X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Ftests%2Fconf-sanity.sh;h=1f22874ffa46e32163188091e1680035e11eeee2;hp=ae77611338d96116343e41bb0412a4bb99fceeb2;hb=658d349a5acac33d0e5b1d27c979873ae4c585b4;hpb=4a3a9c5be5732d448c3c2d255b5b779918442f5d;ds=sidebyside diff --git a/lustre/tests/conf-sanity.sh b/lustre/tests/conf-sanity.sh index ae77611..1f22874 100644 --- a/lustre/tests/conf-sanity.sh +++ b/lustre/tests/conf-sanity.sh @@ -871,8 +871,8 @@ test_23a() { # was test_23 "MOUNT_LUSTRE_PID $MOUNT_LUSTRE_PID still not killed in $WAIT secs" ps -ef | grep mount fi - stop_mds || error - stop_ost || error + stop_mds || error "stopping MDSes failed" + stop_ost || error "stopping OSSes failed" } run_test 23a "interrupt client during recovery mount delay" @@ -1325,11 +1325,6 @@ t32_check() { exit 0 fi - if [ -n "$($LCTL list_nids | grep -v '\(tcp\|lo\)[[:digit:]]*$')" ]; then - skip "LU-2200: Test cannot run over Infiniband" - exit 0 - fi - local IMGTYPE=$(facet_fstype $SINGLEMDS) tarballs=$($r find $RLUSTRE/tests -maxdepth 1 -name \'disk*-$IMGTYPE.tar.bz2\') @@ -1495,6 +1490,7 @@ t32_test() { local tarball=$1 local writeconf=$2 local dne_upgrade=${dne_upgrade:-"no"} + local ff_convert=${ff_convert:-"no"} local shall_cleanup_mdt=false local shall_cleanup_mdt1=false local shall_cleanup_ost=false @@ -1535,6 +1531,9 @@ t32_test() { echo " Kernel: $img_kernel" echo " Arch: $img_arch" + local version=$(version_code $img_commit) + [[ $version -gt $(version_code 2.4.0) ]] && ff_convert="no" + $r $LCTL set_param debug="$PTLDEBUG" $r $TUNEFS --dryrun $tmp/mdt || { @@ -1550,6 +1549,20 @@ t32_test() { } fi else + if [ -n "$($LCTL list_nids | grep -v '\(tcp\|lo\)[[:digit:]]*$')" ]; then + [[ $(lustre_version_code mgs) -ge $(version_code 2.3.59) ]] || + { skip "LU-2200: Cannot run over Inifiniband w/o lctl replace_nids " + "(Need MGS version at least 2.3.59)"; return 0; } + + local osthost=$(facet_active_host ost1) + local ostnid=$(do_node $osthost $LCTL list_nids | head -1) + + $r mount -t lustre -o loop,nosvc $tmp/mdt $tmp/mnt/mdt + $r lctl replace_nids $fsname-OST0000 $ostnid + $r lctl replace_nids $fsname-MDT0000 $nid + $r umount $tmp/mnt/mdt + fi + mopts=loop,exclude=$fsname-OST0000 fi @@ -1654,6 +1667,30 @@ t32_test() { return 1 } + if [ "$ff_convert" != "no" -a $(facet_fstype ost1) == "ldiskfs" ]; then + $r $LCTL lfsck_start -M $fsname-OST0000 || { + error_noexit "Start OI scrub on OST0" + return 1 + } + + # The oi_scrub should be on ost1, but for test_32(), + # all on the SINGLEMDS. + wait_update_facet $SINGLEMDS "$LCTL get_param -n \ + osd-ldiskfs.$fsname-OST0000.oi_scrub | + awk '/^status/ { print \\\$2 }'" "completed" 30 || { + error_noexit "Failed to get the expected 'completed'" + return 1 + } + + local UPDATED=$($r $LCTL get_param -n \ + osd-ldiskfs.$fsname-OST0000.oi_scrub | + awk '/^updated/ { print $2 }') + [ $UPDATED -ge 1 ] || { + error_noexit "Only $UPDATED objects have been converted" + return 1 + } + fi + if [ "$dne_upgrade" != "no" ]; then $r $LCTL conf_param \ $fsname-MDT0001.mdc.max_rpcs_in_flight=9 || { @@ -1760,8 +1797,8 @@ t32_test() { # on an architecture with different number of bits per # "long". # - if [ $(t32_bits_per_long $(uname -m)) != \ - $(t32_bits_per_long $img_arch) ]; then + if [ $(t32_bits_per_long $(uname -m)) != \ + $(t32_bits_per_long $img_arch) ]; then echo "Different number of bits per \"long\" from the disk image" for list in list.orig list; do sed -i -e 's/^[0-9]\+[ \t]\+//' $tmp/$list @@ -1834,7 +1871,7 @@ t32_test() { error_noexit "tunefs.lustre before remounting the MDT" return 1 } - $r mount -t lustre -o loop,exclude=$fsname-OST0000 $tmp/mdt \ + $r mount -t lustre -o loop,exclude=$fsname-OST0000 $tmp/mdt \ $tmp/mnt/mdt || { error_noexit "Remounting the MDT" return 1 @@ -1850,7 +1887,7 @@ test_32a() { t32_check for tarball in $tarballs; do - t32_test $tarball || rc=$? + t32_test $tarball || let "rc += $?" done return $rc } @@ -1863,7 +1900,7 @@ test_32b() { t32_check for tarball in $tarballs; do - t32_test $tarball writeconf || rc=$? + t32_test $tarball writeconf || let "rc += $?" done return $rc } @@ -1883,6 +1920,19 @@ test_32c() { } run_test 32c "dne upgrade test" +test_32d() { + local tarballs + local tarball + local rc=0 + + t32_check + for tarball in $tarballs; do + ff_convert=yes t32_test $tarball || rc=$? + done + return $rc +} +run_test 32d "convert ff test" + test_33a() { # bug 12333, was test_33 local rc=0 local FSNAME2=test-123 @@ -1999,9 +2049,10 @@ test_35a() { # bug 12459 log "Set up a fake failnode for the MDS" FAKENID="127.0.0.2" - local device=$(do_facet $SINGLEMDS "lctl get_param -n devices" | awk '($3 ~ "mdt" && $4 ~ "MDT") { print $4 }' | head -1) - do_facet mgs "$LCTL conf_param ${device}.failover.node=" \ - "$(h2$NETTYPE $FAKENID)" || return 4 + local device=$(do_facet $SINGLEMDS "lctl get_param -n devices" | + awk '($3 ~ "mdt" && $4 ~ "MDT") { print $4 }' | head -1) + do_facet mgs "$LCTL conf_param \ + ${device}.failover.node=$(h2$NETTYPE $FAKENID)" || return 4 log "Wait for RECONNECT_INTERVAL seconds (10s)" sleep 10 @@ -2053,10 +2104,10 @@ test_35b() { # bug 18674 log "Set up a fake failnode for the MDS" FAKENID="127.0.0.2" - local device=$(do_facet $SINGLEMDS "$LCTL get_param -n devices" | \ - awk '($3 ~ "mdt" && $4 ~ "MDT") { print $4 }' | head -1) - do_facet mgs "$LCTL conf_param ${device}.failover.node=" \ - "$(h2$NETTYPE $FAKENID)" || return 1 + local device=$(do_facet $SINGLEMDS "$LCTL get_param -n devices" | + awk '($3 ~ "mdt" && $4 ~ "MDT") { print $4 }' | head -1) + do_facet mgs "$LCTL conf_param \ + ${device}.failover.node=$(h2$NETTYPE $FAKENID)" || return 1 local at_max_saved=0 # adaptive timeouts may prevent seeing the issue @@ -3231,7 +3282,7 @@ thread_sanity() { tmin=$(do_facet $facet "lctl get_param -n ${paramp}.threads_min") tmax=$(do_facet $facet "lctl get_param -n ${paramp}.threads_max") tstarted=$(do_facet $facet "lctl get_param -n ${paramp}.threads_started") - lassert 28 "$msg" '(($tstarted == $tmin && $tstarted == $tmax ))' || return $? + lassert 28 "$msg" '(($tstarted >= $tmin && $tstarted <= $tmax ))' || return $? cleanup load_modules @@ -3353,6 +3404,7 @@ run_test 56 "check big indexes" test_57a() { # bug 22656 local NID=$(do_facet ost1 "$LCTL get_param nis" | tail -1 | awk '{print $1}') writeconf_or_reformat + [ $(facet_fstype ost1) == zfs ] && import_zpool ost1 do_facet ost1 "$TUNEFS --failnode=$NID `ostdevname 1`" || error "tunefs failed" start_mgsmds start_ost && error "OST registration from failnode should fail" @@ -3363,6 +3415,7 @@ run_test 57a "initial registration from failnode should fail (should return errs test_57b() { local NID=$(do_facet ost1 "$LCTL get_param nis" | tail -1 | awk '{print $1}') writeconf_or_reformat + [ $(facet_fstype ost1) == zfs ] && import_zpool ost1 do_facet ost1 "$TUNEFS --servicenode=$NID `ostdevname 1`" || error "tunefs failed" start_mgsmds start_ost || error "OST registration from servicenode should not fail" @@ -3705,6 +3758,110 @@ test_66() { } run_test 66 "replace nids" +test_67() { #LU-2950 + local legacy="$TMP/legacy_lnet_config" + local new="$TMP/new_routes_test" + local out="$TMP/config_out_file" + local verify="$TMP/conv_verify" + local verify_conf="$TMP/conf_verify" + + # Create the legacy file that will be run through the + # lustre_routes_conversion script + cat <<- LEGACY_LNET_CONFIG > $legacy + tcp1 23 192.168.213.1@tcp:1; tcp5 34 193.30.4.3@tcp:4; + tcp2 54 10.1.3.2@tcp; + tcp3 10.3.4.3@tcp:3; + tcp4 10.3.3.4@tcp; + LEGACY_LNET_CONFIG + + # Create the verification file to verify the output of + # lustre_routes_conversion script against. + cat <<- VERIFY_LNET_CONFIG > $verify + tcp1: { gateway: 192.168.213.1@tcp, hop: 23, priority: 1 } + tcp5: { gateway: 193.30.4.3@tcp, hop: 34, priority: 4 } + tcp2: { gateway: 10.1.3.2@tcp, hop: 54 } + tcp3: { gateway: 10.3.4.3@tcp, priority: 3 } + tcp4: { gateway: 10.3.3.4@tcp } + VERIFY_LNET_CONFIG + + # Create the verification file to verify the output of + # lustre_routes_config script against + cat <<- VERIFY_LNET_CONFIG > $verify_conf + lctl --net tcp1 add_route 192.168.213.1@tcp 23 1 + lctl --net tcp5 add_route 193.30.4.3@tcp 34 4 + lctl --net tcp2 add_route 10.1.3.2@tcp 54 4 + lctl --net tcp3 add_route 10.3.4.3@tcp 1 3 + lctl --net tcp4 add_route 10.3.3.4@tcp 1 3 + VERIFY_LNET_CONFIG + + lustre_routes_conversion $legacy $new > /dev/null + if [ -f $new ]; then + # verify the conversion output + cmp -s $new $verify > /dev/null + if [ $? -eq 1 ]; then + error "routes conversion failed" + fi + + lustre_routes_config --dry-run --verbose $new > $out + # check that the script succeeded + cmp -s $out $verify_conf > /dev/null + if [ $? -eq 1 ]; then + error "routes config failed" + fi + else + error "routes conversion test failed" + fi + # remove generated files + rm -f $new $legacy $verify $verify_conf $out +} +run_test 67 "test routes conversion and configuration" + +test_68() { + local fid + local seq + local START + local END + + [ $(lustre_version_code $SINGLEMDS) -ge $(version_code 2.4.53) ] || + { skip "Need MDS version at least 2.4.53"; return 0; } + + umount_client $MOUNT || error "umount client failed" + + start_mdt 1 || error "MDT start failed" + start_ost + + # START-END - the sequences we'll be reserving + START=$(do_facet $SINGLEMDS \ + lctl get_param -n seq.ctl*.space | awk -F'[[ ]' '{print $2}') + END=$((START + (1 << 30))) + do_facet $SINGLEMDS \ + lctl set_param seq.ctl*.fldb="[$START-$END\):0:mdt" + + # reset the sequences MDT0000 has already assigned + do_facet $SINGLEMDS \ + lctl set_param seq.srv*MDT0000.space=clear + + # remount to let the client allocate new sequence + mount_client $MOUNT || error "mount client failed" + + touch $DIR/$tfile + do_facet $SINGLEMDS \ + lctl get_param seq.srv*MDT0000.space + $LFS path2fid $DIR/$tfile + + local old_ifs="$IFS" + IFS='[:]' + fid=($($LFS path2fid $DIR/$tfile)) + IFS="$old_ifs" + let seq=${fid[1]} + + if [[ $seq < $END ]]; then + error "used reserved sequence $seq?" + fi + cleanup || return $? +} +run_test 68 "be able to reserve specific sequences in FLDB" + test_70a() { [ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return local MDTIDX=1 @@ -3994,15 +4151,16 @@ run_test 72 "test fast symlink with extents flag enabled" test_73() { #LU-3006 load_modules - do_facet ost1 "$TUNEFS --failnode=1.2.3.4@tcp $(ostdevname 1)" || + [ $(facet_fstype ost1) == zfs ] && import_zpool ost1 + do_facet ost1 "$TUNEFS --failnode=1.2.3.4@$NETTYPE $(ostdevname 1)" || error "1st tunefs failed" start_mgsmds || error "start mds failed" start_ost || error "start ost failed" mount_client $MOUNT || error "mount client failed" lctl get_param -n osc.*OST0000-osc-[^M]*.import | grep failover_nids | - grep 1.2.3.4@tcp || error "failover nids haven't changed" + grep 1.2.3.4@$NETTYPE || error "failover nids haven't changed" umount_client $MOUNT || error "umount client failed" - stop_all + stopall reformat } run_test 73 "failnode to update from mountdata properly" @@ -4018,6 +4176,133 @@ test_74() { # LU-1606 } run_test 74 "Lustre client api program can compile and link" +test_75() { # LU-2374 + [[ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.4.1) ]] && + skip "Need MDS version at least 2.4.1" && return + + local index=0 + local opts_mds="$(mkfs_opts mds1 $(mdsdevname 1)) \ + --reformat $(mdsdevname 1) $(mdsvdevname 1)" + local opts_ost="$(mkfs_opts ost1 $(ostdevname 1)) \ + --reformat $(ostdevname 1) $(ostvdevname 1)" + + #check with default parameters + add mds1 $opts_mds || error "add mds1 failed for default params" + add ost1 $opts_ost || error "add ost1 failed for default params" + + opts_mds=$(echo $opts_mds | sed -e "s/--mdt//") + opts_mds=$(echo $opts_mds | + sed -e "s/--index=$index/--index=$index --mdt/") + opts_ost=$(echo $opts_ost | sed -e "s/--ost//") + opts_ost=$(echo $opts_ost | + sed -e "s/--index=$index/--index=$index --ost/") + + add mds1 $opts_mds || error "add mds1 failed for new params" + add ost1 $opts_ost || error "add ost1 failed for new params" + return 0 +} +run_test 75 "The order of --index should be irrelevant" + +test_76() { + [[ $(lustre_version_code mgs) -ge $(version_code 2.4.52) ]] || + { skip "Need MDS version at least 2.4.52" && return 0; } + setup + local MDMB_PARAM="osc.*.max_dirty_mb" + echo "Change MGS params" + local MAX_DIRTY_MB=$($LCTL get_param -n $MDMB_PARAM | + head -1) + echo "max_dirty_mb: $MAX_DIRTY_MB" + local NEW_MAX_DIRTY_MB=$((MAX_DIRTY_MB + MAX_DIRTY_MB)) + echo "new_max_dirty_mb: $NEW_MAX_DIRTY_MB" + do_facet mgs $LCTL set_param -P $MDMB_PARAM=$NEW_MAX_DIRTY_MB + wait_update $HOSTNAME "lctl get_param -n $MDMB_PARAM | + head -1" $NEW_MAX_DIRTY_MB + MAX_DIRTY_MB=$($LCTL get_param -n $MDMB_PARAM | head -1) + echo "$MAX_DIRTY_MB" + [ $MAX_DIRTY_MB = $NEW_MAX_DIRTY_MB ] || + error "error while apply max_dirty_mb" + + echo "Check the value is stored after remount" + stopall + setupall + wait_update $HOSTNAME "lctl get_param -n $MDMB_PARAM | + head -1" $NEW_MAX_DIRTY_MB + MAX_DIRTY_MB=$($LCTL get_param -n $MDMB_PARAM | head -1) + [ $MAX_DIRTY_MB = $NEW_MAX_DIRTY_MB ] || + error "max_dirty_mb is not saved after remount" + + echo "Change OST params" + CLIENT_PARAM="obdfilter.*.client_cache_count" + local CLIENT_CACHE_COUNT + CLIENT_CACHE_COUNT=$(do_facet ost1 $LCTL get_param -n $CLIENT_PARAM | + head -1) + echo "client_cache_count: $CLIENT_CACHE_COUNT" + NEW_CLIENT_CACHE_COUNT=$((CLIENT_CACHE_COUNT+CLIENT_CACHE_COUNT)) + echo "new_client_cache_count: $NEW_CLIENT_CACHE_COUNT" + do_facet mgs $LCTL set_param -P $CLIENT_PARAM=$NEW_CLIENT_CACHE_COUNT + wait_update $(facet_host ost1) "lctl get_param -n $CLIENT_PARAM | + head -1" $NEW_CLIENT_CACHE_COUNT + CLIENT_CACHE_COUNT=$(do_facet ost1 $LCTL get_param -n $CLIENT_PARAM | + head -1) + echo "$CLIENT_CACHE_COUNT" + [ $CLIENT_CACHE_COUNT = $NEW_CLIENT_CACHE_COUNT ] || + error "error while apply client_cache_count" + + echo "Check the value is stored after remount" + stopall + setupall + wait_update $(facet_host ost1) "lctl get_param -n $CLIENT_PARAM | + head -1" $NEW_CLIENT_CACHE_COUNT + CLIENT_CACHE_COUNT=$(do_facet ost1 $LCTL get_param -n $CLIENT_PARAM | + head -1) + echo "$CLIENT_CACHE_COUNT" + [ $CLIENT_CACHE_COUNT = $NEW_CLIENT_CACHE_COUNT ] || + error "client_cache_count is not saved after remount" + stopall +} +run_test 76 "set permanent params set_param -P" + +test_77() { # LU-3445 + local server_version=$(lustre_version_code $SINGLEMDS) + + [[ $server_version -ge $(version_code 2.2.60) ]] && + [[ $server_version -le $(version_code 2.4.0) ]] && + skip "Need MDS version < 2.2.60 or > 2.4.0" && return + + if [[ -z "$fs2ost_DEV" || -z "$fs2mds_DEV" ]]; then + is_blkdev $SINGLEMDS $(mdsdevname ${SINGLEMDS//mds/}) && + skip_env "mixed loopback and real device not working" && return + fi + + local fs2mdsdev=$(mdsdevname 1_2) + local fs2ostdev=$(ostdevname 1_2) + local fs2mdsvdev=$(mdsvdevname 1_2) + local fs2ostvdev=$(ostvdevname 1_2) + local fsname=test1234 + local mgsnid + local failnid="$(h2$NETTYPE 1.2.3.4),$(h2$NETTYPE 4.3.2.1)" + + add fs2mds $(mkfs_opts mds1 $fs2mdsdev) --mgs --fsname=$fsname \ + --reformat $fs2mdsdev $fs2mdsvdev || error "add fs2mds failed" + start fs2mds $fs2mdsdev $MDS_MOUNT_OPTS && trap cleanup_24a EXIT INT || + error "start fs2mds failed" + + mgsnid=$(do_facet fs2mds $LCTL list_nids | xargs | tr ' ' ,) + [[ $mgsnid = *,* ]] || mgsnid+=",$mgsnid" + + add fs2ost $(mkfs_opts ost1 $fs2ostdev) --mgsnode=$mgsnid \ + --failnode=$failnid --fsname=$fsname \ + --reformat $fs2ostdev $fs2ostvdev || + error "add fs2ost failed" + start fs2ost $fs2ostdev $OST_MOUNT_OPTS || error "start fs2ost failed" + + mkdir -p $MOUNT2 + mount -t lustre $mgsnid:/$fsname $MOUNT2 || error "mount $MOUNT2 failed" + DIR=$MOUNT2 MOUNT=$MOUNT2 check_mount || error "check $MOUNT2 failed" + cleanup_24a +} +run_test 77 "comma-separated MGS NIDs and failover node NIDs" + if ! combined_mgs_mds ; then stop mgs fi