Details
-
Bug
-
Resolution: Fixed
-
Minor
-
None
-
None
-
Muli-node Lustre test environment.
-
3
-
9223372036854775807
Description
Sanity test 807 stuck on multi-nodes test environment when CLIENTS is empty.
[root@lustre-aio tests]# PTLDEBUG=-1 ./auster -vr sanity --only 807 2... 3 4== sanity test 807: verify LSOM syncing tool ============= 02:19:11 (1641349151) 5mdd.lustre-MDT0000.changelog_mask=+hsm 6mdd.lustre-MDT0001.changelog_mask=+hsm 7Registered 2 changelog users: 'cl1 cl1' 8llite.lustre-fffff7ba8ec63000.xattr_cache=0 91+0 records in 101+0 records out 111048576 bytes (1.0 MB, 1.0 MiB) copied, 0.101602 s, 10.3 MB/s 12Test SOM for multi-client (0) writes 13No such rcmd module "(PATH=$PATH" 14pdsh@lustre-aio: Failed to register rcmd "(PATH=$PATH" for "/lib64/lustre/utils:/lib64/lustre/tests:/sbin:/usr/sbin; cd /lib64/lustre/tests; LUSTRE="/lib64/lustre" mds1_FSTYPE=ldiskfs ost1_FSTYPE=ldiskfs VERBOSE=false FSTYPE=ldiskfs NETTYPE=tcp sh -c "sync ; sleep 5 ; sync")" 15Start to sync 3 records. 16lustre-MDT0000: clear the changelog for cl1 of all records 17lustre-MDT0000: Deregistered changelog user #1 18 19...stucking...
But defining var “CLIENTS=$(hostname)” in cfg/local.sh, the test case passes.
sanity test 807 source code
27113 test_807() { 27114 [ -n "$FILESET" ] && skip "Not functional for FILESET set" 27115 [ $MDS1_VERSION -lt $(version_code 2.11.52) ] && 27116 skip "Need MDS version at least 2.11.52" 27117 27118 # Registration step 27119 changelog_register || error "changelog_register failed" 27120 local cl_user="${CL_USERS[$SINGLEMDS]%% *}" 27121 changelog_users $SINGLEMDS | grep -q $cl_user || 27122 error "User $cl_user not found in changelog_users" 27123 27124 local save="$TMP/$TESTSUITE-$TESTNAME.parameters" 27125 save_lustre_params client "llite.*.xattr_cache" > $save 27126 lctl set_param llite.*.xattr_cache=0 27127 stack_trap "restore_lustre_params < $save; rm -f $save" EXIT 27128 27129 rm -rf $DIR/$tdir || error "rm $tdir failed" 27130 mkdir_on_mdt0 $DIR/$tdir || error "mkdir $tdir failed" 27131 touch $DIR/$tdir/trunc || error "touch $tdir/trunc failed" 27132 $TRUNCATE $DIR/$tdir/trunc 1024 || error "truncate $tdir/trunc failed" 27133 $TRUNCATE $DIR/$tdir/trunc 1048576 || 27134 error "truncate $tdir/trunc failed" 27135 27136 local bs=1048576 27137 dd if=/dev/zero of=$DIR/$tdir/single_dd bs=$bs count=1 conv=fsync || 27138 error "write $tfile failed" 27139 27140 # multi-client wirtes 27141 local num=$(get_node_count ${CLIENTS//,/ }) 27142 local offset=0 27143 local i=0 27144 27145 echo "Test SOM for multi-client ($num) writes" 27146 touch $DIR/$tfile || error "touch $tfile failed" 27147 $TRUNCATE $DIR/$tfile 0 27148 for client in ${CLIENTS//,/ }; do 27149 do_node $client $MULTIOP $DIR/$tfile Oz${offset}w${bs}c & 27150 local pids[$i]=$! 27151 i=$((i + 1)) 27152 offset=$((offset + $bs)) 27153 done 27154 for (( i=0; i < $num; i++ )); do 27155 wait ${pids[$i]} 27156 done 27157 27158 do_rpc_nodes "$CLIENTS" cancel_lru_locks osc 27159 do_nodes "$CLIENTS" "sync ; sleep 5 ; sync" 27160 $LSOM_SYNC -u $cl_user -m $FSNAME-MDT0000 $MOUNT 27161 check_lsom_data $DIR/$tdir/trunc 27162 check_lsom_data $DIR/$tdir/single_dd 27163 check_lsom_data $DIR/$tfile 27164 27165 rm -rf $DIR/$tdir 27166 # Deregistration step 27167 changelog_deregister || error "changelog_deregister failed" 27168 } 27169 run_test 807 "verify LSOM syncing tool"