search

Bahaaqm / bgpsimple

Automatically exported from code.google.com/p/bgpsimple
0 stars 0 forks source link

Convergence Problem for large MRT files ! #5

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago 
What steps will reproduce the problem?
1. I aim to inject some MRT snapshots to my Quagga machines. My feeder 
(bgp_simple) is supposed to advertise a MRT of almost 400,000 lines (taken from 
what my actual transit is sending to me)
2. But bgp_simple never stops sending updates, It keeps resetting the session 
and sending the updates again and again, and finally after 4-5 hours suddenly 
completely tears down the session.
3. during this time my Quagga's "sh ip bgp summ" shows that the neighbourship 
forms, nothing in the queues (router can process all received messages), while 
prefix counter never stops increasing since the session breaks and then become 
Idle, Active and again increasing. Therefore, I can say my routing never 
converges.

What is the expected output? What do you see instead?
Although I am advertising a full BGP dump, I expect it to converge at some 
point! My question is is bgp_simple is able to inject this size of MRT ?

What version of the product are you using? On what operating system?
I am using bgp_simple on a linux machine (customized kernel for Netkit)
and my router is Quagga, I have also tested Dynamips

Please provide any additional information below.
I have also tried this:
http://code.google.com/p/bgpsimple/issues/detail?id=4&can=1

Many thanks,

Original issue reported on code.google.com by mhd.s...@gmail.com on 19 Sep 2012 at 2:02

GoogleCodeExporter commented 9 years ago 
Sorry, I could understand right after finishing the full update injector resets 
the session, bgp_simple's terminal:

Full update sent.
Error occured: type [Hold Timer Expired]
Connection reset with peer 10.10.10.1, AS 65000.

Any help is highly appreciated.

Original comment by mhd.s...@gmail.com on 19 Sep 2012 at 3:15

GoogleCodeExporter commented 9 years ago 
Please try increasing the BGP hold time on both peers. Apparently, the 
bgp_simple.pl script will start sending keepalives only after it sends the full 
MRT dump. If it can't do that within 180 seconds (the default hold time) the 
season will go down. I usually put a very large hold time (like 10k seconds) to 
avoid problems like these.

Original comment by renatowe...@gmail.com on 2 Oct 2012 at 1:18

GoogleCodeExporter commented 9 years ago 
Thank you very much for you help.

Original comment by mhd.s...@gmail.com on 3 Oct 2012 at 3:50