[quagga-users 13111] Re: OSPF Top Router Disconnection Issue! Urgent!

Arif mohdarifsaifi at gmail.com
Wed Dec 26 08:35:57 GMT 2012


> 
> Thanks!
> 
> Regards,
> Arif
> 
> On Fri, Dec 21, 2012 at 12:45 PM, sachi ka <sachika1102 at ...> wrote:
> > Hi Arif,
> >
> >
> >
> >> >Did you ever come across any issue like this?
> >> >Is it related to any of the closed or open issues of Quagga?
> >>
> >> >Is there anything special about the topmost router that has only one
> >> > neighbor?
> >> >Are there special functions in OSPF source that that handles the vertex
> >> >(topmost) routers?
> >>
> >> >Any hint/suggestion from your side could be of great help for me.
> >>
> > Please enable the all debug traces of ospfd (debug ospf xxx) and share the
> > log. Also share the routing table (ip route show) & ifconfig details.  So
> > that we can debug this problem.
> > As far my knowledge, i haven't come across this issue.
> >
> > Thanks
> > SK.
> 

Hi Scott, Sachi K, Vishal & All Quagga Users,

Thanks a lot for your active responses.
I'm still investigating the "disconnection problem" but i think i need your 
suggestions now to move ahead.

I have managed to get the test logs.

Mohd. Arif <mohdarifsaifi at ...> writes:

> 
> Thank you all for your responses.
> 
> Initially, i was thinking that the same issue may be faced by other
> quagga users/developers.
> If that is not the case, i will have to post more information to get
> your views on the issue.
> 
> I will do the same as soon as possible.

The test logs primarily include wireshark logs & zebra, osfpd logs.
Here is the link for the logs: 
https://docs.google.com/open?id=0B0j1jJ4Zi5y3eDNSRUg2Vmt3SW8

The problem_statement is as follows:
TOPOLOGY:
=========

10.48.179.11/29				eth6: 10.48.179.9/29
PC#2---------------------------------------048y
					    |eth0: 10.248.48.9/29
					    |
					    |eth0: 10.248.48.10/29
				          043d
					    |eth1: 10.248.43.9/29
					    |
					    |eth0: 10.248.43.10/29
					  043s
					    |eth1: 10.248.43.1/29
					    |
					    |eth0: 10.248.43.1/29
PC#1---------------------------------------002k
10.2.179.3/29				eth6: 10.2.179.1/29
	
STEPS TO REPRODUCE:
===================

1] Power ON all routers: 048y, 043d, 043s, 002k
2] Test PING is successful from PC1 (10.2.179.3) to 048y (10.48.179.9)
3] Restart 048y router
4] Wait for 2-3 minutes
5] Again Test PING from PC1 (10.2.179.3) to 048y (10.48.179.9)
   It won't be successful if the problem has reproduced.
6] Check routing table of 048y. It shows only the DIRECTLY CONNECTED routes. 
   It won't show other routes of the complete network
   Nothing is displayed in show ip ospf database command and in 048y
   "show ip ospf neighbor" shows router 043d as it's neighbor.
7] Restart 043d router to recover from the failure situation.
8] Wait for 2-3 minutes
9] Again Test PING from PC1 (10.2.179.3) to 048y (10.48.179.9)
   It would be successful
   
MY OBSERVATION:
===============
- This problem is reproducible only once in ten attempts.
- No route summarization, virtual links, stub areas are used in the complete 
topology.
  Only the basic network, area configuration is done with only area 0 for 
complete topology.
- During all the above steps, it is checked that PING from 048y to 043d is 
always successful.
  So there is no physical disconnection.
- On checking ospfd.log, i realized that after restart of 048y, 048y has 
successfully detected 043d as it's neighbor.
  But this neighbor relationship has never reached to FULL state. It has stuck 
to LOADING state.
  But i am yet not 100% sure that this could be the root cause reason or not.
  I didn't understanding why it stuck to LOADING state.
- It is also observed that during problem, if we restart 048y, it won't 
recover from the situation.
  The only way to recover from the situation is to restart 043d.

I will be grateful if you can suggest something to me.
Your views will help me in carrying out the investigation.

Thanks!

Best Regards,
Arif





More information about the Quagga-users mailing list