-#@IgnoreInspection BashAddShebang
+#!/bin/bash
# Activate robotframework virtualenv
# ${ROBOT_VENV} comes from the integration-install-robotframework.sh
# script.
[ml2]
# Needed for VLAN provider tests - because our provider networks are always encapsulated in VXLAN (br-physnet1)
-# MTU(1440) + VXLAN(50) + VLAN(4) = 1494 < MTU eth0/br-physnet1(1500)
-physical_network_mtus = ${PUBLIC_PHYSICAL_NETWORK}:1440
-path_mtu = 1490
+# MTU(1400) + VXLAN(50) + VLAN(4) = 1454 < MTU eth0/br-physnet1(1458)
+physical_network_mtus = ${PUBLIC_PHYSICAL_NETWORK}:1400
+path_mtu = 1458
# workaround for port-status not working due to https://bugs.opendaylight.org/show_bug.cgi?id=9092
[ml2_odl]
${SSH} ${ip} "sudo find /etc > /tmp/find.etc.txt"
${SSH} ${ip} "sudo find /opt/stack > /tmp/find.opt.stack.txt"
${SSH} ${ip} "sudo find /var > /tmp/find2.txt"
- rsync --rsync-path="sudo rsync" -arv --list-only ssh ${ip}:/var/ > ${folder}/rsync.var.txt
+ ${SSH} ${ip} "sudo find /var > /tmp/find.var.txt"
+ rsync --rsync-path="sudo rsync" --list-only -arvhe ssh ${ip}:/etc/ > ${folder}/rsync.etc.txt
+ rsync --rsync-path="sudo rsync" --list-only -arvhe ssh ${ip}:/opt/stack/ > ${folder}/rsync.opt.stack.txt
+ rsync --rsync-path="sudo rsync" --list-only -arvhe ssh ${ip}:/var/ > ${folder}/rsync.var.txt
scp ${ip}:/tmp/find.etc.txt ${folder}
scp ${ip}:/tmp/find.opt.stack.txt ${folder}
scp ${ip}:/tmp/find2.txt ${folder}
+ scp ${ip}:/tmp/find.var.txt ${folder}
}
function collect_logs () {
set +e # We do not want to create red dot just because something went wrong while fetching logs.
- for i in `seq 1 ${NUM_ODL_SYSTEM}`; do
- CONTROLLERIP=ODL_SYSTEM_${i}_IP
- echo "Lets's take the karaf thread dump again..."
- KARAF_PID=$(ssh ${!CONTROLLERIP} "ps aux | grep ${KARAF_ARTIFACT} | grep -v grep | tr -s ' ' | cut -f2 -d' '")
- ssh ${!CONTROLLERIP} "jstack $KARAF_PID"> ${WORKSPACE}/karaf_${i}_threads_after.log || true
- echo "killing karaf process..."
- ${SSH} "${!CONTROLLERIP}" bash -c 'ps axf | grep karaf | grep -v grep | awk '"'"'{print "kill -9 " $1}'"'"' | sh'
- done
-
cat > extra_debug.sh << EOF
echo -e "/usr/sbin/lsmod | /usr/bin/grep openvswitch\n"
/usr/sbin/lsmod | /usr/bin/grep openvswitch
env
EOF
+ # Since this log collection work is happening before the archive build macro which also
+ # creates the ${WORKSPACE}/archives dir, we have to do it here first. The mkdir in the
+ # archives build step will essentially be a noop.
+ mkdir -p ${WORKSPACE}/archives
+
sleep 5
# FIXME: Do not create .tar and gzip before copying.
for i in `seq 1 ${NUM_ODL_SYSTEM}`; do
CONTROLLERIP=ODL_SYSTEM_${i}_IP
- ${SSH} "${!CONTROLLERIP}" "cp -r /tmp/${BUNDLEFOLDER}/data/log /tmp/odl_log"
- ${SSH} "${!CONTROLLERIP}" "tar -cf /tmp/odl${i}_karaf.log.tar /tmp/odl_log/*"
- scp "${!CONTROLLERIP}:/tmp/odl${i}_karaf.log.tar" "${WORKSPACE}/odl${i}_karaf.log.tar"
- ${SSH} "${!CONTROLLERIP}" "tar -cf /tmp/odl${i}_zrpcd.log.tar /tmp/zrpcd.init.log"
- scp "${!CONTROLLERIP}:/tmp/odl${i}_zrpcd.log.tar" "${WORKSPACE}/odl${i}_zrpcd.log.tar"
- tar -xvf ${WORKSPACE}/odl${i}_karaf.log.tar -C . --strip-components 2 --transform s/karaf/odl${i}_karaf/g
- grep "ROBOT MESSAGE\| ERROR " odl${i}_karaf.log > odl${i}_err.log
- # Print ROBOT lines, Print Caused by...Exception: lines,
- # Print Exception{ lines as well as the previous line that has the timestamp for context
- sed -n -e '/ROBOT MESSAGE/P' -e '/Caused by.*Exception:/P' -e '$!N;/Exception:/P;D' -e '$!N;/Exception{/P;D' odl${i}_karaf.log > odl${i}_exception.log
- grep "ROBOT MESSAGE\| ERROR \| WARN \|Exception" odl${i}_karaf.log > odl${i}_err_warn_exception.log
- rm ${WORKSPACE}/odl${i}_karaf.log.tar
+ echo "collect_logs: for opendaylight controller ip: ${!CONTROLLERIP}"
+ NODE_FOLDER="odl_${i}"
+ mkdir -p ${NODE_FOLDER}
+ echo "Lets's take the karaf thread dump again..."
+ ssh ${!CONTROLLERIP} "sudo ps aux" > ${WORKSPACE}/ps_after.log
+ pid=$(grep org.apache.karaf.main.Main ${WORKSPACE}/ps_after.log | grep -v grep | tr -s ' ' | cut -f2 -d' ')
+ echo "karaf main: org.apache.karaf.main.Main, pid:${pid}"
+ ssh ${!CONTROLLERIP} "jstack ${pid}" > ${WORKSPACE}/karaf_${i}_${pid}_threads_after.log || true
+ echo "killing karaf process..."
+ ${SSH} "${!CONTROLLERIP}" bash -c 'ps axf | grep karaf | grep -v grep | awk '"'"'{print "kill -9 " $1}'"'"' | sh'
+ ${SSH} ${!CONTROLLERIP} "sudo journalctl > /tmp/journalctl.log"
+ scp ${!CONTROLLERIP}:/tmp/journalctl.log ${NODE_FOLDER}
+ ${SSH} ${!CONTROLLERIP} "dmesg -T > /tmp/dmesg.log"
+ scp ${!CONTROLLERIP}:/tmp/dmesg.log ${NODE_FOLDER}
+ ${SSH} ${!CONTROLLERIP} "cp -r /tmp/${BUNDLEFOLDER}/data/log /tmp/odl_log"
+ ${SSH} ${!CONTROLLERIP} "tar -cf /tmp/odl${i}_karaf.log.tar /tmp/odl_log/*"
+ scp ${!CONTROLLERIP}:/tmp/odl${i}_karaf.log.tar ${NODE_FOLDER}
+ ${SSH} ${!CONTROLLERIP} "tar -cf /tmp/odl${i}_zrpcd.log.tar /tmp/zrpcd.init.log"
+ scp ${!CONTROLLERIP}:/tmp/odl${i}_zrpcd.log.tar ${NODE_FOLDER}
+ tar -xvf ${NODE_FOLDER}/odl${i}_karaf.log.tar -C ${NODE_FOLDER} --strip-components 2 --transform s/karaf/odl${i}_karaf/g
+ grep "ROBOT MESSAGE\| ERROR " ${NODE_FOLDER}/odl${i}_karaf.log > ${NODE_FOLDER}/odl${i}_err.log
+ grep "ROBOT MESSAGE\| ERROR \| WARN \|Exception" \
+ ${NODE_FOLDER}/odl${i}_karaf.log > ${NODE_FOLDER}/odl${i}_err_warn_exception.log
+ # Print ROBOT lines and print Exception lines. For exception lines also print the previous line for context
+ sed -n -e '/ROBOT MESSAGE/P' -e '$!N;/Exception/P;D' ${NODE_FOLDER}/odl${i}_karaf.log > ${NODE_FOLDER}/odl${i}_exception.log
+ rm ${NODE_FOLDER}/odl${i}_karaf.log.tar
+ mv *_threads* ${NODE_FOLDER}
+ mv ps_* ${NODE_FOLDER}
+ mv ${NODE_FOLDER} ${WORKSPACE}/archives/
done
- # Since this log collection work is happening before the archive build macro which also
- # creates the ${WORKSPACE}/archives dir, we have to do it here first. The mkdir in the
- # archives build step will essentially be a noop.
- mkdir -p ${WORKSPACE}/archives
-
print_job_parameters > ${WORKSPACE}/archives/params.txt
# Control Node
for i in `seq 1 ${NUM_OPENSTACK_CONTROL_NODES}`; do
OSIP=OPENSTACK_CONTROL_NODE_${i}_IP
+ echo "collect_logs: for openstack control node ip: ${!OSIP}"
NODE_FOLDER="control_${i}"
mkdir -p ${NODE_FOLDER}
+ scp ${!OSIP}:/etc/dnsmasq.conf ${NODE_FOLDER}
scp ${!OSIP}:/etc/keystone/keystone.conf ${NODE_FOLDER}
scp ${!OSIP}:/etc/keystone/keystone-uwsgi-admin.ini ${NODE_FOLDER}
scp ${!OSIP}:/etc/keystone/keystone-uwsgi-public.ini ${NODE_FOLDER}
scp ${!OSIP}:/opt/stack/requirements/upper-constraints.txt ${NODE_FOLDER}
scp ${!OSIP}:/opt/stack/tempest/etc/tempest.conf ${NODE_FOLDER}
scp ${!OSIP}:/tmp/get_devstack.sh.txt ${NODE_FOLDER}
- scp ${!OSIP}:/var/log/httpd/keystone_access.log ${NODE_FOLDER}
- scp ${!OSIP}:/var/log/httpd/keystone.log ${NODE_FOLDER}
scp ${!OSIP}:/var/log/openvswitch/ovs-vswitchd.log ${NODE_FOLDER}
scp ${!OSIP}:/var/log/openvswitch/ovsdb-server.log ${NODE_FOLDER}
list_files "${!OSIP}" "${NODE_FOLDER}"
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/etc/hosts ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/usr/lib/systemd/system/haproxy.service ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/audit/audit.log ${NODE_FOLDER}
- rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/dmesg.log ${NODE_FOLDER}
+ rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/httpd/keystone_access.log ${NODE_FOLDER}
+ rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/httpd/keystone.log ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/messages ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/rabbitmq ${NODE_FOLDER}
rsync -avhe ssh ${!OSIP}:/opt/stack/logs/* ${NODE_FOLDER} # rsync to prevent copying of symbolic links
scp ${!OSIP}:/tmp/extra_debug.log ${NODE_FOLDER}
scp ${!OSIP}:/tmp/journalctl.log ${NODE_FOLDER}
scp ${!OSIP}:/tmp/*.xz ${NODE_FOLDER}
+ ${SSH} ${!CONTROLLERIP} "dmesg -T > /tmp/dmesg.log"
+ scp ${!CONTROLLERIP}:/tmp/dmesg.log ${NODE_FOLDER}
mv local.conf_control_${!OSIP} ${NODE_FOLDER}/local.conf
+ mv /tmp/qdhcp ${NODE_FOLDER}
mv ${NODE_FOLDER} ${WORKSPACE}/archives/
done
# Compute Nodes
for i in `seq 1 ${NUM_OPENSTACK_COMPUTE_NODES}`; do
OSIP=OPENSTACK_COMPUTE_NODE_${i}_IP
+ echo "collect_logs: for openstack compute node ip: ${!OSIP}"
NODE_FOLDER="compute_${i}"
mkdir -p ${NODE_FOLDER}
scp ${!OSIP}:/etc/nova/nova.conf ${NODE_FOLDER}
list_files "${!OSIP}" "${NODE_FOLDER}"
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/etc/hosts ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/audit/audit.log ${NODE_FOLDER}
- rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/dmesg.log ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/libvirt ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/messages ${NODE_FOLDER}
rsync --rsync-path="sudo rsync" -avhe ssh ${!OSIP}:/var/log/nova-agent.log ${NODE_FOLDER}
scp ${!OSIP}:/tmp/extra_debug.log ${NODE_FOLDER}
scp ${!OSIP}:/tmp/journalctl.log ${NODE_FOLDER}
scp ${!OSIP}:/tmp/*.xz ${NODE_FOLDER}/
+ ${SSH} ${!OSIP} "dmesg -T > /tmp/dmesg.log"
+ scp ${!OSIP}:/tmp/dmesg.log ${NODE_FOLDER}
mv local.conf_compute_${!OSIP} ${NODE_FOLDER}/local.conf
mv ${NODE_FOLDER} ${WORKSPACE}/archives/
done
set -e
}
+# Check if rabbitmq is ready by looking for a pid in it's status.
+# The function returns the status of the grep command which callers can check.
+function is_rabbitmq_ready() {
+ local -r ip=$1
+ rm -f rabbit.txt
+ ${SSH} ${ip} "sudo rabbitmqctl status" > rabbit.txt
+ grep pid rabbit.txt
+}
+
+# retry the given command ($3) until success for a number of iterations ($1)
+# sleeping ($2) between tries.
+function retry() {
+ set +e
+ local -r -i max_tries=${1}
+ local -r -i sleep_time=${2}
+ local -r cmd=${3}
+ local -i retries=1
+ local -i rc=1
+ while true; do
+ echo "retry ${cmd}: attempt: ${retries}"
+ ${cmd}
+ rc=$?
+ if ((${rc} == 0)); then
+ break;
+ else
+ if ((${retries} == ${max_tries})); then
+ break
+ else
+ ((retries++))
+ sleep ${sleep_time}
+ fi
+ fi
+ done
+ set -e
+ return ${rc}
+}
+
# if we are using the new netvirt impl, as determined by the feature name
# odl-netvirt-openstack (note: old impl is odl-ovsdb-openstack) then we
# want PROVIDER_MAPPINGS to be used -- this should be fixed if we want to support
# AccessRefused: (0, 0): (403) ACCESS_REFUSED - Login was refused using authentication mechanism AMQPLAIN. For details see the broker logfile.
# Compare that timestamp to this log in the control stack.log: sudo rabbitmqctl set_permissions -p nova_cell1 stackrabbit
# If the n-cpu.log is earlier than the control stack.log timestamp then the failure condition is likely hit.
-# TODO: modify devstack to wait for rabbitmq to be available on the controller before starting nova-compute.
-echo "Sleeping for 360s to allow controller to create nova_cell1 before the computes need it"
-sleep 360
+WAIT_FOR_RABBITMQ_MINUTES=60
+echo "Wait a maximum of ${WAIT_FOR_RABBITMQ_MINUTES}m until rabbitmq is ready to allow the controller to create nova_cell1 before the computes need it"
+retry ${WAIT_FOR_RABBITMQ_MINUTES} 60 "is_rabbitmq_ready ${OPENSTACK_CONTROL_NODE_1_IP}"
+rc=$?
+if ((${rc} == 0)); then
+ echo "rabbitmq is ready, starting ${NUM_OPENSTACK_COMPUTE_NODES} compute(s)"
+else
+ echo "rabbitmq was not ready in ${WAIT_FOR_RABBITMQ_MINUTES}m"
+ collect_logs
+ exit 1
+fi
for i in `seq 1 ${NUM_OPENSTACK_COMPUTE_NODES}`; do
NUM_COMPUTES_PER_SITE=$((NUM_OPENSTACK_COMPUTE_NODES / NUM_OPENSTACK_SITES))
sudo ip netns exec pnf_ns ifconfig pnf_veth1 up ${EXTNET_PNF_IP}/24;
sudo ovs-vsctl add-port ${PUBLIC_BRIDGE} pnf_veth0;
"
+ # Control Node - set VXLAN TEP IP for Genius Auto TZ
+ ${SSH} ${!CONTROLIP} "
+ sudo ovs-vsctl set O . external_ids:tep-ip=${!CONTROLIP};
+ "
# Control Node - external net internet address simulation
${SSH} ${!CONTROLIP} "
${SSH} $compute_ip "
sudo ovs-vsctl add-port $PUBLIC_BRIDGE $CONTROLPORT -- set interface $CONTROLPORT type=vxlan options:local_ip=$compute_ip options:remote_ip=${!CONTROLIP} options:dst_port=9876 options:key=flow
"
+ #Compute Node - set VXLAN TEP IP for Genius Auto TZ
+ ${SSH} $compute_ip "
+ sudo ovs-vsctl set O . external_ids:tep-ip=${compute_ip};
+ "
done
done
# Use the testplan if specific SUITES are not defined.
if [ -z "${SUITES}" ]; then
SUITES=`egrep -v '(^[[:space:]]*#|^[[:space:]]*$)' testplan.txt | tr '\012' ' '`
+else
+ newsuites=""
+ workpath="${WORKSPACE}/test/csit/suites"
+ for suite in ${SUITES}; do
+ fullsuite="${workpath}/${suite}"
+ if [ -z ${newsuites} ]; then
+ newsuites+=${fullsuite}
+ else
+ newsuites+=" "${fullsuite}
+ fi
+ done
+ SUITES=${newsuites}
fi
# TODO: run openrc on control node and then scrape the vars from it
echo "Starting Robot test suites ${SUITES} ..."
# please add pybot -v arguments on a single line and alphabetized
-pybot -N ${TESTPLAN} --removekeywords wuks --flattenkeywords for -c critical -e exclude -e skip_if_${DISTROSTREAM} \
+suite_num=0
+for suite in ${SUITES}; do
+ # prepend a incrmental counter to the suite name so that the full robot log combining all the suites as is done
+ # in the rebot step below will list all the suites in chronological order as rebot seems to alphabatize them
+ let "suite_num = suite_num + 1"
+ suite_index="$(printf %02d ${suite_num})"
+ suite_name="$(basename ${suite} | cut -d. -f1)"
+ log_name="${suite_index}_${suite_name}"
+ pybot -N ${log_name} --removekeywords wuks -c critical -e exclude -e skip_if_${DISTROSTREAM} \
+ --log log_${log_name}.html --report None --output output_${log_name}.xml \
-v BUNDLEFOLDER:${BUNDLEFOLDER} \
-v BUNDLE_URL:${ACTUAL_BUNDLE_URL} \
-v CONTROLLER_USER:${USER} \
-v TOOLS_SYSTEM_2_IP:${TOOLS_SYSTEM_2_IP} \
-v USER_HOME:${HOME} \
-v WORKSPACE:/tmp \
- ${TESTOPTIONS} ${SUITES} || true
+ ${TESTOPTIONS} ${suite} || true
+done
+#rebot exit codes seem to be different
+rebot --output ${WORKSPACE}/output.xml --log log_full.html --report None -N openstack output_*.xml || true
echo "Examining the files in data/log and checking file size"
ssh ${ODL_SYSTEM_IP} "ls -altr /tmp/${BUNDLEFOLDER}/data/log/"
Code Review / releng / builder.git / blobdiff