Create Oxygen jobs and bump Nitrogen branch
[releng/builder.git] / jjb / integration / include-raw-integration-start-cluster-run-test.sh
index 014e3b16d280d021b705a0f9ec372c8429410b52..97405ab8dc71492218064018574bec4aff90be7d 100644 (file)
@@ -2,6 +2,7 @@
 # Activate robotframework virtualenv
 # ${ROBOT_VENV} comes from the include-raw-integration-install-robotframework.sh
 # script.
+# shellcheck source=${ROBOT_VENV}/bin/activate disable=SC1091
 source ${ROBOT_VENV}/bin/activate
 
 echo "#################################################"
@@ -30,7 +31,7 @@ while true; do
         echo "Dumping last 500K bytes of karaf log..."
         tail --bytes=500K "/tmp/${BUNDLEFOLDER}/data/log/karaf.log"
         echo "Listing all open ports on controller system"
-        netstat -natu
+        netstat -pnatu
         exit 1
     else
         COUNT=\$(( \${COUNT} + 5 ))
@@ -40,7 +41,7 @@ while true; do
 done
 
 echo "Listing all open ports on controller system.."
-netstat -natu
+netstat -pnatu
 
 function exit_on_log_file_message {
     echo "looking for \"\$1\" in log file"
@@ -67,6 +68,11 @@ do
     ssh ${!CONTROLLERIP} "bash /tmp/verify-cluster-is-up.sh ${i} ${!CONTROLLERIP}"
 done
 
+if [ ${NUM_OPENSTACK_SYSTEM} -gt 0 ]; then
+   echo "Exiting without running tests to deploy openstack for testing"
+   exit
+fi
+
 if [ ${CONTROLLERSCOPE} == 'all' ]; then
     COOLDOWN_PERIOD="180"
 else
@@ -80,6 +86,9 @@ for i in `seq 1 ${NUM_ODL_SYSTEM}`
 do
     CONTROLLERIP=ODL_SYSTEM_${i}_IP
     odl_variables=${odl_variables}" -v ${CONTROLLERIP}:${!CONTROLLERIP}"
+    echo "Lets's take the karaf thread dump"
+    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_before.log || true
 done
 
 echo "Generating mininet variables..."
@@ -102,7 +111,8 @@ cat testplan.txt
 SUITES=`egrep -v '(^[[:space:]]*#|^[[:space:]]*$)' testplan.txt | tr '\012' ' '`
 
 echo "Starting Robot test suites ${SUITES} ..."
-pybot -N ${TESTPLAN} -c critical -e exclude -v BUNDLEFOLDER:${BUNDLEFOLDER} -v WORKSPACE:/tmp -v BUNDLE_URL:${ACTUALBUNDLEURL} \
+pybot -N ${TESTPLAN} --removekeywords wuks -c critical -e exclude -e skip_if_${DISTROSTREAM} -v BUNDLEFOLDER:${BUNDLEFOLDER} \
+-v WORKSPACE:/tmp -v BUNDLE_URL:${ACTUAL_BUNDLE_URL} -v JAVA_HOME:${JAVA_HOME} \
 -v NEXUSURL_PREFIX:${NEXUSURL_PREFIX} -v JDKVERSION:${JDKVERSION} -v ODL_STREAM:${DISTROSTREAM} \
 -v CONTROLLER:${ODL_SYSTEM_IP} -v CONTROLLER1:${ODL_SYSTEM_2_IP} -v CONTROLLER2:${ODL_SYSTEM_3_IP} -v ODL_SYSTEM_IP:${ODL_SYSTEM_IP} \
 ${odl_variables} -v NUM_ODL_SYSTEM:${NUM_ODL_SYSTEM} -v CONTROLLER_USER:${USER} -v ODL_SYSTEM_USER:${USER} -v \
@@ -111,14 +121,21 @@ TOOLS_SYSTEM_IP:${TOOLS_SYSTEM_IP} ${tools_variables} -v NUM_TOOLS_SYSTEM:${NUM_
 -v USER_HOME:${HOME} ${TESTOPTIONS} ${SUITES} || true
 # FIXME: Sort (at least -v) options alphabetically.
 
+echo "Examining the files in data/log and checking filesize"
+ssh ${ODL_SYSTEM_1_IP} "ls -altr /tmp/${BUNDLEFOLDER}/data/log/"
+ssh ${ODL_SYSTEM_1_IP} "du -hs /tmp/${BUNDLEFOLDER}/data/log/*"
+ssh ${ODL_SYSTEM_2_IP} "ls -altr /tmp/${BUNDLEFOLDER}/data/log/"
+ssh ${ODL_SYSTEM_2_IP} "du -hs /tmp/${BUNDLEFOLDER}/data/log/*"
+ssh ${ODL_SYSTEM_3_IP} "ls -altr /tmp/${BUNDLEFOLDER}/data/log/"
+ssh ${ODL_SYSTEM_3_IP} "du -hs /tmp/${BUNDLEFOLDER}/data/log/*"
+
 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 "dumping first 500K bytes of karaf log..." > "odl${i}_karaf.log"
-    ssh "${!CONTROLLERIP}" head --bytes=500K "/tmp/${BUNDLEFOLDER}/data/log/karaf.log" >> "odl${i}_karaf.log"
-    echo "dumping last 500K bytes of karaf log..." >> "odl${i}_karaf.log"
-    ssh "${!CONTROLLERIP}" tail --bytes=500K "/tmp/${BUNDLEFOLDER}/data/log/karaf.log" >> "odl${i}_karaf.log"
+    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
@@ -126,10 +143,21 @@ sleep 5
 for i in `seq 1 ${NUM_ODL_SYSTEM}`
 do
     CONTROLLERIP=ODL_SYSTEM_${i}_IP
-    ssh "${!CONTROLLERIP}" xz -9ekvv "/tmp/${BUNDLEFOLDER}/data/log/karaf.log"
-    scp "${!CONTROLLERIP}:/tmp/${BUNDLEFOLDER}/data/log/karaf.log.xz" "odl${i}_karaf.log.xz"
+    echo "Compressing karaf.log ${i}"
+    ssh ${!CONTROLLERIP} gzip --best /tmp/${BUNDLEFOLDER}/data/log/karaf.log
+    echo "Fetching compressed karaf.log ${i}"
+    scp "${!CONTROLLERIP}:/tmp/${BUNDLEFOLDER}/data/log/karaf.log.gz" "odl${i}_karaf.log.gz" && ssh ${!CONTROLLERIP} rm -f "/tmp/${BUNDLEFOLDER}/data/log/karaf.log.gz"
+    # TODO: Should we compress the output log file as well?
+    scp "${!CONTROLLERIP}:/tmp/${BUNDLEFOLDER}/data/log/karaf_console.log" "odl${i}_karaf_console.log" && ssh ${!CONTROLLERIP} rm -f "/tmp/${BUNDLEFOLDER}/data/log/karaf_console.log"
+    echo "Fetch GC logs"
+    # FIXME: Put member index in filename, instead of directory name.
+    mkdir -p "gclogs-${i}"
+    scp "${!CONTROLLERIP}:/tmp/${BUNDLEFOLDER}/data/log/*.log" "gclogs-${i}/" && ssh ${!CONTROLLERIP} rm -f "/tmp/${BUNDLEFOLDER}/data/log/*.log"
 done
+
+echo "Examine copied files"
+ls -lt
+
 true  # perhaps Jenkins is testing last exit code
 
 # vim: ts=4 sw=4 sts=4 et ft=sh :
-