03-23-2017 04:55 PM - edited 03-01-2019 04:37 AM
Hi everybody,
I hope someone would have an answer for me. I deployed the APIC-EM, and so far it's working fine, except that I'm scratching my head over one thing. After, performing many discoveries (using both cdp, and range discovery), when I go to Topology, there are devices with no links, ( 0 connections ), even though, they have few neighbors connected to them. So, in the map they are just sitting by themselves, I thought aggregating them with their respective neighbors would create some type of dependency, but then quickly, I realized that I was wrong . These devices have cdp enabled, and sh cdp neigh confirms it, I don't know if I'm doing something wrong with the discovery process, which is pretty straight forward, or what.....it's even affected the path trace app, because the trace goes nowhere from these devices, and it errors out with this message:
"Not able to locate unique interface or host for source ip address, Not able to locate unique interface or host for source ip address"
Below is just one of few devices ( edge routers ) that have missing links ( no connection to their neighbors ), which is killer to path trace
THANK YOU,
03-23-2017 10:46 PM
Normally this "just works". Can you confirm the following:
1) Version of APIC-EM
2) cdp in both directions (you need to have it bi-dir)
3) the interface type of the link at both ends (is it a sub-interface, routed interface etc).
4) What device is this supposed to be connected to
Adam
03-27-2017 03:44 AM
I have the exact same issue, with APIC-EM Version 1.4.0.1959.
While in the cases where I added the devices one-by-one manually the connections seem to be ok, on other cases where I used the Discovery feature, I got broken connections.
03-28-2017 01:49 PM
Hi Adam,
Sorry for the delay. thanks for the help. Enjoyed, your conferences,
1) APCI-EM ver: 1.4.0.1959
2) a show cdp neigh , show all the respective neighbors ( both direction ), however when I did: debug cdp packets on the those troubled devices, no cdp debug data is returned ( which should be returned every 60 sec - default - ).
3) links affected: L2/L3. Devices: ASR, ISRs, Cat switches, two of our ASAs. I'm waiting for the APIC-EM support engineer to respond to my case, I will let you guys know.
Thank you,
04-04-2017 06:57 PM
aradford,
Just a quick update, I've been back and forth between PI/APIC engineers for the past week, for this same issue, because I'm experiencing broken connections ( 0 connections), missing links, even though the devices exist in the inventory, and their discovery was cdp based ( and successful ).
I was told by the engineers, that it is possible my prime 3.0 is the cause of the issue ( APIC is integrated with PI), because I have broken install, and apparently, a lot people have had a PI 3.0 broken install, because Cisco pushed the wrong updates, that screwed up the database.
At this point there's still no solution to this problem, it's like a broken network, because packets can't flow. If anybody, got a fix please let us know.
Thanks,
04-04-2017 11:17 PM
In my case however there's no prime infrastructure integration and I still have those issues, which are similar to what you described. Furthermore, even in cases where links appear to be normal, path trace won't work.
I might give v1.3 a try before I give up..
04-04-2017 11:23 PM
This is strange. The integration between APIC-EM and PI is one way for inventory. PI just reads APIC-EM API, so I am curious as to how TAC thinks that is impacting APIC-EM.
04-12-2017 04:15 PM
I have upgraded yesterday to APIC-Em version 1.4.1159. And it does look kind of better as far as missing those links.
Do you think NETCONF might be the culprit here?
04-12-2017 04:31 PM
I notice on the path trace that you have a number of unknown devices? Have you tried adding them to inventory?
04-12-2017 05:42 PM
Those unknown devices are PE devices.
I upgraded to 1.4.1.1159 and the link discovery seems a bit better (was able to found more links) but still far from perfect.
I have sites with identical hardware and IOS versions and while on the first one links/path trace works, on others don't.
Topology Service > Instance Logs look like:
2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 805fa1d2-7b60-4491-b24c-6c9fad047d54 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 80a9187f-4754-4630-8d29-fc75b9b4f590 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 80e19231-b85b-4bb1-8cd9-567cfb7d11b1 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 8327235d-dd11-44ae-a381-09a5e8d4a506 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 83edb604-b274-4bc5-a527-e56af5313a49 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 84b7a83d-4e4c-4c04-b6d7-6e8e33a59327 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 84bf28a4-1260-43cf-8d12-ce88514fa241 | 2017-04-13 00:28:25,009 | WARN | qtp830918697-12453 | topology-service | c.c.e.t.impl.CustomTopologyService | NodeCustom not found in RAS for node id: 8501e56f-bdf5-4385-8bad-b184845c0078 |
03-28-2017 02:20 AM
Is there any way to force link re-discovery on the devices? Even deleting them and adding them again one-by-one doesn't work.
The weird thing is that the same devices that in the lab server were connected properly, on the live server refuse to form links
03-28-2017 06:38 AM
Hi Vag,
if you do a rediscovery of the device, it should rediscover the links. Are you doing a CDP based rediscovery, or an IP range?
1) Do you have CDP neighbour on both sides (show cdp ne)?
2) What is the type of the device that does not have a connection and it's version of code (is it supported)? e.g 2960X - 15.2.5e
3) How are the devices that do not have links connected? L2/L3/ Subinterface?
Adam
03-28-2017 05:37 PM
Hi araford,
1) CDP is enabled on both sides. On the lab server, the same devices formed links between each other properly and path trace worked.
2) The 1st site consists of WS-C3560-24PS-E (12.2(55)SE6), WS-C2960G-48TC-L (15.0(2)SE7) switches and CISCO2911/K9 (15.4(3)M2) routers. While the switches are not in the support list, they did work properly on the lab server.
The second site has a 2960X switch stack(15.0.2-EX5), 887 and 2911 15.2(4)M3 routers.
All links are Layer 2, no sub interfaces.
Re-discovery didn't fix the missing links. I tried CDP based, IP range based and even adding the devices one-by-one at the device inventory.
That's a path trace from the lab server at the same site:
04-01-2017 09:35 PM
Hi Vag, have you configured a domain-name on the AP?
04-02-2017 05:39 AM
Hi aragord,
All of the devices and the apic-em server are configured with a domain name.
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