Transceiver DDMI stats via SNMP

[jeronimo]

I found several posts on this, most of the time with a final comment like "not YET available".
So when will it be? 🙂

This would be great to have on EOS and EXOS platforms.
I see that you can at least do webservice-kindof requests to EXOS, that alleviates the problem a little bit. But I still think this doesn't "abolish" SNMP. 🙂

Why we need this: We want to know what is happening on the fibers without having to drive there, take down the links, plug in the power meter, clean the fiber afterwards, drive to the other DC, same thing, etc. etc.

:)

[jeronimo]

FWIW Here is an expect script for the show transceiver command

I highly recommand using the ro login for the switch!!

#! /usr/bin/expect

set host [lrange $argv 0 0]
set port [lrange $argv 1 1]

spawn ssh ro@$host
expect {
{word:?} {send "password\r"}
}
expect {
{)->} {send "show port transceiver $port\r"}
}
expect {
{)->} {send "exit\r"}
}
exit
And here is the wrapper in bash

#!/bin/bash

if [ $# -ne 2 ]; then
echo "Syntax: $0 host port"
exit 1
fi

TMPF=/dev/shm/$0.$1.$2.tmp
rm -f $TMP

host=$1
port=$2

/usr/local/nagios/libexec/show-transceiver.xp $1 $2 > $TMPF

VOLT=$(awk '{if ($2=="Voltage") print $4}' $TMPF)
TEMP=$(awk '{if ($2=="Temp") print $4}' $TMPF)
TX=$(awk '{if ($2=="TX" && $4=="(dBm)") print $5}' $TMPF)
RX=$(awk '{if ($2=="RX" && $4=="(dBm)") print $5}' $TMPF)

echo "$TEMP degrees C, $VOLT Volt, $RX dBm RX, $TX dBm TX|v=$VOLT t=$TEMP tx=$RX rx=$TX"
This can be used as Nagios plugin (if you add thresholds and appropriate return codes).

Have fun 

[jeronimo]

Oh and remember to do an SSH as the nagios user to the switch once because it wants you to validate the host key.

[jeronimo]

I just noticed that you might want to implement the wrapper in a more sophisticated language than a shell script because (at least when using bash) it won't even work with decimals.... Anyway here is my current version of a nagios plugin

#!/bin/bash

if [ $# -ne 2 ]; then
echo "Syntax: $0 host port"
exit 1
fi

STATE_UNKN=3
STATE_CRIT=2
STATE_WARN=1
STATE_OK=0

TMPF=/dev/shm/$(basename $0).$1.$2.tmp
rm -f $TMP

host=$1
port=$2

/usr/local/nagios/libexec/show-transceiver.xp $1 $2 > $TMPF

VOLT=$(awk '{if ($2=="Voltage") print $4}' $TMPF)
VOLTHW=$(awk '{if ($2=="Voltage") print $7}' $TMPF)
VOLTLW=$(awk '{if ($2=="Voltage") print $8}' $TMPF)
VOLTHC=$(awk '{if ($2=="Voltage") print $6}' $TMPF)
VOLTLC=$(awk '{if ($2=="Voltage") print $9}' $TMPF)

TEMP=$(awk '{if ($2=="Temp") print $4}' $TMPF)
TEMPHW=$(awk '{if ($2=="Temp") print $7}' $TMPF)
TEMPLW=$(awk '{if ($2=="Temp") print $8}' $TMPF)
TEMPHC=$(awk '{if ($2=="Temp") print $6}' $TMPF)
TEMPLC=$(awk '{if ($2=="Temp") print $9}' $TMPF)

TX=$(awk '{if ($2=="TX" && $4=="(dBm)") print $5}' $TMPF)
TXHW=$(awk '{if ($2=="TX" && $4=="(dBm)") print $8}' $TMPF)
TXLW=$(awk '{if ($2=="TX" && $4=="(dBm)") print $9}' $TMPF)
TXHC=$(awk '{if ($2=="TX" && $4=="(dBm)") print $7}' $TMPF)
TXLC=$(awk '{if ($2=="TX" && $4=="(dBm)") print $10}' $TMPF)

RX=$(awk '{if ($2=="RX" && $4=="(dBm)") print $5}' $TMPF)
RXHW=$(awk '{if ($2=="RX" && $4=="(dBm)") print $8}' $TMPF)
RXLW=$(awk '{if ($2=="RX" && $4=="(dBm)") print $9}' $TMPF)
RXHC=$(awk '{if ($2=="RX" && $4=="(dBm)") print $7}' $TMPF)
RXLC=$(awk '{if ($2=="RX" && $4=="(dBm)") print $10}' $TMPF)

STATE=$STATE_UNKN

function set_state() {
# only set it if it gets worse... (UNKNOWN is an exception)
if [ $STATE -lt $1 -o $STATE -eq $STATE_UNKN ]; then
STATE=$1
fi
}

for i in VOLT TEMP TX RX; do
# it=i*1000 without decimals
# must use (/1) for it to consider scale=0
it=$(echo "scale=0; ${!i}*1000/1" | bc)
# get content of variable "${i}XXX"
LC="$(eval "echo \${$(echo ${i}LC)}")"
LW="$(eval "echo \${$(echo ${i}LW)}")"
HC="$(eval "echo \${$(echo ${i}HC)}")"
HW="$(eval "echo \${$(echo ${i}HW)}")"
# bash can't do float
# must use (/1) for it to consider scale=0
LC=$(echo "scale=0; $LC*1000/1" | bc)
LW=$(echo "scale=0; $LW*1000/1" | bc)
HC=$(echo "scale=0; $HC*1000/1" | bc)
HW=$(echo "scale=0; $HW*1000/1" | bc)
if [ $it -le $LC ]; then
set_state $STATE_CRIT
prob_crit="$prob_crit ${i}:LOW"
elif [ $it -le $LW ]; then
set_state $STATE_WARN
prob_warn="$prob_warn ${i}:LOW"
fi
if [ $it -ge $HC ]; then
set_state $STATE_CRIT
prob_crit="$prob_crit ${i}:HIGH"
elif [ $it -ge $HW ]; then
set_state $STATE_WARN
prob_warn="$prob_warn ${i}:HIGH"
fi
done

set_state $STATE_OK

case $STATE in
$STATE_CRIT)
STATUS="CRITICAL"
;;
$STATE_WARN)
STATUS="WARN"
;;
$STATE_OK)
STATUS="OK"
;;
*)
STATE=$STATE_UNKN
STATUS="UNKN"
;;
esac

if [ ! -z "$prob_warn" ]; then
STATUS="$STATUS $prob_warn"
fi
if [ ! -z "$prob_crit" ]; then
STATUS="$STATUS $prob_crit"
fi

echo "$STATUS: $TEMP degrees C, $VOLT Volt, $RX dBm RX, $TX dBm TX|v=$VOLT t=$TEMP tx=$RX rx=$TX"

chown nagios: $TMPF

exit $STATE
Example:

# ./show-transceiver.sh 1.2.3.4 tg.1.22
OK: 36 degrees C, 3.237 Volt, -9.281 dBm RX, -2.000 dBm TX|v=3.237 t=36 tx=-9.281 rx=-2.000

I know there are inefficient parts in it. Don't tell me 🙂

[jeronimo]

If you could at least do:

root@nms:~$ ssh rw@1.2.3.4 show transceiver
rw@1.2.3.4's password:
Write failed: Broken pipe

Do we really have to work with "expect"??