> Date: Tue, 11 Dec 2001 08:59:30 +0530
> From: "Hanumanth Jerbandi (SSD-BLRAKS-IC&NI)" <Hanu@wipro.co.in>
> To: "'R. Drew Davis'" <drew@research.bell-labs.com>, alaerte@embrate
l.com.br
> Subject: RE: Cflowd Message: failed to create flows
> Cc: cflowd@caida.org
>
>hi... Drew...
>
> I hv a query on cflowd, Hv u tried to run multiple instances of
>cflowd/flowscan.
> I want to collect flows saperately for 2/more different
>routers/serials, or want to
> collect saperate(stored in respective ip address dir) flows for 2 or
>more serials on the same router.
>
> In fact i had posted this to mailing list... but i didnt get any
>reply...
> hope if u can help me.
>
> I hv seen that graphs reported for i0services is reversed, i.e in/out are
>swaped, how to fix it. Does the ifindex on the router changes!!!, i hv
>noticed this
>
>
> thanks,
> hanu
I think I remember seeing your e-mail to the list. My own cflowd setup
is pretty simple. Had 2 routers feeding one data collector. At the
moment, I'm not even collecting cflowd data.
Not only that, but our routers were nearly all ethernet, except for one DS3
to Holmdel, so I've very little experience with serial lines and netflow.
Sometimes I'm able to help people on the list by making informed guesses,
but I try to stop myself from making uninformed guesses to the list.
(My Dad's advice was "Better to let people think you are dumb than to
open your mouth and prove it to them").
So, I hope someone else on the list has a bit more relevant experience and
a smidge of free time to try to help you out with your question. Sorry
I can't be of better service to you.
You want to run 2 instances of cflowd on the same server? Going to
need a ton of shared memory and disk space and bandwidth to make that
work, assuming traffic is going to be heavy. As long as you make the
config files such that the streams don't cross each other, seems like it
should work. But remember Egon Spengler's advice: Crossing the
streams would be bad.
Editorializing a bit --- It isn't obvious to me: Why do you want
separate collectors? You might get more attention from the list if
you can convince people its an interesting problem that they might want
to pursue for their own purposes.
Drew
R. Drew Davis, Room MH 2C-302A E-mail: drew@bell-labs.com
Bell Laboratories Voice: 908-582-7280
600 Mountain Ave. Fax: 908-582-3340
Murray Hill, NJ 07974
Bell Laboratories: The Research & Development Unit of Lucent Technologies.
-- cflowd mailing list cflowd@caida.org
This archive was generated by hypermail 2b29 : Mon Dec 10 2001 - 21:21:03 PST