X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Ftests%2Finsanity.sh;h=74ca4a24c360eeefc26f8891ea4e9788ecf1a2b5;hb=46021d556b67557d545e01522c41b9fad20120e1;hp=4d4f7703a704730c82461e11d0e528000ba530c3;hpb=fa8df3b66d6f4e13d0b6076bb087a52b20d4166c;p=fs%2Flustre-release.git

diff --git a/lustre/tests/insanity.sh b/lustre/tests/insanity.sh
index 4d4f770..74ca4a2 100755
--- a/lustre/tests/insanity.sh
+++ b/lustre/tests/insanity.sh
@@ -1,4 +1,7 @@
-#!/bin/sh
+#!/bin/bash
+# -*- mode: Bash; tab-width: 4; indent-tabs-mode: t; -*-
+# vim:shiftwidth=4:softtabstop=4:tabstop=4:
+#
 # Test multiple failures, AKA Test 17
 
 set -e
@@ -9,17 +12,16 @@ LUSTRE=${LUSTRE:-`dirname $0`/..}
 init_test_env $@
 
 . ${CONFIG:=$LUSTRE/tests/cfg/$NAME.sh}
+init_logging
 #              
 ALWAYS_EXCEPT="10  $INSANITY_EXCEPT"
 
 if [ "$FAILURE_MODE" = "HARD" ]; then
-    mixed_ost_devs && CONFIG_EXCEPTIONS="0 2 4 5 6 8" && \
-        echo -n "Several ost services on one ost node are used with FAILURE_MODE=$FAILURE_MODE. " && \
-        echo "Except the tests: $CONFIG_EXCEPTIONS" && \
-        ALWAYS_EXCEPT="$ALWAYS_EXCEPT $CONFIG_EXCEPTIONS"
+	skip_env "$TESTSUITE: is not functional with FAILURE_MODE = HARD, " \
+		"please use recovery-double-scale, bz20407"
+	exit 0
 fi
 
-#
 [ "$SLOW" = "no" ] && EXCEPT_SLOW=""
 
 SETUP=${SETUP:-""}
@@ -31,17 +33,18 @@ SINGLECLIENT=${SINGLECLIENT:-$HOSTNAME}
 LIVE_CLIENT=${LIVE_CLIENT:-$SINGLECLIENT}
 FAIL_CLIENTS=${FAIL_CLIENTS:-$RCLIENTS}
 
-assert_env mds_HOST MDS_MKFS_OPTS
-assert_env ost_HOST OST_MKFS_OPTS OSTCOUNT
+assert_env mds_HOST
+assert_env ost_HOST OSTCOUNT
 assert_env LIVE_CLIENT FSNAME
 
-remote_mds_nodsh && skip "remote MDS with nodsh" && exit 0
-remote_ost_nodsh && skip "remote OST with nodsh" && exit 0
+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
@@ -61,12 +64,12 @@ set_fail_client() {
 }
 
 fail_clients() {
-    num=$1
+	num=$1
 
-    log "Request clients to fail: ${num}. Num of clients to fail: ${FAIL_NUM}, already failed: $DOWN_NUM"
-    if [ -z "$num"  ] || [ "$num" -gt $((FAIL_NUM - DOWN_NUM)) ]; then
-	num=$((FAIL_NUM - DOWN_NUM)) 
-    fi
+	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!"
@@ -84,57 +87,57 @@ fail_clients() {
 
     echo "down clients: $DOWN_CLIENTS"
 
-    for client in $DOWN_CLIENTS; do
-	boot_node $client
-    done
-    DOWN_NUM=`echo $DOWN_CLIENTS | wc -w`
-    client_rmdirs
+	for client in $DOWN_CLIENTS; do
+		boot_node $client
+	done
+	DOWN_NUM=`echo $DOWN_CLIENTS | wc -w`
+	client_rmdirs
 }
 
 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
+	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
 }
 
 start_ost() {
-    start ost$1 `ostdevname $1` $OST_MOUNT_OPTS
+	start ost$1 `ostdevname $1` $OST_MOUNT_OPTS
 }
 
 trap exit INT
 
 client_touch() {
-    file=$1
-    for c in $LIVE_CLIENT $FAIL_CLIENTS;  do
-	if echo $DOWN_CLIENTS | grep -q $c; then continue; fi
-	$PDSH $c touch $MOUNT/${c}_$file || return 1
-    done
+	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
 }
 
 client_rm() {
-    file=$1
-    for c in $LIVE_CLIENT $FAIL_CLIENTS;  do
-	$PDSH $c rm $MOUNT/${c}_$file
-    done
+	file=$1
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		$PDSH $c rm $TESTDIR/${c}_$file
+	done
 }
 
 client_mkdirs() {
-    for c in $LIVE_CLIENT $FAIL_CLIENTS;  do
-	echo "$c mkdir $MOUNT/$c"
-	$PDSH $c "mkdir $MOUNT/$c"
-	$PDSH $c "ls -l $MOUNT/$c" 
-    done
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		echo "$c mkdir $TESTDIR/$c"
+		$PDSH $c "mkdir $TESTDIR/$c && ls -l $TESTDIR/$c"
+	done
 }
 
 client_rmdirs() {
-    for c in $LIVE_CLIENT $FAIL_CLIENTS;  do
-	echo "rmdir $MOUNT/$c"
-	$PDSH $LIVE_CLIENT "rmdir $MOUNT/$c"
-    done
+	for c in $LIVE_CLIENT $FAIL_CLIENTS; do
+		echo "rmdir $TESTDIR/$c"
+		$PDSH $LIVE_CLIENT "rmdir $TESTDIR/$c"
+	done
 }
 
 clients_recover_osts() {
@@ -144,6 +147,9 @@ clients_recover_osts() {
 
 check_and_setup_lustre
 
+rm -rf $TESTDIR
+mkdir -p $TESTDIR
+
 # 9 Different Failure Modes Combinations
 echo "Starting Test 17 at `date`"
 
@@ -167,10 +173,13 @@ run_test 1 "MDS/MDS failure"
 
 ############### Second Failure Mode ###############
 test_2() {
-    echo "Verify Lustre filesystem is up and running"
-    [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
+	echo "Verify Lustre filesystem is up and running"
+	[ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+	[ "$(facet_fstype ost1)" = "zfs" ] &&
+		skip "LU-2059: no local config for ZFS OSTs" && return
+
+    clients_up
 
     shutdown_facet $SINGLEMDS
     reboot_facet $SINGLEMDS
@@ -179,7 +188,7 @@ test_2() {
     change_active $SINGLEMDS
     reboot_facet $SINGLEMDS
 
-    client_df &
+    clients_up &
     DFPID=$!
     sleep 5
 
@@ -187,17 +196,17 @@ test_2() {
 
     echo "Reintegrating OST"
     reboot_facet ost1
-    wait_for ost1
+    wait_for_facet ost1
     start_ost 1 || return 2
 
-    wait_for $SINGLEMDS
+    wait_for_facet $SINGLEMDS
     start $SINGLEMDS `mdsdevname 1` $MDS_MOUNT_OPTS || return $?
 
     #Check FS
     wait $DFPID
     clients_recover_osts ost1
     echo "Verify reintegration"
-    client_df || return 1
+    clients_up || return 1
 
 }
 run_test 2 "Second Failure Mode: MDS/OST `date`"
@@ -215,7 +224,7 @@ test_3() {
     #Check FS
 
     echo "Test Lustre stability after MDS failover"
-    client_df
+    clients_up
 
     #CLIENT Portion
     echo "Failing 2 CLIENTS"
@@ -223,13 +232,13 @@ test_3() {
     
     #Check FS
     echo "Test Lustre stability after CLIENT failure"
-    client_df
+    clients_up
     
     #Reintegration
     echo "Reintegrating CLIENTS"
     reintegrate_clients || return 1
 
-    client_df || return 3
+    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`"
@@ -237,14 +246,17 @@ run_test 3  "Thirdb Failure Mode: MDS/CLIENT `date`"
 
 ############### Fourth Failure Mode ###############
 test_4() {
-    echo "Fourth Failure Mode: OST/MDS `date`"
+	echo "Fourth Failure Mode: OST/MDS `date`"
+
+	[ "$(facet_fstype ost1)" = "zfs" ] &&
+		skip "LU-2059: no local config for ZFS OSTs" && return
 
     #OST Portion
     shutdown_facet ost1
  
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     sleep 5
 
@@ -256,17 +268,17 @@ test_4() {
     change_active $SINGLEMDS
     reboot_facet $SINGLEMDS
 
-    client_df &
+    clients_up &
     DFPIDB=$!
     sleep 5
 
     #Reintegration
     echo "Reintegrating OST"
     reboot_facet ost1
-    wait_for ost1
+    wait_for_facet ost1
     start_ost 1
-    
-    wait_for $SINGLEMDS
+
+    wait_for_facet $SINGLEMDS
     start $SINGLEMDS `mdsdevname 1` $MDS_MOUNT_OPTS
     #Check FS
     
@@ -274,14 +286,14 @@ test_4() {
     wait $DFPIDB
     clients_recover_osts ost1
     echo "Test Lustre stability after MDS failover"
-    client_df || return 1
+    clients_up || return 1
 }
 run_test 4 "Fourth Failure Mode: OST/MDS `date`"
 ###################################################
 
 ############### Fifth Failure Mode ###############
 test_5() {
-    [ $OSTCOUNT -lt 2 ] && skip "$OSTCOUNT < 2, not enough OSTs" && return 0
+    [ $OSTCOUNT -lt 2 ] && skip_env "$OSTCOUNT < 2, not enough OSTs" && return 0
 
     echo "Fifth Failure Mode: OST/OST `date`"
 
@@ -289,7 +301,7 @@ test_5() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     
     #OST Portion
     shutdown_facet ost1
@@ -297,7 +309,7 @@ test_5() {
     
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     sleep 5
     
@@ -307,15 +319,15 @@ test_5() {
 
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDB=$!
     sleep 5
 
     #Reintegration
     echo "Reintegrating OSTs"
-    wait_for ost1
+    wait_for_facet ost1
     start_ost 1
-    wait_for ost2
+    wait_for_facet ost2
     start_ost 2
     
     clients_recover_osts ost1
@@ -324,7 +336,7 @@ test_5() {
 
     wait $DFPIDA
     wait $DFPIDB
-    client_df || return 2
+    clients_up || return 2
 }
 run_test 5 "Fifth Failure Mode: OST/OST `date`"
 ###################################################
@@ -337,7 +349,7 @@ test_6() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile || return 2
 	
     #OST Portion
@@ -346,7 +358,7 @@ test_6() {
 
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     echo DFPIDA=$DFPIDA
     sleep 5
@@ -357,25 +369,25 @@ test_6() {
     
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df &
+    clients_up &
     DFPIDB=$!
     echo DFPIDB=$DFPIDB
     sleep 5
     
     #Reintegration
     echo "Reintegrating OST/CLIENTs"
-    wait_for ost1
+    wait_for_facet ost1
     start_ost 1
     reintegrate_clients || return 1
     sleep 5 
 
-    wait_remote_prog df $((TIMEOUT * 3 + 10)) 
+    wait_remote_prog "stat -f" $((TIMEOUT * 3 + 20)) 
     wait $DFPIDA
     wait $DFPIDB
 
     echo "Verifying mount"
     [ -z "$(mounted_lustre_filesystems)" ] && return 3
-    client_df
+    clients_up
 }
 run_test 6 "Sixth Failure Mode: OST/CLIENT `date`"
 ###################################################
@@ -389,7 +401,7 @@ test_7() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile  || return 1
 
     #CLIENT Portion
@@ -398,9 +410,9 @@ test_7() {
     
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
+    clients_up
+    $PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+    $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
     
     #Sleep
     echo "Wait 1 minutes"
@@ -410,19 +422,19 @@ test_7() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && return 2
 
-    client_df
+    clients_up
     client_rm testfile
 
     #MDS Portion
     fail $SINGLEMDS
 
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
+    $PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+    $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
 
     #Reintegration
     echo "Reintegrating CLIENTs"
     reintegrate_clients || return 2
-    client_df
+    clients_up
     
     #Sleep
     echo "wait 1 minutes"
@@ -440,7 +452,7 @@ test_8() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile
 	
     #CLIENT Portion
@@ -449,9 +461,9 @@ test_8() {
 
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
+    clients_up
+    $PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+    $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
 
     #Sleep
     echo "Wait 1 minutes"
@@ -461,7 +473,7 @@ test_8() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile
 
 
@@ -471,20 +483,20 @@ test_8() {
 
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPID=$!
     sleep 5
     #non-failout hangs forever here
-    #$PDSH $LIVE_CLIENT "ls -l $MOUNT"
-    #$PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile"
+    #$PDSH $LIVE_CLIENT "ls -l $TESTDIR"
+    #$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
     
     #Reintegration
     echo "Reintegrating CLIENTs/OST"
     reintegrate_clients || return 3
-    wait_for ost1
+    wait_for_facet ost1
     start_ost 1
     wait $DFPID
-    client_df || return 1
+    clients_up || return 1
     client_touch testfile2 || return 2
 
     #Sleep
@@ -503,7 +515,7 @@ test_9() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile || return 1
 	
     #CLIENT Portion
@@ -512,9 +524,9 @@ test_9() {
 
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT" || return 1
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile" || return 2
+    clients_up
+    $PDSH $LIVE_CLIENT "ls -l $TESTDIR" || return 1
+    $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile" || return 2
 
     #Sleep
     echo "Wait 1 minutes"
@@ -522,7 +534,7 @@ test_9() {
 
     #Create files
     echo "Verify Lustre filesystem is up and running"
-    $PDSH $LIVE_CLIENT df $MOUNT || return 3
+    client_up $LIVE_CLIENT || return 3
     client_touch testfile || return 4
 
     #CLIENT Portion
@@ -531,14 +543,14 @@ test_9() {
     
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
-    $PDSH $LIVE_CLIENT "ls -l $MOUNT" || return 5
-    $PDSH $LIVE_CLIENT "rm -f $MOUNT/*_testfile" || return 6
+    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
-    client_df
+    clients_up
     
     #Sleep
     echo "Wait 1 minutes"
@@ -555,6 +567,6 @@ test_10() {
 }
 run_test 10 "Running Availability for 6 hours..."
 
-equals_msg `basename $0`: test complete, cleaning up
+complete $SECONDS
 check_and_cleanup_lustre
-[ -f "$TESTSUITELOG" ] && cat $TESTSUITELOG && grep -q FAIL $TESTSUITELOG && exit 1 || true
+exit_status