X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Ftests%2Fsanity-lfsck.sh;h=f19d464645d7623ad0ead0a0f5616298d77d85fa;hb=be009cb4a73b3bef7302083bec7d1d6289d515b7;hp=4456975d2ab8595ac1cab65433bcb0bbf128f1f2;hpb=077570483e75e0610fd45149b926097547c434b8;p=fs%2Flustre-release.git diff --git a/lustre/tests/sanity-lfsck.sh b/lustre/tests/sanity-lfsck.sh index 4456975..f19d464 100644 --- a/lustre/tests/sanity-lfsck.sh +++ b/lustre/tests/sanity-lfsck.sh @@ -8,18 +8,31 @@ set -e ONLY=${ONLY:-"$*"} -#Bug number for excepting test -ALWAYS_EXCEPT="$SANITY_LFSCK_EXCEPT" - -[ "$SLOW" = "no" ] && EXCEPT_SLOW="" -# UPDATE THE COMMENT ABOVE WITH BUG NUMBERS WHEN CHANGING ALWAYS_EXCEPT! - -LUSTRE=${LUSTRE:-$(cd $(dirname $0)/..; echo $PWD)} +LUSTRE=${LUSTRE:-$(dirname $0)/..} . $LUSTRE/tests/test-framework.sh init_test_env $@ -. ${CONFIG:=$LUSTRE/tests/cfg/$NAME.sh} init_logging +ALWAYS_EXCEPT="$SANITY_LFSCK_EXCEPT " + +# DNE does not support striped directory on zfs-based backend yet. +[ $(facet_fstype $SINGLEMDS) != ldiskfs ] && +#Bug number for excepting test + ALWAYS_EXCEPT+=" 31" +# UPDATE THE COMMENT ABOVE WITH BUG NUMBERS WHEN CHANGING ALWAYS_EXCEPT! + +[[ $(lustre_version_code $SINGLEMDS) -le $(version_code 2.4.90) ]] && + ALWAYS_EXCEPT="$ALWAYS_EXCEPT 2c" + +[[ $(lustre_version_code ost1) -lt $(version_code 2.5.55) ]] && + ALWAYS_EXCEPT="$ALWAYS_EXCEPT 11 12 13 14 15 16 17 18 19 20 21" + +[[ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.6.50) ]] && + ALWAYS_EXCEPT="$ALWAYS_EXCEPT 2d 2e 3 22 23 24 25 26 27 28 29 30 31" + +[ "$SLOW" = "no" ] && EXCEPT_SLOW="" +build_test_filter + require_dsh_mds || exit 0 load_modules @@ -51,21 +64,6 @@ cleanupall # build up a clean test environment. REFORMAT="yes" check_and_setup_lustre -[[ $(lustre_version_code $SINGLEMDS) -le $(version_code 2.4.90) ]] && - ALWAYS_EXCEPT="$ALWAYS_EXCEPT 2c" - -[[ $(lustre_version_code ost1) -lt $(version_code 2.5.55) ]] && - ALWAYS_EXCEPT="$ALWAYS_EXCEPT 11 12 13 14 15 16 17 18 19 20 21" - -[[ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.6.50) ]] && - ALWAYS_EXCEPT="$ALWAYS_EXCEPT 2d 2e 3 22 23 24 25 26 27 28 29 30 31" - -# DNE does not support striped directory on zfs-based backend yet. -[ $(facet_fstype $SINGLEMDS) != ldiskfs ] && - ALWAYS_EXCEPT="$ALWAYS_EXCEPT 31" - -build_test_filter - MDT_DEV="${FSNAME}-MDT0000" OST_DEV="${FSNAME}-OST0000" MDT_DEVNAME=$(mdsdevname ${SINGLEMDS//mds/}) @@ -81,8 +79,8 @@ SHOW_LAYOUT="do_facet $SINGLEMDS \ $LCTL get_param -n mdd.${MDT_DEV}.lfsck_layout" SHOW_LAYOUT_ON_OST="do_facet ost1 \ $LCTL get_param -n obdfilter.${OST_DEV}.lfsck_layout" -MOUNT_OPTS_SCRUB="-o user_xattr" -MOUNT_OPTS_NOSCRUB="-o user_xattr,noscrub" +MOUNT_OPTS_SCRUB="$MDS_MOUNT_OPTS -o user_xattr" +MOUNT_OPTS_NOSCRUB="$MDS_MOUNT_OPTS -o user_xattr,noscrub" MOUNT_OPTS_SKIP_LFSCK="-o user_xattr,skip_lfsck" lfsck_prep() { @@ -784,7 +782,10 @@ test_6b() { local D_POS1=$($SHOW_NAMESPACE | awk '/^latest_start_position/ { print $4 }') - if [ "$D_POS0" == "N/A" -o "$D_POS1" == "N/A" ]; then + echo "Additional debug for 6b" + $SHOW_NAMESPACE + if [ "$D_POS0" == "N/A" -o "$D_POS0" == "0x0" \ + -o "$D_POS1" == "0x0" -o "$D_POS1" == "N/A" ]; then [[ $O_POS0 -lt $O_POS1 ]] || error "(7.1) $O_POS1 is not larger than $O_POS0" else @@ -1326,7 +1327,7 @@ ost_remove_lastid() { test_11a() { check_mount_and_prep - $SETSTRIPE -c 1 -i 0 $DIR/$tdir + $LFS setstripe -c 1 -i 0 $DIR/$tdir createmany -o $DIR/$tdir/f 64 || error "(0) Fail to create 64 files." echo "stopall" @@ -1367,7 +1368,7 @@ run_test 11a "LFSCK can rebuild lost last_id" test_11b() { check_mount_and_prep - $SETSTRIPE -c 1 -i 0 $DIR/$tdir + $LFS setstripe -c 1 -i 0 $DIR/$tdir echo "set fail_loc=0x160d to skip the updating LAST_ID on-disk" #define OBD_FAIL_LFSCK_SKIP_LASTID 0x160d @@ -2127,6 +2128,8 @@ test_18a() { run_test 18a "Find out orphan OST-object and repair it (1)" test_18b() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" + echo "#####" echo "The target MDT-object is lost. The LFSCK should re-create the" echo "MDT-object under .lustre/lost+found/MDTxxxx. The admin should" @@ -2186,6 +2189,28 @@ test_18b() { cancel_lru_locks mdc cancel_lru_locks osc + # dryrun mode only check orphans, not repaie + echo "Trigger layout LFSCK --dryrun to find out orphan OST-object" + $START_LAYOUT --dryrun -o -r || + error "Fail to start layout LFSCK in dryrun mode" + wait_all_targets_blocked layout completed 2 + + local PARAMS=$($SHOW_LAYOUT | awk '/^param/ { print $2 }') + [ "$PARAMS" == "dryrun,all_targets,orphan" ] || + error "Expect 'dryrun,all_targets,orphan', got '$PARAMS'" + + local orphans=$(do_facet mds1 $LCTL get_param -n \ + mdd.$(facet_svc mds1).lfsck_layout | + awk '/^inconsistent_orphan/ { print $2 }') + [ $orphans -eq 3 ] || + error "Expect 3 found on mds1, but got: $orphans" + + # orphan parents should not be created + local subdir + for subdir in $MOUNT/.lustre/lost+found/*; do + [ ! "$(ls -A $subdir)" ] || error "$subdir not empty" + done + echo "Trigger layout LFSCK on all devices to find out orphan OST-object" $START_LAYOUT -r -o || error "(1) Fail to start LFSCK for layout!" @@ -2262,6 +2287,8 @@ test_18b() { run_test 18b "Find out orphan OST-object and repair it (2)" test_18c() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" + echo "#####" echo "The target MDT-object is lost, and the OST-object FID is missing." echo "The LFSCK should re-create the MDT-object with new FID under the " @@ -2487,6 +2514,8 @@ test_18d() { run_test 18d "Find out orphan OST-object and repair it (4)" test_18e() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" + echo "#####" echo "The target MDT-object layout EA slot is occpuied by some new" echo "created OST-object when repair dangling reference case. Such" @@ -2777,6 +2806,8 @@ test_18f() { run_test 18f "Skip the failed OST(s) when handle orphan OST-objects" test_18g() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" + echo "#####" echo "The target MDT-object is lost, but related OI mapping is there" echo "The LFSCK should recreate the lost MDT-object without affected" @@ -2990,6 +3021,7 @@ PATTERN_WITHOUT_HOLE="raid0" test_20a() { [ $OSTCOUNT -lt 2 ] && skip "needs >= 2 OSTs" && return + [ -n "$FILESET" ] && skip "Not functional for FILESET set" echo "#####" echo "The target MDT-object and some of its OST-object are lost." @@ -3329,6 +3361,7 @@ run_test 20a "Handle the orphan with dummy LOV EA slot properly" test_20b() { [ $OSTCOUNT -lt 2 ] && skip "needs >= 2 OSTs" && return + [ -n "$FILESET" ] && skip "Not functional for FILESET set" echo "#####" echo "The target MDT-object and some of its OST-object are lost." @@ -4011,6 +4044,7 @@ run_test 23c "LFSCK can repair dangling name entry (3)" test_24() { [ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return + [ -n "$FILESET" ] && skip "Not functional for FILESET set" echo "#####" echo "Two MDT-objects back reference the same name entry via their" @@ -4221,6 +4255,8 @@ test_26b() { run_test 26b "LFSCK can add the missing remote name entry back to the namespace" test_27a() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" + echo "#####" echo "The local parent referenced by the MDT-object linkEA is lost." echo "The namespace LFSCK will re-create the lost parent as orphan." @@ -4274,6 +4310,7 @@ run_test 27a "LFSCK can recreate the lost local parent directory as orphan" test_27b() { [ $MDSCOUNT -lt 2 ] && skip "needs >= 2 MDTs" && return + [ -n "$FILESET" ] && skip "Not functional for FILESET set" echo "#####" echo "The remote parent referenced by the MDT-object linkEA is lost." @@ -4604,6 +4641,7 @@ run_test 29c "verify linkEA size limitation" test_30() { [ $(facet_fstype $SINGLEMDS) != ldiskfs ] && skip "ldiskfs only test" && return + [ -n "$FILESET" ] && skip "Not functional for FILESET set" echo "#####" echo "The namespace LFSCK will move the orphans from backend" @@ -4756,7 +4794,7 @@ test_31b() { #define OBD_FAIL_LFSCK_BAD_NAME_HASH 0x1628 $LCTL set_param fail_loc=0x1628 fail_val=1 - createmany -d $DIR/$tdir/striped_dir/d $MDSCOUNT || + createmany -d $DIR/$tdir/striped_dir/d $((MDSCOUNT * 5)) || error "(2) Fail to create file under striped directory" $LCTL set_param fail_loc=0 fail_val=0 @@ -4769,13 +4807,14 @@ test_31b() { local repaired=$(do_facet mds2 $LCTL get_param -n \ mdd.$(facet_svc mds2).lfsck_namespace | awk '/^name_hash_repaired/ { print $2 }') + echo "repaired $repaired name entries with bad hash" [ $repaired -ge 1 ] || error "(5) Fail to repair bad name hash: $repaired" umount_client $MOUNT || error "(6) umount failed" mount_client $MOUNT || error "(7) mount failed" - for ((i = 0; i < $MDSCOUNT; i++)); do + for ((i = 0; i < $((MDSCOUNT * 5)); i++)); do stat $DIR/$tdir/striped_dir/d$i || error "(8) Fail to stat d$i after LFSCK" rmdir $DIR/$tdir/striped_dir/d$i || @@ -5233,6 +5272,11 @@ test_36a() { check_mount_and_prep + lfs df $DIR + lfs df -i $DIR + lctl get_param osc.*.*grant* + stack_trap "lfs df $DIR; lfs df -i $DIR; lctl get_param osc.*.*grant*" + $LFS setstripe -N -E 1M -o 0,1 -E -1 -o 2 -N -E 2M -o 1,2 -E -1 -o 0 \ -N -E 3M -o 2,0 -E -1 -o 1 $DIR/$tdir/f0 || error "(0) Fail to create mirror file $DIR/$tdir/f0" @@ -5350,6 +5394,7 @@ test_36a() { run_test 36a "rebuild LOV EA for mirrored file (1)" test_36b() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" [ $OSTCOUNT -lt 3 ] && skip "needs >= 3 OSTs" && return echo "#####" @@ -5437,6 +5482,7 @@ test_36b() { run_test 36b "rebuild LOV EA for mirrored file (2)" test_36c() { + [ -n "$FILESET" ] && skip "Not functional for FILESET set" [ $OSTCOUNT -lt 3 ] && skip "needs >= 3 OSTs" && return echo "#####" @@ -5554,6 +5600,221 @@ test_37() } run_test 37 "LFSCK must skip a ORPHAN" +test_38() +{ + [[ $MDS1_VERSION -le $(version_code 2.12.51) ]] && + skip "Need MDS version newer than 2.12.51" + + test_mkdir $DIR/$tdir + local uuid1=$(cat /proc/sys/kernel/random/uuid) + local uuid2=$(cat /proc/sys/kernel/random/uuid) + + # create foreign file + $LFS setstripe --foreign=daos --flags 0xda05 \ + -x "${uuid1}@${uuid2}" $DIR/$tdir/$tfile || + error "$DIR/$tdir/$tfile: create failed" + + $LFS getstripe -v $DIR/$tdir/$tfile | + grep "lfm_magic:.*0x0BD70BD0" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign magic" + # lfm_length is LOV EA size - sizeof(lfm_magic) - sizeof(lfm_length) + $LFS getstripe -v $DIR/$tdir/$tfile | grep "lfm_length:.*73" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign size" + $LFS getstripe -v $DIR/$tdir/$tfile | grep "lfm_type:.*daos" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign type" + $LFS getstripe -v $DIR/$tdir/$tfile | + grep "lfm_flags:.*0x0000DA05" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign flags" + $LFS getstripe $DIR/$tdir/$tfile | + grep "lfm_value:.*${uuid1}@${uuid2}" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign value" + + # modify striping should fail + $LFS setstripe -c 2 $DIR/$tdir/$tfile && + error "$DIR/$tdir/$tfile: setstripe should fail" + + $START_NAMESPACE -r -A || error "Fail to start LFSCK for namespace" + + wait_all_targets_blocked namespace completed 1 + + # check that "global" namespace_repaired == 0 !!! + local repaired=$(do_facet mds1 \ + "$LCTL lfsck_query -t all -M ${FSNAME}-MDT0000 | + awk '/^namespace_repaired/ { print \\\$2 }'") + [ $repaired -eq 0 ] || + error "(2) Expect no namespace repair, but got: $repaired" + + $START_LAYOUT -A -r || error "Fail to start LFSCK for layout" + + wait_all_targets_blocked layout completed 2 + + # check that "global" layout_repaired == 0 !!! + local repaired=$(do_facet mds1 \ + "$LCTL lfsck_query -t all -M ${FSNAME}-MDT0000 | + awk '/^layout_repaired/ { print \\\$2 }'") + [ $repaired -eq 0 ] || + error "(2) Expect no layout repair, but got: $repaired" + + echo "post-lfsck checks of foreign file" + + $LFS getstripe -v $DIR/$tdir/$tfile | + grep "lfm_magic:.*0x0BD70BD0" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign magic" + # lfm_length is LOV EA size - sizeof(lfm_magic) - sizeof(lfm_length) + $LFS getstripe -v $DIR/$tdir/$tfile | grep "lfm_length:.*73" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign size" + $LFS getstripe -v $DIR/$tdir/$tfile | grep "lfm_type:.*daos" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign type" + $LFS getstripe -v $DIR/$tdir/$tfile | + grep "lfm_flags:.*0x0000DA05" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign flags" + $LFS getstripe $DIR/$tdir/$tfile | + grep "lfm_value:.*${uuid1}@${uuid2}" || + error "$DIR/$tdir/$tfile: invalid LOV EA foreign value" + + # modify striping should fail + $LFS setstripe -c 2 $DIR/$tdir/$tfile && + error "$DIR/$tdir/$tfile: setstripe should fail" + + # R/W should fail + cat $DIR/$tdir/$tfile && "$DIR/$tdir/$tfile: read should fail" + cat /etc/passwd > $DIR/$tdir/$tfile && + error "$DIR/$tdir/$tfile: write should fail" + + #remove foreign file + rm $DIR/$tdir/$tfile || + error "$DIR/$tdir/$tfile: remove of foreign file has failed" +} +run_test 38 "LFSCK does not break foreign file and reverse is also true" + +test_39() +{ + [[ $(lustre_version_code $SINGLEMDS) -le $(version_code 2.12.51) ]] && + skip "Need MDS version newer than 2.12.51" + + test_mkdir $DIR/$tdir + local uuid1=$(cat /proc/sys/kernel/random/uuid) + local uuid2=$(cat /proc/sys/kernel/random/uuid) + + # create foreign dir + $LFS mkdir --foreign=daos --xattr="${uuid1}@${uuid2}" --flags=0xda05 \ + $DIR/$tdir/${tdir}2 || + error "$DIR/$tdir/${tdir}2: create failed" + + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | + grep "lfm_magic:.*0x0CD50CD0" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA magic" + # lfm_length is LMV EA size - sizeof(lfm_magic) - sizeof(lfm_length) + # - sizeof(lfm_type) - sizeof(lfm_flags) + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | grep "lfm_length:.*73" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA size" + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | grep "lfm_type:.*daos" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA type" + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | + grep "lfm_flags:.*0x0000DA05" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA flags" + $LFS getdirstripe $DIR/$tdir/${tdir}2 | + grep "lfm_value.*${uuid1}@${uuid2}" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA value" + + # file create in dir should fail + touch $DIR/$tdir/${tdir}2/$tfile && + "$DIR/${tdir}2: file create should fail" + + # chmod should work + chmod 777 $DIR/$tdir/${tdir}2 || + error "$DIR/${tdir}2: chmod failed" + + # chown should work + chown $RUNAS_ID:$RUNAS_GID $DIR/$tdir/${tdir}2 || + error "$DIR/${tdir}2: chown failed" + + $START_NAMESPACE -r -A || error "Fail to start LFSCK for namespace" + + wait_all_targets_blocked namespace completed 1 + + # check that "global" namespace_repaired == 0 !!! + local repaired=$(do_facet mds1 \ + "$LCTL lfsck_query -t all -M ${FSNAME}-MDT0000 | + awk '/^namespace_repaired/ { print \\\$2 }'") + [ $repaired -eq 0 ] || + error "(2) Expect nothing to be repaired, but got: $repaired" + + $START_LAYOUT -A -r || error "Fail to start LFSCK for layout" + + wait_all_targets_blocked layout completed 2 + + # check that "global" layout_repaired == 0 !!! + local repaired=$(do_facet mds1 \ + "$LCTL lfsck_query -t all -M ${FSNAME}-MDT0000 | + awk '/^layout_repaired/ { print \\\$2 }'") + [ $repaired -eq 0 ] || + error "(2) Expect no layout repair, but got: $repaired" + + echo "post-lfsck checks of foreign dir" + + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | + grep "lfm_magic:.*0x0CD50CD0" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA magic" + # lfm_length is LMV EA size - sizeof(lfm_magic) - sizeof(lfm_length) + # - sizeof(lfm_type) - sizeof(lfm_flags) + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | grep "lfm_length:.*73" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA size" + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | grep "lfm_type:.*daos" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA type" + $LFS getdirstripe -v $DIR/$tdir/${tdir}2 | + grep "lfm_flags:.*0x0000DA05" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA flags" + $LFS getdirstripe $DIR/$tdir/${tdir}2 | + grep "lfm_value.*${uuid1}@${uuid2}" || + error "$DIR/$tdir/${tdir}2: invalid LMV EA value" + + # file create in dir should fail + touch $DIR/$tdir/${tdir}2/$tfile && + "$DIR/${tdir}2: file create should fail" + + # chmod should work + chmod 777 $DIR/$tdir/${tdir}2 || + error "$DIR/${tdir}2: chmod failed" + + # chown should work + chown $RUNAS_ID:$RUNAS_GID $DIR/$tdir/${tdir}2 || + error "$DIR/${tdir}2: chown failed" + + #remove foreign dir + rmdir $DIR/$tdir/${tdir}2 || + error "$DIR/$tdir/${tdir}2: remove of foreign dir has failed" +} +run_test 39 "LFSCK does not break foreign dir and reverse is also true" + +test_40a() { + [[ $MDSCOUNT -ge 2 ]] || skip "needs >= 2 MDTs" + + check_mount_and_prep + $LFS mkdir -i 1 $DIR/$tdir/dir1 + $LFS setstripe -E 1M -c1 -S 1M -E 128M -c2 -S 4M -E eof $DIR/$tdir/dir1 + + touch $DIR/$tdir/dir1/f1 + local layout1=$(get_layout_param $DIR/$tdir/dir1/f1) + + echo "Migrate $DIR/$tdir/dir1 from MDT1 to MDT0" + $LFS migrate -m 0 $DIR/$tdir/dir1 + + echo "trigger LFSCK for layout" + do_facet $SINGLEMDS $LCTL lfsck_start -M ${MDT_DEV} -t layout -r + + wait_update_facet $SINGLEMDS "$LCTL get_param -n \ + mdd.${MDT_DEV}.lfsck_layout | + awk '/^status/ { print \\\$2 }'" "completed" 32 || { + $SHOW_LAYOUT + error "(2) unexpected status" + } + + local layout2=$(get_layout_param $DIR/$tdir/dir1/f1) + + [[ "$layout1" == "$layout2" ]] || error "layout lost after lfsck" +} +run_test 40a "LFSCK correctly fixes lmm_oi in composite layout" # restore MDS/OST size MDSSIZE=${SAVED_MDSSIZE}