ping_mon/path-loss.sh

126 lines
2.6 KiB
Bash
Raw Normal View History

2016-09-01 13:31:00 +00:00
#!/bin/bash
# rburkholder@quovadis.bm
2016-09-02 20:18:20 +00:00
2016-09-01 13:31:00 +00:00
# raymond@burkholder.net
2016-09-02 20:18:20 +00:00
# http://blog.raymond.burkholder.net
# requires use of (with some modifications, will post):
# https://github.com/enigma-io/pd-trigger
2016-09-01 13:31:00 +00:00
# user defined settings
2016-09-02 20:18:20 +00:00
txtEmail="name@example.com"
2016-09-01 13:31:00 +00:00
txtSubject="Path Loss Report"
txtSeparator="==============================\n"
nTrigger=4
nAttempts=5
2016-09-02 20:18:20 +00:00
txtPdServiceKey="000000000000000"
2016-09-01 13:31:00 +00:00
declare -A nodes
nodes=( \
2016-09-02 20:18:20 +00:00
[8.8.8.8]="google1 member" \
2016-09-01 13:31:00 +00:00
[8.8.4.4]="google2" \
)
# local variables
status="|"
2016-09-02 20:18:20 +00:00
cntNotify=0
2016-09-01 13:31:00 +00:00
tmpLog=$(mktemp)
tmpPing=$(mktemp)
2016-09-02 20:18:20 +00:00
cntNodes=0
cntNodesDown=0
cntMembers=0
cntMembersDown=0
declare -a items
2016-09-01 13:31:00 +00:00
# preload output
date > ${tmpLog}
# loop through nodes and test
2016-09-01 13:31:00 +00:00
for node in ${!nodes[*]}; do
2016-09-02 20:18:20 +00:00
((cntNodes++))
# split out node details
# 0: alias/name
# 1: optional 'member' for determining edge outage
info=${nodes[${node}]}
items[1]="none"
ix=0
for arg in ${info}; do
items[ix]=${arg}
((ix++))
done
name="${items[0]}"
2016-09-01 13:31:00 +00:00
echo -e ${txtSeparator} >> ${tmpLog};
2016-09-02 20:18:20 +00:00
echo "checking node ${name}:" >> ${tmpLog};
2016-09-01 13:31:00 +00:00
echo "" >> ${tmpLog}
2016-09-01 13:31:00 +00:00
ping -W 1 -c ${nAttempts} ${node} > ${tmpPing}
cat ${tmpPing} >> ${tmpLog}
2016-09-01 13:31:00 +00:00
value=$(grep transmitted ${tmpPing} | cut -d ' ' -f 4)
2016-09-02 20:18:20 +00:00
if [[ nTrigger -ge value ]]; then
flagNxt="dn"
((cntNodesDown++))
else
flagNxt="up"
fi
if test "member" == "${items[1]}"; then
((cntMembers++))
if test "dn" = "${flagNxt}"; then ((cntMembersDown++)); fi
fi
flagPrv="na"
if [[ -f /tmp/pl.dn.${node} ]]; then
flagPrv="dn"
if test "up" = "${flagNxt}"; then
rm /tmp/pl.dn.${node}
2016-09-02 20:18:20 +00:00
fi
fi
if [[ -f /tmp/pl.up.${node} ]]; then
flagPrv="up"
if test "dn" = "${flagNxt}"; then
rm /tmp/pl.up.${node}
2016-09-02 20:18:20 +00:00
fi
fi
if test "${flagPrv}" != "${flagNxt}"; then
touch /tmp/pl.${flagNxt}.${node}
((cntNotify++));
2016-09-01 13:31:00 +00:00
mtr -w -b --report ${node} >> ${tmpLog};
2016-09-02 20:18:20 +00:00
status="${status} ${name} ${flagPrv}>${flagNxt} |"
echo "" >> ${tmpLog}
echo "Above State Change: ${flagPrv}>${flagNxt}" >> ${tmpLog}
2016-09-02 20:18:20 +00:00
fi
2016-09-02 20:18:20 +00:00
done
2016-09-01 13:31:00 +00:00
# footer
2016-09-02 20:18:20 +00:00
echo -e ${txtSeparator} >> ${tmpLog}
2016-09-01 13:31:00 +00:00
date >> ${tmpLog}
# notify on failure
2016-09-02 20:18:20 +00:00
if [[ cntNodes -eq cntNodesDown ]]; then
logger "path-loss - all nodes unreachable"
else
# if something to notify
if [[ cntNotify -gt 0 ]]; then
# need a pagerduty alert if all important members are down
if [[ cntMembers -eq cntMembersDown ]]; then
response=$(cat ${tmpLog} | ./pd-trigger.sh -L -s "${txtPdServiceKey}" -d "${status}")
fi
# attempt an email
cat ${tmpLog} | mail -s "${txtSubject}:${status}" ${txtEmail};
fi
fi
2016-09-01 13:31:00 +00:00
# clean up
rm ${tmpLog}
rm ${tmpPing}