_facet mds2
+
+	# prepare for MDS failover
+	change_active mds2
+	reboot_facet mds2
+
+	wait_for_facet mds1
+	start_mdt 1 || return $?
+
+	wait_for_facet mds2
+	start_mdt 2 || return $?
+
+	#Check FS
+	wait $DFPID
+	clients_recover_osts ost1
+	echo "Verify reintegration"
+	clients_up || return 1
+}
+run_test 10 "Tenth Failure Mode: MDT0/OST/MDT1 `date`"
+###################################################
+
+############### Seventh Failure Mode ###############
+test_11() {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	#MDS Portion
+	fail mds1
+	#Check FS
+
+	echo "Test Lustre stability after MDS failover"
+	clients_up
+
+	#CLIENT Portion
+	echo "Failing 2 CLIENTS"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENT failure"
+	clients_up
+
+	#Reintegration
+	echo "Reintegrating CLIENTS"
+	reintegrate_clients || return 1
+
+	fail mds2
+
+	clients_up || return 3
+	sleep 2 # give it a little time for fully recovered before next test
+}
+run_test 11 "Eleventh Failure Mode: MDS0/CLIENT/MDS1 `date`"
+###################################################
+
+test_12() {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	#MDS Portion
+	fail mds1,mds2
+	clients_up
+
+	#OSS Portion
+	fail ost1,ost2
+	clients_up
+
+	#CLIENT Portion
+	echo "Failing 2 CLIENTS"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENT failure"
+	clients_up
+
+	#Reintegration
+	echo "Reintegrating CLIENTS"
+	reintegrate_clients || return 1
+
+	clients_up || return 3
+	sleep 2 # give it a little time for fully recovered before next test
+}
+run_test 12 "Twelve Failure Mode: MDS0,MDS1/OST0, OST1/CLIENTS `date`"
+###################################################
+
+test_13() {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	#MDS Portion
+	fail mds1,mds2
+	clients_up
+
+	#CLIENT Portion
+	echo "Failing 2 CLIENTS"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENT failure"
+	clients_up
+
+	#Reintegration
+	echo "Reintegrating CLIENTS"
+	reintegrate_clients || return 1
+
+	clients_up || return 3
+	sleep 2 # give it a little time for fully recovered before next test
+
+	#OSS Portion
+	fail ost1,ost2
+	clients_up || return 4
+}
+run_test 13 "Thirteen Failure Mode: MDS0,MDS1/CLIENTS/OST0,OST1 `date`"
+###################################################
+
+test_14() {
+	[ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs"
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+
+	#OST Portion
+	fail ost1,ost2
+	clients_up
+
+	#CLIENT Portion
+	echo "Failing 2 CLIENTS"
+	fail_clients 2
+
+	#Check FS
+	echo "Test Lustre stability after CLIENT failure"
+	clients_up
+
+	#Reintegration
+	echo "Reintegrating CLIENTS"
+	reintegrate_clients || return 1
+
+	clients_up || return 3
+	sleep 2 # give it a little time for fully recovered before next test
+
+	#OSS Portion
+	fail mds1,mds2
+	clients_up || return 4
 }
-run_test 10 "Running Availability for 6 hours..."
+run_test 14 "Fourteen Failure Mode: OST0,OST1/CLIENTS/MDS0,MDS1 `date`"
 
-equals_msg "Done, cleaning up"
-cleanup
+complete $SECONDS
+check_and_cleanup_lustre
+exit_status