06-12-2003 04:18 PM - edited 03-02-2019 08:06 AM
I currently am using 3 marking via IP Precedence for mission critical apps, best effort apps, and < best effort apps. The link is highly congested and I do lose EIGRP neighbor relationship relatively frequently due to congestion.
I have received conflicting info. Many say that EIGRP is set to a high precedence value by default, if so what is the value and do I need to specify it's treatment in QOS or is there a background QOS policy for it? Others say it is not marked with IP Precedence values, if this is the case, what's cisco's recommendation on a value when compared to voice/video values?
Any help would be much appreciated, thank you!
06-12-2003 04:55 PM
If I remember right (I've not looked at that code in a bit, and my mind is like a steel sieve, only the big pieces stick). EIGRP hello's are marked with tos 6, or the flash, I think it is, and the remainder of the packets are not marked. On the originating router, these high priority packets (not just eigrp packets, other protocols mark their some of their packets at this high level, as well), are placed in a special outbound queue.
So, I suppose the first question is: Why are the neighbors dropping? Is it because they are losing hello's, or other traffic? ARE you seeing hold timer expirations, or retry interval timeouts? If it's retry interval timeouts, then you probably need to do something about treating eigrp unicast traffic in a special way in your policy, so it doesn't always drop, I'd think.
On the inbound side, these packets are treated as a special case by SPD, but it is possible, through user configuration, to cause the packets to drop. I could, of course, be wrong on the marking level, etc., but that should point you in the right direction.
Russ
06-12-2003 05:05 PM
Russ,
The neighbor loss is due to holding timer expirations, see output below.
Jun 12 22:20:44: %DUAL-5-NBRCHANGE: IP-EIGRP 100: Neighbor 10.225.140.1 (Serial0
/0.608) is down: holding time expired
Jun 12 22:20:46: %DUAL-5-NBRCHANGE: IP-EIGRP 100: Neighbor 10.225.140.1 (Serial0
/0.608) is up: new adjacency
NU2W0001# sh ip eigrp ne
IP-EIGRP neighbors for process 100
H Address Interface Hold Uptime SRTT RTO Q Seq Type
(sec) (ms) Cnt Num
2 10.225.140.1 Se0/0.608 9 00:41:20 124 744 0 350832
Any suggestions on what to do? Should I mark and treat all EIGRP packets? Will that help?
06-13-2003 06:34 AM
You can't mark all eigrp packets outbound, at least there's not any way that I know of, other than possibly using policy routing with a local policy configured. You could account for eigrp packets in your queueing policy inbound and outbound on your boxes, that would probably help. Just give preferential treatment ot packets to 224.0.0.10, or, if you are using an access list to classify packets, match on eigrp, rather, than ip.
:-)
Russ
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide