--- /dev/null
+#!/bin/bash
+
+# client failure does not affect other clients
+
+# Start load on clients (each client works on it's own directory).
+# At defined (5-10 minutes) interval fail one random client and then fail mds.
+# Reintegrate failed client after recovery completed,
+# application errors are allowed for that client but not on other clients.
+# 10 minute intervals and verify that no application errors occur.
+
+# Test runs one of CLIENT_LOAD progs on remote clients.
+
+LUSTRE=${LUSTRE:-`dirname $0`/..}
+SETUP=${SETUP:-""}
+CLEANUP=${CLEANUP:-""}
+. $LUSTRE/tests/test-framework.sh
+
+init_test_env $@
+
+. ${CONFIG:=$LUSTRE/tests/cfg/$NAME.sh}
+
+TESTSUITELOG=${TESTSUITELOG:-$TMP/recovery-random-scale}
+DEBUGLOG=$TESTSUITELOG.debug
+exec 2>$DEBUGLOG
+echo "--- env ---" >&2
+env >&2
+echo "--- env ---" >&2
+set -x
+
+[ "$SHARED_DIRECTORY" ] || \
+ { skip "$0: Empty SHARED_DIRECTORY" && exit 0; }
+
+[ -n "$CLIENTS" ] || { skip "$0 Need two or more remote clients" && exit 0; }
+[ $CLIENTCOUNT -ge 3 ] || \
+ { skip "$0 Need two or more clients, have $CLIENTCOUNT" && exit 0; }
+
+END_RUN_FILE=${END_RUN_FILE:-$SHARED_DIRECTORY/end_run_file}
+LOAD_PID_FILE=${LOAD_PID_FILE:-$TMP/client-load.pid}
+
+remote_mds_nodsh && skip "remote MDS with nodsh" && exit 0
+
+build_test_filter
+
+check_and_setup_lustre
+rm -rf $DIR/[df][0-9]*
+
+# the test node needs to be insulated from a lustre failure as much as possible,
+# so not even loading the lustre modules is ideal.
+# -- umount lustre
+# -- remove hostname from clients list
+zconf_umount $(hostname) $MOUNT
+NODES_TO_USE=${NODES_TO_USE:-$CLIENTS}
+NODES_TO_USE=$(exclude_items_from_list $NODES_TO_USE $(hostname))
+
+check_progs_installed $NODES_TO_USE ${CLIENT_LOADS[@]}
+
+MDTS=$(get_facets MDS)
+
+if [ "$SLOW" = "no" ]; then
+ DURATION=${DURATION:-$((60 * 30))}
+ SERVER_FAILOVER_PERIOD=${SERVER_FAILOVER_PERIOD:-$((60 * 5))}
+else
+ DURATION=${DURATION:-$((60 * 60 * 24))}
+ SERVER_FAILOVER_PERIOD=${SERVER_FAILOVER_PERIOD:-$((60 * 10))} # 10 minutes
+fi
+
+rm -f $END_RUN_FILE
+
+vmstatLOG=${TESTSUITELOG}_$(basename $0 .sh).vmstat
+
+numfailovers () {
+ local facet
+ local var
+
+ for facet in $MDTS ${failed_clients//,/ }; do
+ var=${facet}_nums
+ val=${!var}
+ if [ "$val" ] ; then
+ echo "$facet failed over $val times"
+ fi
+ done
+}
+
+# list is comma separated
+print_logs () {
+ local list=$1
+
+ do_nodes $list "node=\\\$(hostname)
+var=\\\${node}_load
+log=${TESTSUITELOG}_run_${!var}.sh-\\\$node.debug
+if [ -e \\\$log ] ; then
+echo Node \\\$node debug log:
+cat \\\$log
+fi"
+}
+
+summary_and_cleanup () {
+
+ local rc=$?
+ local var
+ trap 0
+
+ # Having not empty END_RUN_FILE means the failed loads only
+ if [ -s $END_RUN_FILE ]; then
+ echo "Found the END_RUN_FILE file: $END_RUN_FILE"
+ cat $END_RUN_FILE
+ local END_RUN_NODE=
+ read END_RUN_NODE < $END_RUN_FILE
+
+ # a client load will end (i.e. fail) if it finds
+ # the end run file. that does not mean that that client load
+ # actually failed though. the first node in the END_RUN_NODE is
+ # the one we are really interested in.
+ if [ -n "$END_RUN_NODE" ]; then
+ var=${END_RUN_NODE}_load
+ echo "Client load failed on node $END_RUN_NODE"
+ echo
+ echo "client $END_RUN_NODE load stdout and debug files :
+ ${TESTSUITELOG}_run_${!var}.sh-${END_RUN_NODE}
+ ${TESTSUITELOG}_run_${!var}.sh-${END_RUN_NODE}.debug"
+ fi
+ rc=1
+ fi
+
+
+ echo $(date +'%F %H:%M:%S') Terminating clients loads ...
+ echo "$0" >> $END_RUN_FILE
+ local result=PASS
+ [ $rc -eq 0 ] || result=FAIL
+
+ log "Duraion: $DURATION
+Server failover period: $SERVER_FAILOVER_PERIOD seconds
+Exited after: $ELAPSED seconds
+Number of failovers before exit:
+$(numfailovers)
+Status: $result: rc=$rc"
+
+ # stop the vmstats on the OSTs
+ if [ "$VMSTAT" ]; then
+ do_nodes $(comma_list $(osts_nodes)) "test -f /tmp/vmstat.pid && \
+ { kill -s TERM \$(cat /tmp/vmstat.pid); rm -f /tmp/vmstat.pid; \
+ gzip -f9 $vmstatLOG-\$(hostname); }"
+ fi
+
+ # make sure the client loads die
+ do_nodes $NODES_TO_USE "set -x; test -f $LOAD_PID_FILE && \
+ { kill -s TERM \$(cat $LOAD_PID_FILE) || true; }"
+
+ # and free up the pdshes that started them, if any are still around
+ if [ -n "$CLIENT_LOAD_PIDS" ]; then
+ kill $CLIENT_LOAD_PIDS || true
+ sleep 5
+ kill -9 $CLIENT_LOAD_PIDS || true
+ fi
+
+ if [ $rc -ne 0 ]; then
+ print_logs $NODES_TO_USE
+ fi
+
+ if [ $rc -eq 0 ]; then
+ zconf_mount $(hostname) $MOUNT
+ else
+ error "exited with rc=$rc"
+ fi
+ exit $rc
+}
+
+#
+# MAIN
+#
+log "-----============= $0 starting =============-----"
+
+trap summary_and_cleanup EXIT # INT
+
+ELAPSED=0
+
+# vmstat the osts
+if [ "$VMSTAT" ]; then
+ do_nodes $(comma_list $(osts_nodes)) "vmstat 1 > $vmstatLOG-\$(hostname) 2>/dev/null </dev/null & echo \$! > /tmp/vmstat.pid"
+fi
+
+# Start client loads.
+start_client_loads $NODES_TO_USE
+
+echo clients load pids:
+if ! do_nodes $NODES_TO_USE "set -x; echo \$(hostname): && cat $LOAD_PID_FILE"; then
+ if [ -e $DEBUGLOG ]; then
+ exec 2<&-
+ cat $DEBUGLOG
+ exit 3
+ fi
+fi
+
+START_TS=$(date +%s)
+CURRENT_TS=$START_TS
+
+MINSLEEP=${MINSLEEP:-120}
+REQFAIL_PERCENT=${REQFAIL_PERCENT:-3} # bug17839 comment 62
+REQFAIL=${REQFAIL:-$(( DURATION / SERVER_FAILOVER_PERIOD * REQFAIL_PERCENT / 100))}
+reqfail=0
+sleep=0
+
+# This is used for FAIL_CLIENT only
+ERRORS_OK="yes"
+while [ $ELAPSED -lt $DURATION -a ! -e $END_RUN_FILE ]; do
+
+ # In order to perform the
+ # expected number of failovers, we need to account the following :
+ # 1) the time that has elapsed during the client load checking
+ # 2) time takes for failover
+
+ it_time_start=$(date +%s)
+
+ FAIL_CLIENT=$(get_random_entry $NODES_TO_USE)
+ client_var=${FAIL_CLIENT}_nums
+
+ # store the list of failed clients
+ # lists are comma separated
+ failed_clients=$(expand_list $failed_clients $FAIL_CLIENT)
+
+ SERVERFACET=$(get_random_entry $MDTS)
+ var=${SERVERFACET}_nums
+
+ # Check that our client loads are still running. If any have died,
+ # that means they have died outside of recovery, which is unacceptable.
+
+ log "==== Checking the clients loads BEFORE failover -- failure NOT OK \
+ ELAPSED=$ELAPSED DURATION=$DURATION PERIOD=$SERVER_FAILOVER_PERIOD"
+
+ if ! check_client_loads $NODES_TO_USE; then
+ exit 4
+ fi
+
+ log "FAIL CLIENT $FAIL_CLIENT ... "
+ shutdown_client $FAIL_CLIENT
+
+ log "Starting failover on $SERVERFACET"
+
+ facet_failover "$SERVERFACET" || exit 1
+ if ! wait_recovery_complete $SERVERFACET $((TIMEOUT * 10)); then
+ echo "$SERVERFACET recovery is not completed!"
+ exit 7
+ fi
+
+ boot_node $FAIL_CLIENT
+ echo "Reintegrating $FAIL_CLIENT"
+ zconf_mount $FAIL_CLIENT $MOUNT || exit $?
+
+ # Increment the number of failovers
+ val=$((${!var} + 1))
+ eval $var=$val
+ val=$((${!client_var} + 1))
+ eval $client_var=$val
+
+ # load script on failed clients could create END_RUN_FILE
+ # We shuold remove it and ignore the failure if this
+ # file contains the failed client only.
+ # We can not use ERRORS_OK when start all loads at the start of this script
+ # because the application errors allowed for random failed client only, but
+ # not for all clients.
+ if [ -e $END_RUN_FILE ]; then
+ read END_RUN_NODE < $END_RUN_FILE
+ [[ $END_RUN_NODE = $FAIL_CLIENT ]] &&
+ rm -f $END_RUN_FILE || exit 13
+ fi
+
+ restart_client_loads $FAIL_CLIENT $ERRORS_OK || exit $?
+
+ # Check that not failed clients loads are still running.
+ # No application failures should occur on clients that was not failed.
+
+ log "==== Checking the clients loads AFTER failed client reintegrated -- failure NOT OK"
+ if ! ERRORS_OK= check_client_loads $(exclude_items_from_list $NODES_TO_USE $FAIL_CLIENT); then
+ log "Client load failed. Exiting"
+ exit 5
+ fi
+
+ CURRENT_TS=$(date +%s)
+ ELAPSED=$((CURRENT_TS - START_TS))
+
+ sleep=$((SERVER_FAILOVER_PERIOD-(CURRENT_TS - it_time_start)))
+
+ # keep count the number of itterations when
+ # time spend to failover and two client loads check exceeded
+ # the value ( SERVER_FAILOVER_PERIOD - MINSLEEP )
+ if [ $sleep -lt $MINSLEEP ]; then
+ reqfail=$((reqfail +1))
+ log "WARNING: failover, client reintegration and check_client_loads time
+exceeded SERVER_FAILOVER_PERIOD - MINSLEEP !
+Failed to meet interval $reqfail times ( REQFAIL=$REQFAIL ); have sleep=$sleep"
+ [ $reqfail -gt $REQFAIL ] && exit 6
+ fi
+
+ log " Number of failovers:
+$(numfailovers) and counting..."
+
+ if [ $sleep -gt 0 ]; then
+ echo "sleeping $sleep seconds ... "
+ sleep $sleep
+ fi
+done
+
+exit 0
git://git.whamcloud.com / fs / lustre-release.git / commitdiff