Branch b1_8

b=19023
i=rread
i=johann

don't use df to determine recovery result

diff --git a/lustre/tests/conf-sanity.sh b/lustre/tests/conf-sanity.sh
index 7dd16d3..510b63a 100644 (file)
--- a/lustre/tests/conf-sanity.sh
+++ b/lustre/tests/conf-sanity.sh
@@ -1660,7 +1660,7 @@ test_47() { #17674
 
         facet_failover ost1
         facet_failover mds
-        df -h $MOUNT || return 3
+        client_up || return 3
 
         count=0
         for ns in $($LCTL get_param ldlm.namespaces.$FSNAME-*-*-*.lru_size); do

diff --git a/lustre/tests/insanity.sh b/lustre/tests/insanity.sh
index eb5dd79..d1345f1 100755 (executable)
--- a/lustre/tests/insanity.sh
+++ b/lustre/tests/insanity.sh
@@ -177,7 +177,7 @@ test_2() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
 
     shutdown_facet mds
     reboot_facet mds
@@ -186,7 +186,7 @@ test_2() {
     change_active mds
     reboot_facet mds
 
-    client_df &
+    clients_up &
     DFPID=$!
     sleep 5
 
@@ -204,7 +204,7 @@ test_2() {
     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`"
@@ -216,28 +216,28 @@ test_3() {
     #Create files
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
-    
+
     #MDS Portion
     facet_failover mds
     wait $DFPID || echo df failed: $?
     #Check FS
 
     echo "Test Lustre stability after MDS failover"
-    client_df
+    clients_up
 
     #CLIENT Portion
     echo "Failing 2 CLIENTS"
     fail_clients 2
-    
+
     #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
 }
 run_test 3  "Thirdb Failure Mode: MDS/CLIENT `date`"
 ###################################################
@@ -248,10 +248,10 @@ test_4() {
 
     #OST Portion
     shutdown_facet ost1
- 
+
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     sleep 5
 
@@ -263,7 +263,7 @@ test_4() {
     change_active mds
     reboot_facet mds
 
-    client_df &
+    clients_up &
     DFPIDB=$!
     sleep 5
 
@@ -272,16 +272,16 @@ test_4() {
     reboot_facet ost1
     wait_for ost1
     start_ost 1
-    
+
     wait_for mds
     start mds $MDSDEV $MDS_MOUNT_OPTS
     #Check FS
-    
+
     wait $DFPIDA
     wait $DFPIDB
     clients_recover_osts ost1
     echo "Test Lustre stability after MDS failover"
-    client_df || return 1
+    clients_up || return 1
 }
 run_test 4 "Fourth Failure Mode: OST/MDS `date`"
 ###################################################
@@ -296,25 +296,25 @@ test_5() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
-    
+    clients_up
+
     #OST Portion
     shutdown_facet ost1
     reboot_facet ost1
-    
+
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     sleep 5
-    
+
     #OST Portion
     shutdown_facet ost2
     reboot_facet ost2
 
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDB=$!
     sleep 5
 
@@ -324,14 +324,14 @@ test_5() {
     start_ost 1
     wait_for ost2
     start_ost 2
-    
+
     clients_recover_osts ost1
     clients_recover_osts ost2
     sleep $TIMEOUT
 
     wait $DFPIDA
     wait $DFPIDB
-    client_df || return 2
+    clients_up || return 2
 }
 run_test 5 "Fifth Failure Mode: OST/OST `date`"
 ###################################################
@@ -344,16 +344,16 @@ test_6() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile || return 2
-       
+
     #OST Portion
     shutdown_facet ost1
     reboot_facet ost1
 
     #Check FS
     echo "Test Lustre stability after OST failure"
-    client_df &
+    clients_up &
     DFPIDA=$!
     echo DFPIDA=$DFPIDA
     sleep 5
@@ -361,28 +361,28 @@ test_6() {
     #CLIENT Portion
     echo "Failing CLIENTs"
     fail_clients
-    
+
     #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
     start_ost 1
     reintegrate_clients || return 1
-    sleep 5 
+    sleep 5
 
-    wait_remote_prog df $((TIMEOUT * 3 + 10)) 
+    wait_remote_prog df $((TIMEOUT * 3 + 10))
     wait $DFPIDA
     wait $DFPIDB
 
     echo "Verifying mount"
     [ -z "$(mounted_lustre_filesystems)" ] && return 3
-    client_df
+    clients_up
 }
 run_test 6 "Sixth Failure Mode: OST/CLIENT `date`"
 ###################################################
@@ -396,19 +396,19 @@ test_7() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && error "Lustre is not running"
 
-    client_df
+    clients_up
     client_touch testfile  || return 1
 
     #CLIENT Portion
     echo "Part 1: Failing CLIENT"
     fail_clients 2
-    
+
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
+    clients_up
     $PDSH $LIVE_CLIENT "ls -l $TESTDIR"
     $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
-    
+
     #Sleep
     echo "Wait 1 minutes"
     sleep 60
@@ -417,7 +417,7 @@ test_7() {
     echo "Verify Lustre filesystem is up and running"
     [ -z "$(mounted_lustre_filesystems)" ] && return 2
 
-    client_df
+    clients_up
     client_rm testfile
 
     #MDS Portion
@@ -432,8 +432,8 @@ test_7() {
     #Reintegration
     echo "Reintegrating CLIENTs"
     reintegrate_clients || return 2
-    client_df
-    
+    clients_up
+
     #Sleep
     echo "wait 1 minutes"
     sleep 60
@@ -450,16 +450,16 @@ 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
     echo "Failing CLIENTs"
     fail_clients 2
 
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
+    clients_up
     $PDSH $LIVE_CLIENT "ls -l $TESTDIR"
     $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
 
@@ -471,7 +471,7 @@ 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
 
 
@@ -481,20 +481,20 @@ 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 $TESTDIR"
     #$PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile"
-    
+
     #Reintegration
     echo "Reintegrating CLIENTs/OST"
     reintegrate_clients || return 3
     wait_for ost1
     start_ost 1
     wait $DFPID
-    client_df || return 1
+    clients_up || return 1
     client_touch testfile2 || return 2
 
     #Sleep
@@ -513,16 +513,16 @@ 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
     echo "Failing CLIENTs"
     fail_clients 2
 
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
+    clients_up
     $PDSH $LIVE_CLIENT "ls -l $TESTDIR" || return 1
     $PDSH $LIVE_CLIENT "rm -f $TESTDIR/*_testfile" || return 2
 
@@ -533,24 +533,24 @@ test_9() {
     #Create files
     echo "Verify Lustre filesystem is up and running"
     $PDSH $LIVE_CLIENT "grep -e $MOUNT /proc/mounts" || return 3
-    $PDSH $LIVE_CLIENT df $MOUNT
+    client_up $LIVE_CLIENT
     client_touch testfile || return 4
 
     #CLIENT Portion
     echo "Failing CLIENTs"
     fail_clients 2
-    
+
     #Check FS
     echo "Test Lustre stability after CLIENTs failure"
-    client_df
+    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"
     sleep 60

diff --git a/lustre/tests/large-scale.sh b/lustre/tests/large-scale.sh
index 5acb144..65f5d29 100644 (file)
--- a/lustre/tests/large-scale.sh
+++ b/lustre/tests/large-scale.sh
@@ -90,7 +90,7 @@ run_test 1b "VBR: connect $CLIENTCOUNT clients with delayed exports"
 # fail fn does not do df on all clients
 fail_mds () {
     facet_failover mds
-    client_df
+    clients_up
 }
 
 test_1c() {

diff --git a/lustre/tests/ost-pools.sh b/lustre/tests/ost-pools.sh
index 6bfb3f6..d646123 100644 (file)
--- a/lustre/tests/ost-pools.sh
+++ b/lustre/tests/ost-pools.sh
@@ -1323,7 +1323,7 @@ test_25() {
                stop $SINGLEMDS || return 1
                start $SINGLEMDS $MDSDEV $MDS_MOUNT_OPTS  || \
                        { error "Failed to start $SINGLEMDS after stopping" && break; }
-               client_df
+               clients_up
 
         # Veriy that the pool got created and is usable
                echo "Creating a file in pool$i"

diff --git a/lustre/tests/recovery-small.sh b/lustre/tests/recovery-small.sh
index 4a8a6f8..4a97037 100755 (executable)
--- a/lustre/tests/recovery-small.sh
+++ b/lustre/tests/recovery-small.sh
@@ -303,7 +303,7 @@ test_18b() {
     df $MOUNT > /dev/null 2>&1
     sleep 2
     # my understanding is that there should be nothing in the page
-    # cache after the client reconnects?     
+    # cache after the client reconnects?
     rc=0
     pgcache_empty || rc=2
     rm -f $f
@@ -339,7 +339,7 @@ test_18c() {
     df $MOUNT > /dev/null 2>&1
     sleep 2
     # my understanding is that there should be nothing in the page
-    # cache after the client reconnects?     
+    # cache after the client reconnects?
     rc=0
     pgcache_empty || rc=2
     rm -f $f
@@ -623,7 +623,7 @@ run_test 21h "drop open request and close reply while close and open are both in
 test_22() {
     f1=$DIR/${tfile}-1
     f2=$DIR/${tfile}-2
-    
+
     do_facet mds "lctl set_param fail_loc=0x80000115"
     multiop $f2 Oc &
     close_pid=$!
@@ -719,7 +719,7 @@ test_26b() {      # bug 10140 - evict dead exports by pinger
        fi
 
        check_timeout || return 1
-       client_df
+       clients_up
        zconf_mount `hostname` $MOUNT2 ||
                 { error "Failed to mount $MOUNT2"; return 2; }
        sleep 1 # wait connections being established
@@ -781,9 +781,9 @@ test_28() {      # bug 6086 - error adding new clients
        #define OBD_FAIL_MDS_ADD_CLIENT 0x12f
        do_facet mds lctl set_param fail_loc=0x8000012f
        # fail once (evicted), reconnect fail (fail_loc), ok
-       df || (sleep 1; df) || (sleep 1; df) || error "reconnect failed"
+       client_up || client_up || client_up || error "reconnect failed"
        rm -f $DIR/$tfile
-       fail mds                # verify MDS last_rcvd can be loaded
+       fail mds # verify MDS last_rcvd can be loaded
 }
 run_test 28 "handle error adding new clients (bug 6086)"
 

diff --git a/lustre/tests/replay-dual.sh b/lustre/tests/replay-dual.sh
index eb830dd..1f923b3 100755 (executable)
--- a/lustre/tests/replay-dual.sh
+++ b/lustre/tests/replay-dual.sh
@@ -198,7 +198,7 @@ test_11() {
     #sleep for while, let both clients reconnect and timeout
     sleep $((TIMEOUT * 2))
     do_facet mds lctl set_param fail_loc=0
-    client_df
+    clients_up
     while [ -z "$(ls $MOUNT1/$tfile-[1-5] 2>/dev/null)" ]; do
        sleep 5
        echo -n "."
@@ -220,7 +220,7 @@ test_12() {
     do_facet mds lctl set_param fail_loc=0x80000302
     facet_failover mds
     do_facet mds lctl set_param fail_loc=0
-    df $MOUNT || { kill -USR1 $MULTIPID  && return 1; }
+    clients_up || { kill -USR1 $MULTIPID  && return 1; }
 
     ls $DIR/$tfile
     kill -USR1 $MULTIPID || return 3
@@ -245,7 +245,7 @@ test_13() {
     do_facet mds lctl set_param fail_loc=0x80000115
     facet_failover mds
     do_facet mds lctl set_param fail_loc=0
-    df $MOUNT || return 1
+    clients_up || return 1
 
     ls $DIR/$tfile
     $CHECKSTAT -t file $DIR/$tfile || return 2
@@ -264,7 +264,7 @@ test_14a() {
 
     facet_failover mds
     # expect recovery to fail due to missing client 2
-    df $MOUNT1 && return 1
+    client_evicted || return 1
     sleep 1
 
     # first 25 files should have been replayed
@@ -286,9 +286,7 @@ test_14b() {
     createmany -o $MOUNT1/$tfile-3- 5
     umount $MOUNT2
 
-    facet_failover mds
-    # expect recovery don't fail due to VBR
-    df $MOUNT1 || return 1
+    fail mds
 
     # first 25 files should have been replayed
     unlinkmany $MOUNT1/$tfile- 5 || return 2
@@ -311,8 +309,7 @@ test_15a() {        # was test_15
     createmany -o $MOUNT2/$tfile-2- 1
     umount $MOUNT2
 
-    facet_failover mds
-    df $MOUNT || return 1
+    fail mds
 
     unlinkmany $MOUNT1/$tfile- 25 || return 2
     [ -e $MOUNT1/$tfile-2-0 ] && error "$tfile-2-0 exists"
@@ -329,9 +326,7 @@ test_15c() {
     done
 
     umount $MOUNT2
-    facet_failover mds
-
-    df $MOUNT || return 1
+    fail mds
 
     zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
     return 0
@@ -346,8 +341,7 @@ test_16() {
 
     facet_failover mds
     sleep $TIMEOUT
-    facet_failover mds
-    df $MOUNT || return 1
+    fail mds
 
     unlinkmany $MOUNT1/$tfile- 25 || return 2
 
@@ -369,8 +363,7 @@ test_17() {
 
     facet_failover ost1
     sleep $TIMEOUT
-    facet_failover ost1
-    df $MOUNT || return 1
+    fail ost1
 
     unlinkmany $MOUNT1/$tfile- 25 || return 2
 
@@ -421,8 +414,7 @@ test_20() { #16389
     touch $MOUNT1/a
     touch $MOUNT2/b
     umount $MOUNT2
-    facet_failover mds
-    df $MOUNT1 || return 1
+    fail mds
     rm $MOUNT1/a
     zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
     TIER1=$((`date +%s` - BEFORE))
@@ -431,8 +423,7 @@ test_20() { #16389
     touch $MOUNT1/a
     touch $MOUNT2/b
     umount $MOUNT2
-    facet_failover mds
-    df $MOUNT1 || return 1
+    fail mds
     rm $MOUNT1/a
     zconf_mount `hostname` $MOUNT2 || error "mount $MOUNT2 fail"
     TIER2=$((`date +%s` - BEFORE))

diff --git a/lustre/tests/replay-single.sh b/lustre/tests/replay-single.sh
index 034aa8d..79d7380 100755 (executable)
--- a/lustre/tests/replay-single.sh
+++ b/lustre/tests/replay-single.sh
@@ -63,7 +63,7 @@ test_0c() {
     umount $DIR
     facet_failover mds
     zconf_mount `hostname` $DIR || error "mount fails"
-    df $DIR || error "post-failover df failed"
+    clients_up || error "post-failover df failed"
 }
 run_test 0c "expired recovery with no clients"
 
@@ -408,10 +408,9 @@ test_20b() { # bug 10480
     lfs getstripe $DIR/$tfile || return 1
     rm -f $DIR/$tfile || return 2       # make it an orphan
     mds_evict_client
-    df -P $DIR || df -P $DIR || true    # reconnect
+    client_up || client_up || true    # reconnect
 
     fail mds                            # start orphan recovery
-    df -P $DIR || df -P $DIR || true    # reconnect
     wait_recovery_complete mds || error "MDS recovery not done"
 
     # For interop with 2.0 only:
@@ -436,8 +435,7 @@ test_20c() { # bug 10480
     ls -la $DIR/$tfile
 
     mds_evict_client
-
-    df -P $DIR || df -P $DIR || true    # reconnect
+    client_up || client_up || true    # reconnect
 
     kill -USR1 $pid
     wait $pid || return 1
@@ -651,7 +649,7 @@ test_32() {
     multiop_bg_pause $DIR/$tfile O_c || return 3
     pid2=$!
     mds_evict_client
-    df $MOUNT || sleep 1 && df $MOUNT || return 1
+    client_up || client_up || return 1
     kill -USR1 $pid1
     kill -USR1 $pid2
     wait $pid1 || return 4
@@ -958,7 +956,7 @@ test_47() { # bug 2824
     # OBD_FAIL_OST_CREATE_NET 0x204
     fail ost1
     do_facet ost1 "lctl set_param fail_loc=0x80000204"
-    df $MOUNT || return 2
+    client_up || return 2
 
     # let the MDS discover the OST failure, attempt to recover, fail
     # and recover again.
@@ -982,7 +980,7 @@ test_48() {
     facet_failover mds
     #define OBD_FAIL_OST_EROFS 0x216
     do_facet ost1 "lctl set_param fail_loc=0x80000216"
-    df $MOUNT || return 2
+    client_up || return 2
 
     createmany -o $DIR/$tfile 20 20 || return 2
     unlinkmany $DIR/$tfile 40 || return 3
@@ -1489,8 +1487,7 @@ test_62() { # Bug 15756 - don't mis-drop resent replay
     createmany -o $DIR/$tdir/$tfile- 25
 #define OBD_FAIL_TGT_REPLAY_DROP         0x707
     do_facet mds "lctl set_param fail_loc=0x80000707"
-    facet_failover mds
-    df $MOUNT || return 1
+    fail mds
     do_facet mds "lctl set_param fail_loc=0"
     unlinkmany $DIR/$tdir/$tfile- 25 || return 2
     return 0
@@ -1831,7 +1828,7 @@ test_71a() {
     umount $DIR
     facet_failover mds
     zconf_mount `hostname` $DIR || error "mount fails"
-    df $DIR || error "post-failover df failed"
+    client_up || error "post-failover df failed"
     do_facet mds "lctl get_param -n mds.${mds_svc}.stale_exports|grep $UUID" || \
         error "no delayed exports"
     OLD_AGE=$(do_facet mds "lctl get_param -n mds.${mds_svc}.stale_export_age")
@@ -2080,6 +2077,7 @@ test_84() {
     PID=$!
     mds_evict_client
     wait $PID || true
+    client_up || client_up || true    # reconnect
 }
 run_test 84 "stale open during export disconnect"
 

diff --git a/lustre/tests/replay-vbr.sh b/lustre/tests/replay-vbr.sh
index cc172a1..e62e061 100644 (file)
--- a/lustre/tests/replay-vbr.sh
+++ b/lustre/tests/replay-vbr.sh
@@ -109,7 +109,7 @@ test_0b() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -a $DIR/$tdir/$tfile; then
         error "open succeeded unexpectedly"
     fi
@@ -128,8 +128,8 @@ test_0c() {
     rmultiop_start $CLIENT1 $DIR/$tdir/$tfile o_c
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
+    client_up $CLIENT1 || error "$CLIENT1 evicted"
 
-    do_node $CLIENT1 df $MOUNT || error "$CLIENT1 evicted"
     rmultiop_stop $CLIENT1 || error "close failed"
     zconf_mount $CLIENT2 $MOUNT
 }
@@ -158,7 +158,7 @@ test_0e() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -a $DIR/$tdir/$tfile; then
         error "create succeeded unexpectedly"
     fi
@@ -191,7 +191,7 @@ test_0g() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if do_node $CLIENT1 $CHECKSTAT -a $DIR/$tdir/$tfile; then
         error "unlink succeeded unexpectedly"
     fi
@@ -241,7 +241,7 @@ test_0j() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -u \\\#$UID $file; then
         error "setattr of UID succeeded unexpectedly"
     fi
@@ -261,7 +261,7 @@ test_0k() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -g \\\#$UID $file; then
         error "setattr of GID succeeded unexpectedly"
     fi
@@ -296,7 +296,7 @@ test_0m() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -p 0644 $file; then
         error "setattr of permission succeeded unexpectedly"
     fi
@@ -345,7 +345,7 @@ test_0o() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     checkattr $CLIENT1 i $file
     rc=$?
     do_node $CLIENT1 chattr -i $file
@@ -414,7 +414,7 @@ test_0r() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT || error "$CLIENT1 evicted"
+    client_up $CLIENT1 || error "$CLIENT1 evicted"
     if (($mtime_pre >= $mtime_post)); then
         error "time not changed: pre $mtime_pre, post $mtime_post"
     fi
@@ -462,7 +462,7 @@ test_0t() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -a $DIR/$tdir/$tfile; then
         error "link should fail"
     fi
@@ -481,7 +481,7 @@ test_0u() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if ! do_node $CLIENT1 $CHECKSTAT -a $DIR/$tdir/$tfile; then
         error "link should fail"
     fi
@@ -536,7 +536,7 @@ test_0x() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if do_node $CLIENT1 $CHECKSTAT -a $DIR/$tfile; then
         error "rename should fail"
     fi
@@ -555,7 +555,7 @@ test_0y() {
     zconf_umount $CLIENT2 $MOUNT
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1 not evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1 not evicted"
     if do_node $CLIENT1 $CHECKSTAT -a $DIR/$tfile; then
         error "rename should fail"
     fi
@@ -579,7 +579,7 @@ test_1() {
 
     facet_failover mds
     # recovery shouldn't fail due to missing client 2
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # All 50 files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 2
@@ -610,7 +610,7 @@ test_2a() { # was test_2
 
     facet_failover mds
     # recovery shouldn't fail due to missing client 2
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # All 50 files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 2
@@ -667,8 +667,8 @@ test_2b() {
     zconf_umount $CLIENT2 $MOUNT2
     facet_failover mds
 
-    do_node $CLIENT1 df $MOUNT && error "$CLIENT1:$MOUNT not evicted"
-    do_node $CLIENT2 df $MOUNT1 || error "$CLIENT2:$MOUNT1 evicted"
+    client_evicted $CLIENT1 || error "$CLIENT1:$MOUNT not evicted"
+    client_up $CLIENT2 || error "$CLIENT2:$MOUNT1 evicted"
 
     #
     # Check the MDT epoch.  $post must be the first transaction
@@ -717,7 +717,7 @@ test_3a() {
 
     facet_failover mds
     # recovery shouldn't fail due to missing client 2
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
     do_node $CLIENT1 $CHECKSTAT $DIR/$tfile && return 2
 
     zconf_mount $CLIENT2 $DIR || error "mount $CLIENT2 $DIR fail"
@@ -748,7 +748,7 @@ test_3b() {
 
     facet_failover mds
     # recovery should fail due to missing client 2
-    do_node $CLIENT1 df $DIR && return 1
+    client_evicted $CLIENT1 || return 1
 
     do_node $CLIENT1 $CHECKSTAT -p 0755 $DIR/$tfile && return 2
     zconf_mount $CLIENT2 $DIR || error "mount $CLIENT2 $DIR fail"
@@ -792,8 +792,7 @@ test_3c() {
 
     facet_failover mds
     # recovery shouldn't fail due to missing client 2
-    do_node $CLIENT1 df $DIR || return 1
-    sleep 1
+    client_up $CLIENT1 || return 1
 
     zconf_mount $CLIENT2 $DIR || error "mount $CLIENT2 $DIR fail"
     do_node $CLIENT1 $RUNAS cat $DIR/d3c/sub/$tfile && return 6
@@ -840,14 +839,14 @@ test_4a() {
     vbr_deactivate_client $CLIENT2
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # All 50 files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 2
     do_node $CLIENT1 unlinkmany $DIR/$tfile-3- 25 || return 3
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 4
+    client_up $CLIENT2 || return 4
     # All 25 files from client2 should have been replayed
     do_node $CLIENT2 unlinkmany $DIR/$tdir/$tfile-2- 25 || return 5
 
@@ -871,13 +870,13 @@ test_4b() {
     vbr_deactivate_client $CLIENT2
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # create another set of files
     do_node $CLIENT1 createmany -o $DIR/$tfile-3- 25
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 2
+    client_up $CLIENT2 || return 2
 
     # All files from should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 3
@@ -903,13 +902,13 @@ test_4c() {
     vbr_deactivate_client $CLIENT2
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # create another set of files
     do_node $CLIENT1 createmany -m $DIR/$tfile-3- 25
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 2
+    client_up $CLIENT2 || return 2
 
     # All files from should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 3
@@ -936,10 +935,10 @@ test_5a() {
     vbr_deactivate_client $CLIENT2
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR && return 1
+    client_evicted $CLIENT1 || return 1
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 2
+    client_up $CLIENT2 || return 2
 
     # First 25 files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 3
@@ -967,14 +966,14 @@ test_5b() {
     vbr_deactivate_client $CLIENT2
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
     do_node $CLIENT1 $CHECKSTAT $DIR/$tfile-2-0 && error "$tfile-2-0 exists"
 
     # create another set of files
     do_node $CLIENT1 createmany -o $DIR/$tfile-3- 25
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR && return 4
+    client_evicted $CLIENT2 || return 4
     # file from client2 should fail
     do_node $CLIENT2 $CHECKSTAT $DIR/$tfile-2-0 && error "$tfile-2-0 exists"
 
@@ -1007,13 +1006,13 @@ test_6a() {
     do_node $CLIENT2 "sysctl -w lustre.fail_val=5"
 #define OBD_FAIL_PTLRPC_REPLAY        0x50e
     do_node $CLIENT2 "sysctl -w lustre.fail_loc=0x2000050e"
-    do_node $CLIENT2 df $DIR
+    client_up $CLIENT2
     # vbr_activate_client $CLIENT2
     # need way to know that client stops replays
     sleep 5
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # All files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 2
@@ -1043,10 +1042,10 @@ test_7a() {
 
     facet_failover mds
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 4
+    client_up $CLIENT2 || return 4
 
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 1
+    client_up $CLIENT1 || return 1
 
     # All files should have been replayed
     do_node $CLIENT1 unlinkmany $DIR/$tfile- 25 || return 2
@@ -1074,10 +1073,10 @@ test_8a() {
 
     vbr_deactivate_client $CLIENT2
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 3
+    client_up $CLIENT1 || return 3
     #client1 is back and will try to open orphan
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 4
+    client_up $CLIENT2 || return 4
 
     do_node $CLIENT2 $CHECKSTAT $DIR/$tfile && error "$tfile exists"
     zconf_umount_clients $CLIENTS $DIR
@@ -1100,10 +1099,10 @@ test_8b() {
 
     vbr_deactivate_client $CLIENT2
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 2
+    client_up $CLIENT1 || return 2
     #client1 is back and will try to open orphan
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 3
+    client_up $CLIENT2 || return 3
 
     rmultiop_stop $CLIENT2 || return 1
     do_node $CLIENT2 $CHECKSTAT $DIR/$tfile && error "$tfile exists"
@@ -1128,10 +1127,10 @@ test_8c() {
 
     vbr_deactivate_client $CLIENT2
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 3
+    client_up $CLIENT1 || return 3
     #client1 is back and will try to open orphan
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 4
+    client_up $CLIENT2 || return 4
 
     do_node $CLIENT2 $CHECKSTAT $DIR/$tfile && error "$tfile exists"
     zconf_umount_clients $CLIENTS $DIR
@@ -1157,11 +1156,11 @@ test_8d() {
 
     vbr_deactivate_client $CLIENT2
     facet_failover mds
-    do_node $CLIENT1 df $DIR || return 6
+    client_up $CLIENT1 || return 6
 
     #client1 is back and will try to open orphan
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 8
+    client_up $CLIENT2 || return 8
 
     do_node $CLIENT2 $CHECKSTAT $DIR/$tfile && error "$tfile exists"
     zconf_umount_clients $CLIENTS $DIR
@@ -1183,7 +1182,7 @@ test_8e() {
 
     zconf_umount $CLIENT1 $DIR
     facet_failover mds
-    do_node $CLIENT2 df $DIR || return 6
+    client_up $CLIENT2 || return 6
 
     do_node $CLIENT2 rm $DIR/$tfile || error "$tfile doesn't exists"
     zconf_umount_clients $CLIENTS $DIR
@@ -1205,7 +1204,7 @@ test_8f() {
 
     zconf_umount $CLIENT1 $DIR
     facet_failover mds
-    do_node $CLIENT2 df $DIR || return 6
+    client_up $CLIENT2 || return 6
 
     do_node $CLIENT2 rm $DIR/$tfile || error "$tfile doesn't exists"
     zconf_umount $CLIENT2 $DIR
@@ -1264,7 +1263,7 @@ test_10 () {
     sleep $TIMEOUT
 
     vbr_activate_client $CLIENT2
-    do_node $CLIENT2 df $DIR || return 4
+    client_up $CLIENT2 || return 4
 
     for CLIENT in ${CLIENTS//,/ }; do
         PID=`cat pid.$CLIENT`

diff --git a/lustre/tests/test-framework.sh b/lustre/tests/test-framework.sh
index 5adcfea..bb112d5 100644 (file)
--- a/lustre/tests/test-framework.sh
+++ b/lustre/tests/test-framework.sh
@@ -983,15 +983,31 @@ wait_remote_prog () {
     return $rc
 }
 
-client_df() {
+clients_up() {
     # not every config has many clients
+    sleep 1
     if [ -n "$CLIENTS" ]; then
-        $PDSH $CLIENTS "df $MOUNT" > /dev/null
+        $PDSH $CLIENTS "stat -f $MOUNT" > /dev/null
     else
-       df $MOUNT > /dev/null
+        stat -f $MOUNT > /dev/null
     fi
 }
 
+client_up() {
+    local client=$1
+    # usually checked on particular client or locally
+    sleep 1
+    if [ ! -z "$client" ]; then
+        $PDSH $client "stat -f $MOUNT" > /dev/null
+    else
+        stat -f $MOUNT > /dev/null
+    fi
+}
+
+client_evicted() {
+    ! client_up $1
+}
+
 client_reconnect() {
     uname -n >> $MOUNT/recon
     if [ -z "$CLIENTS" ]; then
@@ -1012,7 +1028,7 @@ facet_failover() {
     shutdown_facet $facet
     [ -n "$sleep_time" ] && sleep $sleep_time
     reboot_facet $facet
-    client_df &
+    clients_up &
     DFPID=$!
     RECOVERY_START_TIME=`date +%s`
     echo "df pid is $DFPID"