clearer names for actions, and infer actions better

[monitor.git] / automate-default.sh

diff --git a/automate-default.sh b/automate-default.sh
index d01650e..d691c66 100755 (executable)
--- a/automate-default.sh
+++ b/automate-default.sh
@@ -1,21 +1,33 @@
 #!/bin/bash
 
+export PATH=$PATH:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin
+
 # NOTE: Must be an absolute path to guarantee it is read.
 INSTALLPATH=/usr/share/monitor/
-# Generate an 'sh' style file full of variables in monitor.conf
-$INSTALLPATH/shconfig.py >  $INSTALLPATH/monitorconfig.sh
+$INSTALLPATH/commands/shconfig.py >  $INSTALLPATH/monitorconfig.sh
 source $INSTALLPATH/monitorconfig.sh
 cd ${MONITOR_SCRIPT_ROOT}
 set -e
 DATE=`date +%Y-%m-%d-%T`
 MONITOR_PID="${MONITOR_SCRIPT_ROOT}/SKIP"
 
+function send_mail ()
+{
+    subject=$1
+    body=$2
+    mail -s "$subject" $exception_email <<EOF
+$body
+EOF
+}
+
+
 echo "#######################################"; echo "Running Monitor at $DATE"; echo "######################################"
 echo "Performing API test"
-API=$(./testapi.py)
+API=$(${MONITOR_SCRIPT_ROOT}/tools/testapi.py)
 if [ "$API" != "ok" ] ; then 
        # NOTE: Do not try to run any commands if the API is obviously broken.
        echo "API IS DOWN : "`date`
+    send_mail "API IS DOWN: canceled monitor run for `date`" "have a nice day..."
        exit 1
 fi
 
@@ -25,7 +37,7 @@ if [ -f $MONITOR_PID ] ; then
                PID=`cat $MONITOR_PID`
                rm -f $MONITOR_PID
                if [ -z $PID ] ; then
-                       ${MONITOR_SCRIPT_ROOT}/kill.cmd.sh $PID
+                       ${MONITOR_SCRIPT_ROOT}/tools/kill.cmd.sh $PID
                        echo "done."
                else
                        echo "No PID to be killed."
@@ -54,49 +66,51 @@ if [ -z "$AGENT" ] ; then
         # if no agent is running, set it up.
         ssh-agent > ${MONITOR_SCRIPT_ROOT}/agent.sh
         source ${MONITOR_SCRIPT_ROOT}/agent.sh
+        ssh-add /etc/planetlab/myops_ssh_key.rsa
         ssh-add /etc/planetlab/debug_ssh_key.rsa
         ssh-add /etc/planetlab/root_ssh_key.rsa
 fi
 #TODO: should add a call to ssh-add -l to check if the keys are loaded or not.
 source ${MONITOR_SCRIPT_ROOT}/agent.sh
 
+# CHECK AGENT IS UP AND RUNNING
+count=$( ssh-add -l | wc -l ) 
+if [ $count -lt 3 ] ; then
+    send_mail "ssh-agent is not up and running." "Add keys before monitoring can continue"
+       exit
+fi
 
-echo "Performing Findbad Nodes"
+${MONITOR_SCRIPT_ROOT}/commands/syncwithplc.py $DATE || :
+service plc restart monitor
+
+echo "Performing FindAll Nodes"
 #########################
 # 1. FINDBAD NODES 
-${MONITOR_SCRIPT_ROOT}/findbad.py --increment $DATE || :
+${MONITOR_SCRIPT_ROOT}/commands/findall.py $DATE || :
 ps ax | grep BatchMode | grep -v grep | awk '{print $1}' | xargs -r kill || :
-
-echo "Performing Findbad PCUs"
-#########################
-# 2. FINDBAD PCUS
-${MONITOR_SCRIPT_ROOT}/findbadpcu.py --increment $DATE || :
 # clean up stray 'locfg' processes that hang around inappropriately...
 ps ax | grep locfg | grep -v grep | awk '{print $1}' | xargs -r kill || :
 
-echo "Performing uptime changes for sites, nodes, and pcus"
-########################
-# 3. record last-changed for sites, nodes and pcus.
-${MONITOR_SCRIPT_ROOT}/sitebad.py --increment || :
-${MONITOR_SCRIPT_ROOT}/nodebad.py --increment || :
-${MONITOR_SCRIPT_ROOT}/pcubad.py --increment || :
 
-echo "Archiving pkl files"
-#########################
-# Archive pkl files.
-for f in findbad act_all findbadpcus l_plcnodes site_persistflags node_persistflags pcu_persistflags ; do
-       if [ -f ${MONITOR_DATA_ROOT}/production.$f.pkl ] ; then
-               cp ${MONITOR_DATA_ROOT}/production.$f.pkl ${MONITOR_ARCHIVE_ROOT}/`date +%F-%H:%M`.production.$f.pkl
-       else
-               echo "Warning: It failed to archive ${MONITOR_DATA_ROOT}/production.$f.pkl"
-       fi
-done
+${MONITOR_SCRIPT_ROOT}/commands/repair.py $DATE || :
+${MONITOR_SCRIPT_ROOT}/commands/policy.py $DATE || :
+curl -s 'http://summer.cs.princeton.edu/status/tabulator.cgi?table=table_nodeview&formatcsv' > /var/lib/monitor/comon/$DATE.comon.csv || :
 
-#echo "Running grouprins on all dbg nodes"
-############################
-# 5. Check if there are any nodes in dbg state.  Clean up afterward.
-#${MONITOR_SCRIPT_ROOT}/grouprins.py --mail=1 --reboot --nodeselect 'state=DOWN&&boot_state=(boot|rins|dbg|diag)' --stopselect "state=BOOT" || :
-#${MONITOR_SCRIPT_ROOT}/grouprins.py --mail=1 --reboot --nodeselect 'state=DEBUG&&boot_state=(rins|dbg|boot)' --stopselect 'state=BOOT' || :
+#${MONITOR_SCRIPT_ROOT}/statistics/add-google-record.py --email email  --password password --database MonitorStats --sheet NodeHistory `${MONITOR_SCRIPT_ROOT}/statistics/get-records.py nodes`
+#${MONITOR_SCRIPT_ROOT}/statistics/add-google-record.py --email email --password password --database MonitorStats --sheet SiteHistory `${MONITOR_SCRIPT_ROOT}/statistics/get-records.py sites`
 
 cp ${MONITOR_SCRIPT_ROOT}/monitor.log ${MONITOR_ARCHIVE_ROOT}/`date +%F-%H:%M`.monitor.log
+service plc restart monitor || :
 rm -f $MONITOR_PID
+
+D=`date +%F-%H:%M`
+
+# NOTE: check log for major sections.
+wc=`grep -E "^(findbad|findbadpcu|nodebad|pcubad|sitebad|apply-policy)$" ${MONITOR_SCRIPT_ROOT}/monitor.log | wc -l`
+if [[ $wc -ge 6 ]] ; then 
+    send_mail "A:finished monitor run for $SD at $D" "Thank you..."
+else
+    send_mail "ERROR finished monitor run for $SD at $D" "Missing some sections:
+        $(grep -E "findbad|findbadpcu|nodebad|pcubad|sitebad|apply-policy" ${MONITOR_SCRIPT_ROOT}/monitor.log)"
+fi
+