OK! I made these scripts for a couple Redback SmartEdge SE400 machines.
#1: Subscribers per Context (SE400 manages the Quality of Service for different subscribers, so this basically keeps track of how many subscribers are online in every context - a context can be a certain bandwidth plan, for instance).
It's a wee bit... raw, and I didn't add any "usage" prompts, as I intended only Cacti to run it. I could put them in though, and I might do it later.
In the XML file, I used the <arg_prepend>|host_hostname| |host_snmp_community| |host_snmp_port| |host_max_oids|</arg_prepend> option, to have those values sent to the script before the "get" or "index" option that Cacti uses to poll data. (hence the $1 $2 $3 $4 usage in the script).
The 5th argument is "numconts" (specified like that in the XML file), or index, query, etc. The documentation said that it would go "script query <field>", but it actually did "script <field>" directly, so I added both options just in case. (
This problem ended up being a missing field in the XML file: <arg_query>, which is now added) I used grep to filter out one context that came out blank ("") and had no corresponding subscriber number, so everything would match. I just used the numbers 1-9 or whatever, as the index. I used awk to print the line number before the line (to simulate the index, and also for filtering the requested value). I used snmpbulkwalk, as it helps reduce network overhead, packing more SNMP requests into one IP packet. You can change it to snmpwalk if you'd like, removing the 4th parameter from the arg_prepend and changing the argument numbers in the script.
Code: Select all
case "$5" in
numconts) /usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.6.3.16.1.1.1.1 | grep -v \"\" | wc -l ;; # count number of values in the SNMPWALK
index) seq `/usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.6.3.16.1.1.1.1 | grep -v \"\" | wc -l` ;;
query)
if [ $6 = "context" ]
then
/usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.6.3.16.1.1.1.1 | grep -v \"\" | cut -d "\"" -f 2 | awk '{print NR ":" $0}'
elif [ $6 = "subs" ]
then
/usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.4.1.2352.2.27.1.2.1.1.1 | cut -d ":" -f 2 | awk '{print NR ":" $1}'
fi
;;
get)
if [ $6 = "context" ]
then
/usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.6.3.16.1.1.1.1 | grep -v \"\" | cut -d "\"" -f 2 | awk '{print NR ":" $0}' | grep $7: | cut -d ":" -f 2
elif [ $6 = "subs" ]
then
/usr/bin/snmpbulkwalk -Cr$4 -v 2c -c $2 $1:$3 .1.3.6.1.4.1.2352.2.27.1.2.1.1.1 | cut -d ":" -f 2 | awk '{print NR ":" $1}' | grep $7: | cut -d ":" -f 2
fi
;;
*) exit 1 ;;
esac
By the way, the $0 or $1 inside the AWK command, are NOT the command line arguments, but fields split by AWK from the input. Bash does not replace the command line arguments $0, $1, if they are inside apostrophes ('$0' will literally print $0, but "$0" will print the script name, anywhere in a Bash script).
The XML file looks like this:
Code: Select all
<query>
<name>SmartEdge - Subscribers per context</name>
<description>Monitors # of subscribers in each service context</description>
<script_path>/bin/bash |path_cacti|/scripts/SE400subs.sh</script_path>
<arg_prepend>|host_hostname| |host_snmp_community| |host_snmp_port| |host_max_oids|</arg_prepend>
<arg_index>index</arg_index>
<arg_query>query</arg_query>
<arg_get>get</arg_get>
<arg_num_indexes>numconts</arg_num_indexes>
<output_delimeter>:</output_delimeter>
<index_order_type>numeric</index_order_type>
<index_title_format>|context|</index_title_format>
<fields>
<context>
<name>Context</name>
<direction>input</direction>
<query_name>context</query_name>
</context>
<subs>
<name>Subscribers</name>
<direction>output</direction>
<query_name>subs</query_name>
</subs>
</fields>
</query>
I don't REALLY understand why a couple of things are necessary (like the index_title_format), but it works. The query_name is the string Cacti will pass as the argument to get an index of the field.
I was thinking of posting some more scripts and XML files (I have a couple more; in the other one I used "sed" instead of awk and cut, since the index was part of the OID - an IP address, not just a number). If you'd like me to post those, I can... just didn't want to make this post too long
).