Issue with PCs moving on LACP link

[Tom404]

Hello,

So I just opened a case with Meraki but thought I would post here as well. Here is what I sent in for the case :

I recently started having an issue on our HQ network.

We have one stack for L3 routing and L3 out and 2 stack for distribution. The distribution switches are connected to the core throught one aggregat LACP of 2 links one for each node.

Every morning, when people connect to the network and packets start flowing through the network, I have to close the two links to the MS410-01. otherwise some users have no network

It's pretty difficult to troubleshoot and do some tests since when there is nobody connected to the network everything look fine.

If you have any idea 🙂

[alemabrahao]

Are you sure that LACP was configured property?

[alemabrahao]

Have you checked the STP configuration?

[Brash]

Definitely sounds like your port channel either isn't configured properly, or isn't applying properly.

[Tom404]

LACP is properly configured (speed/duplex/vlan match...) root guest is enable on the core switches... This problem appears after a migration of a L3 interco, I just changed the default route then rollback.

I'm used to Cisco catalyst and never faced this kind of issue...

Really strange

[alemabrahao]

I have never worked with Interco, maybe It's an Interco issue.

[JacekJ]

Just a side note, this bug can be very nasty (has been fixed on MS 15.20):

"Stack routing tables can be incorrect after removing an SVI (present since MS 14.32)"

[Tom404]

Very interesting... I'm in 14.33. I'll do the upgrade and tell if it resolves this issue.

Thanks !

[JacekJ]

You can reboot the MS410 stack to see if this is causing you issues (done that quite a few times when that wasn't reported as a known issue). After a reboot everything should run ok, as you set it up in the dashboard.

So the current work around if you need to remove a SVI - is either not remove it at all, or remove and then do a reboot of the stack AFAIK.

The MS15.21 is the first 15.xx that got into stable, so I would wait a bit with the upgrade just to be on the safe side.

[cmr]

We had the issue with an MS355 stack running 15.18/19 and rebooting it didn't help...   The upgrade to 15.20 did fix it.

[Tom404]

Hello,

I had to do the same migration than last time last thursday.

- Shut down the two former ethernet links to our former datacenter

- Turn on the new ethernet link

- Change the default gateway to the new link

Everything worked fine after the migration (oowh)

But since last monday I can ping (from an other location each gateway on the branch switch stack meraki (MS410) but when I ping the endpoints, or even if I ping the endpoints from the local gateway on the meraki switch I've got very high latency.

I can't understand how it can be so unstable

0
0/ 100 = 0% |
1 1ms 0/ 100 = 0% 0/ 100 = 0% 10.20.x.x (HQ gateway MS410)
0/ 100 = 0% |
2 4ms 0/ 100 = 0% 0/ 100 = 0% 10.0.x.x (DC Firewall <> HQ interco)
3/ 100 = 3% |
3 8ms 3/ 100 = 3% 0/ 100 = 0% 10.0.x.x (Branch Meraki MS410)
52/ 100 = 52% |
4 465ms 55/ 100 = 55% 0/ 100 = 0% 10.10.x:x (Endpoint)

Not sure if it's clear enough, and sorry about my english 🙂

[cmr]

@Tom404 what firmware version are you running?  If on 15.x and lower than 15.20 I would upgrade to the current release.

[Tom404]

Hi CMR, I'm running 14.33, I would do it but there are many switches/AP in the same tenant and did not find if it's possible to choose which device I want to upgrade.

[JacekJ]

You can't upgrade single switches, you go either all or none for a given network in your dashboard.

The only thing you can do is to stage the switch upgrades, so they won't go all at one time:

https://documentation.meraki.com/MS/Other_Topics/MS_Firmware_Upgrades#Staged_Upgrades