8600 Redundancy Issue

[R.J.]

Hi!

I recently commissioned an 8300 that serves as a distribution switch at a remote site supporting a number of 5520s.  Two 100 base LR links configured as a MLT are run back to the Cores (two 8600s).  Each of the 8600s have an SMLT configured back to the 8300.  When the second port of the MLT is activated on the 8300, connectivity to it and/or some of the remote sites coming off that 8300 are lost.  What makes this even more peculiar is that the person sitting next to me would be able to access the sites I cannot and vice versa.  We have had this issue on a few 5500s, not all, that have been set up with a MLT back to our cores and SMLTs configured on the cores.

We suspect that there are issues somewhere in the VRRP/IST/Core relationship but have been unable to see anything out of the ordinary.

Cores are two 8600s (5.1.6.0)  8 ISTs spanning 3 different 8630GBR boards.
Distribution switch 8300 (4.1.3.1)
End switches 5520-24/48T (5.1.5.009)


Any suggestions you have would be greatly appreciated!  Thanks in advance!

R.J.

[Michael McNamara]

Hi R.J. and welcome to the forums!

Let me explain why your co-worker is working and you aren't... this is the hashing algorithm that (tries) to load-balance packets across an MLT/SMLT connection. The algorithm looks at the traffic and essentially splits the traffic based on source IP address (and sometimes destination IP address). The traffic from your desktop/laptop is being routed/bridged across a path that is "broken" when the traffic from your co-workers desktop/laptop is being routed/bridged across a path that is working.

I would verify that you have the ports configured properly (auto-negotiate - you have link on both sides right, VLACP, tagging, etc) and I would also confirm that you have the same VLANs tagged across both ports on both sides of the links. It's also best practice to set the PVID to the same on all sides, usually the management VLAN if you have one.

If you need additional help please upload a "show config" and "show tech" from all three switches and highlight which ports you are using.

Good Luck!

[R.J.]

Thanks for the info!   I will start comparing the links on both cores.  As far as the remote distribution switch, both ports of the MLT and both ends are set up the same.  I definitely believe that the problem lies within our ISTs, so again, going with your recommendation, I will verify those. 

Thanks again!
R.J.

[Michael McNamara]

I have yet to see any issues myself with 5.1.x software that involve some of the previous FDB/ARP issues and the IST links.

When the link comes up... you have VLACP configured on the ports so VLACP enters an UP state?
When the link comes up... you have STP disabled on the ports?
When the link comes up... you get link on both ends of the connection?
When the link comes up... you get ifOctetsIn and ifOctetsOut on both ports incrementing?

It's really difficult to try and guess what's wrong without getting some hints. Perhaps you could post an excerpt of the logs from both switches for the timeframe you tried to bring up the additional link?

Cheers!


Perhaps you could post an excerpt from your logs... it's very difficult to guess what might be wrong.

[R.J.]

We don't have VLACP enabled globally, STP isn't enabled on any of those trunk ports, we do get link on both ends, and they are accumulating.  Do you have an email that I can send any configs to?  I can't post those here in a public forum.  Thanks again!

[Michael McNamara]

I've responded to you with a PM.

Cheers!

[Jon Hurtt]

Not sure if you checked and if you have solved this but have you tried changing the ports on each link... could be as simple as a bad port or bad cable...

I am also assuming that if you have two links , Link A and Link B when you plug Link B you have some traffic loss. What happens if you only have Link B active, do you have complete traffic loss or still some traffic loss?