ATM EIGRP connection flapping

Sharkey13 · ‎06-05-2008

Summary: we recently upgraded a router and circuit at one site. Router was upgraded to a 3825 from a 3640, circuit upgraded from a 3xATM/IMA to a 15 Mbit ATM DS3. All prior connections reconnected, and are working fine, except one.

Once a minute, every minute, a get the syslog EIGRP message (truncated) on this pvc:

06-05-2008 09:29:21 Neighbor 192.168.101.122 (ATM1/0.35) is up: new adjacency

06-05-2008 09:29:21 Neighbor 192.168.101.122 (ATM1/0.35) is down: retry limit exceeded

All EIGRP timers are default settings; I have run pings without any losses; I am not getting any other messages related to the other pvc on this new router.

SH IP EIGRP NEI shows a Q CNT of 1 on this link.

I am getting quite a few input errors and giants on the other end of this circuit, which is a frame-relay btw. It also has an mtu of 1500, whereas I think the 3825 has a mtu of 4470. No other errors seen on 3825 at interface or pvc (subinterface) level.

Thanks in advance. Patrick

sdoremus33 · ‎06-05-2008

What does sh controller tell you?

Sharkey13 · ‎06-05-2008

sh controller has a rather lenghty output. Is there something in particular to look for?

ruwhite · ‎06-09-2008

This is probably layer 2.... Try a couple of things. First, ping to 224.0.0.10. I'm guessing this is working, because you're building an adjacency in the first place, but it's always where I start anyway. Then pint unicast between the two routers, say 10000 packets, with a very low timeout, say 0 seconds.

Next, try pinging with a range of sizes (it's under the advanced options in the ping command--don't put in an address, just do ping, then follow the prompts, and when it asks you for advanced options, say yes). Make certain you put the final size as what the MTU for the link should be.

I'm guessing you either have unidirectional unicast only, unidirectional multicast only, or an MTU problem on the link. It's far too consistent to be random errors, I think.

:-)

Russ

Sharkey13 · ‎06-09-2008

Russ - thank you for the reply.

Ping to 224.0.0.10 is successful on both routers.

Unicast ping from either router to the other, 5000 packets, 0 timeout, 0% successful.

Ping range of sizes: (the results were the same from each router, only one posted):

#ping

Protocol [ip]:

Target IP address: 192.168.101.121

Repeat count [5]:

Datagram size [100]:

Timeout in seconds [2]: 0

Extended commands [n]: y

Source address or interface: 192.168.101.122

Type of service [0]:

Set DF bit in IP header? [no]:

Validate reply data? [no]:

Data pattern [0xABCD]:

Loose, Strict, Record, Timestamp, Verbose[none]:

Sweep range of sizes [n]: y

Sweep min size [36]:

Sweep max size [18024]: 1500

Sweep interval [1]:

Success rate is 0 percent (0/7325)

I look forward to your response and interpretation of these results.

Thanks, Patrick

Sharkey13 · ‎06-09-2008

Russ - your post led me to the solution for this issue.

I changed the mtu on the ATM subinterface in question of the new router to 1500, which matches the frame-relay interface mtu on the corresponding router. Syslog errors stopped immediately.

Thanks for providing some insight on where to look for a solution.

Patrick

ruwhite · ‎06-10-2008

Great! Glad you solved it....

:-)

Russ