I have learned how to execute arbitrary commands, invoked by a
pre-defined OID that I create using an "exec" command in the snmpd.conf
file. Example:
exec .1.3.6.1.4.1.2001.65.1.101.1.1 hoplist_1 /usr/bin/cat
/export/home/hopuser/traceroute.host_1.hoplist
This would produce a list of the traceroute hops between the host and a
remote host. However, it is effectively "hard-coded" for every host and
producing a single hop time requires yet more hard-coded OID's.
What I am trying to do is, create a "dynamic" set of extensions that
function similar to IF-MIB. I have a pre-defined list of hosts that I
want to do a traceroute to. I have created an OID like the one above
that produces a list of these targets when an snmpwalk is issued against
it. Example:
snmpwalk -O n -c public -v 2c myhost.att.com .1.3.6.1.4.1.2001.65.1.101
Produces:
.1.3.6.1.4.1.2001.65.1.1.101.1 = STRING: "remote.host_1.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.2 = STRING: "remote.host_2.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.3 = STRING: "remote.host_3.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.4 = STRING: "remote.host_4.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.5 = STRING: "remote.host_5.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.6 = STRING: "remote.host_6.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.7 = STRING: "remote.host_7.att.com"
.1.3.6.1.4.1.2001.65.1.1.101.8 = STRING: "remote.host_8.att.com"
Now, I can create an entry that will "enumerate" the preceding (this is
where things are getting a bit fuzzy for me). Example:
exec .1.3.6.1.4.1.2001.65.1.101.1.3 targetCount
/export/home/hopuser/countHopTargets.ksh
Using an snmpwalk against this OID produces an "index" of the of all the
hosts I want to trace. Example
snmpwalk -O n -c public -v 2c myhost.att.com
.1.3.6.1.4.1.2001.65.1.3.101
Produces:
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.1 = STRING: "1"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.2 = STRING: "2"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.3 = STRING: "3"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.4 = STRING: "4"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.5 = STRING: "5"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.6 = STRING: "6"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.7 = STRING: "7"
.1.3.6.1.4.1.2001.65.1.1.101.1.3.101.8 = STRING: "8"
Furthermore, I have an OID that will produce a count of the number of
trace targets.
exec .1.3.6.1.4.1.2001.65.1.101.1.1 numberOfTargets
/export/home/hopuser/numberOfTargets.ksh
Using an snmpget on the above displays a count of the trace targets.
snmpget -O n -c public -v 2c myhost.att.com
.1.3.6.1.4.1.2001.65.1.1.101.1
Produces:
.1.3.6.1.4.1.2001.65.1.1.101.1 = STRING: "11"
So, I have in place the elements that will provide a list of the trace
targets, enumerate each, and provide a count of the total number of
targets.
I have a script, run periodically by cron, that performs the traceroute,
then places the hop time into files named specifically for that hop and
located in a directory named for the trace target. For example,
traceroute www.ibm.com would produce the following files (incomplete
list):
/export/home/hopuser/www.ibm.com/135.294.13.5
/export/home/hopuser/www.ibm.com/4.69.145.243
/export/home/hopuser/www.ibm.com/4.68.132.104
/export/home/hopuser/www.ibm.com/4.53.0.90
Each file contains the hop times in milliseconds for each host traced.
So, to the crux of my question, what I want to be able to is walk an OID
(child of the root OID, assumedly .1.3.6.1.4.1.2001.65.1.1.101.1 from
above) in a manner that will allow me to bring back all data for all
traced hosts in a "structured" manner, such that an external program can
read and display it. In this case, I'm using Cacti.
I have looked at the IF-MIB and related files in Cacti. It seems to do
what I want in that it dynamically determines all of the interfaces on
the server, then from that can display interface-specific information
for each. From the above, the equivalent "child" on IF-MIB would be:
My OID From Above (walk or get) IF-MIB OID
============================
=========================================
.1.3.6.1.4.1.2001.65.1.101 .1.3.6.1.2.1.2.2.1.2
(IF-MIB::ifDescr.X)
.1.3.6.1.4.1.2001.65.1.1.101.1 .1.3.6.1.2.1.2.1.0
(IF-MIB::ifNumber.0)
.1.3.6.1.4.1.2001.65.1.3.101 .1.3.6.1.2.1.2.2.1.1
(IF-MIB::ifIndex.X)
The interfaces referenced by IF-MIB would be equivalent to my "traced
host" name. Each would provide a "branch" off the root OID that could
be queried for a list of hops, which in turn would have their individual
hop times.
I realize that the above structure is wrong, so that's why I'm coming to
the experts. I want to be able to do this for other functions, without
having to write an agent or build MIB's. I see it structured something
like this:
<Number Of Traced Hosts>
|
--<Trace Host Enumeration>
|
--<Trace Host Name (Descr)>
|
--<Number Of Hops>
|
--<Hop Enumeration>
|
--<Hop Name (Descr)>
|
--<Hops Time>
Cacti allows me to provide it structured index (or indices), and it will
go and get the appropriate "terminal" OID's for all branches.
What I need is the "OID structure" that will allow me to treat my hop
times as IF-MIB treats such things like ifInsOctets. I'm aware that
what I may be asking cannot be done using the "exec" functions in
snmpd.conf. If it does work, then it would make lots of structured data
available via this universally supported protocol.
Thanks for wading through all of this and for any help you can offer.
Michael
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Net-snmp-users mailing list
Net-snmp-users@lists.sourceforge.net
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users
