RFC L2 (switch) topology. Mainly off-topic.

Subhendu Ghosh sghosh at sghosh.org
Wed Apr 7 16:17:08 CEST 2004


Hi Stanley

I'm interested...

-sg

On Fri, 5 Mar 2004, Stanley Hopcroft wrote:

> Dear Ladies and Gentlemen,
> 
> <this is at most marginally on topic>
> 
> I am writing to ask if there is any interest in using layer two 
> topological information with Nagios.
> 
> Potential uses include (that I will not probably not contribute to)
> 
> . custom status maps that show the bridge/switch layout with respect to 
> the spanning tree, which in many cases is the inter switch path viewed 
> from the point of the root bridge.
> 
> This is perhaps less useful than it may seem because the status map 
> would ideally show the relationship of the hosts to the switches and 
> while this is doable, it is slow with more than a small number of 
> switches (say less than 10).
> 
> . custom checks that analyse the spanning tree (and whine if the 
> favourite/most powerful/least powerful/XP in bridge mode/some stray 
> wireless node bridge is not or is the root bridge.
> 
> What you would get ?
> 
> 1 Net::SNMP (an absolute joy of a Perl module) pollers that ascertain
> 
>   - bridge MAC addresses (for relating to name and or IP)
>   - descendant bridges
>   - the switch 'designated bridge'
>   - nodes in a netblock that are switches (ie respond to the 
>     dot1dBaseType with a value of 'transparent')
> 
> 2 A Perl CGI that displays them (primitively) in a tree form with the
> root at the top
> 
> This is the only application I have at the moment, although in 
> principle, generating a dot map and having graphviz generate a .png for 
> a better looking web display 'shouldn't be too hard' (TM).
> 
> Switch topology.
> 
>         SCBR21-C5K-2
>                 DBR21-C5K-1
>                         DNGR21-C29-10
>                                 DNGR22-C28-11
>                         DN1R21-C29-20
>                                 DN1R22-C28-21
>                         DN1R23-C28-22
>                                 DN1R24-C28-23
>                         DNR21-C29-30
>                                 DN2R23-C28-32
>                                         DN2R24-C28-33
>                         DN2R22-C28-31
>                         DN3R21-C29-40
>                                 DN3R22-C28-41
>                                         DN3R23-C28-42
>                         DN3R24-C29-43
>                         DN4R21-C29-50
>                                 DN4R22-C28-51
>                                         DN4R23-C28-52
>   .. snip ..
> 
>                 SC1R41-C29-65
>                         SC1R42-C28-66
>                                 SC1R43-C28-67
>                 SC1R44-C28-68
>                 SC1R45-C28-69
>                 SC1R46-C29-162
> 
> What about the bad bits ?
> 
> 1 The pollers don't reliably discover switches since 
> 
> 1.1 agents don't respond if they have better things to do
> 
> 1.2 the switch discovery wants to be fast; I haven't bothered to retry 
> the requests; maybe I should.
> 
> 2 The code is not suitable for publication ie supportable or without 
> known showstoppers (how can you have showstoppers in 300 lines of Perl ? 
> well, I can).
> 
> In fact, there may even be conceptual flaws also.
> 
> 3 It almost certainly doesn't scale.
> 
> At the moment the switch discovery and polling of ~ 80 switches in a /24 
> takes less than 5 seconds - without connectivy or utilisation problems.
> 
> wins> /home/anwsmh/perl5/dotime 5 ./update_switch_topology
> Running ./update_switch_topology 5 times
> 1 2 3 4 5 done
>          Avg        1        2        3        4        5
>       ------   ------   ------   ------   ------   ------
> real   3.730     3.52     4.17     3.69     3.51     3.76
> user   1.120     1.19     1.18     1.08     1.08     1.07
> sys    0.062     0.07     0.07     0.07     0.04     0.06
> wins> 
> 
> 
> Please would you let me know if there is any interest or that
> 
> - this is stupid; L2 sucks (sure does)
> 
> - we already use big name product foo to do this and foo is great
> 
> - why would you want to do this
> 
> - we need it done for big_number k switches across n netblocks
> 
> - does it work for Cisco rapid spanning tree (probably not).
> 
> 
> </this is at most marginally on topic>
> 
> Yours sincerely.
> 
> 

-- 




-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null





More information about the Users mailing list