MX105 HA pair with stacked 225/210 switches

[IT_Magician]

Okay getting our butt kicked here. Ticket originally opened with Meraki support but was given a lot of run around so hoping Meraki support can see this/community weigh in.

I want to start by saying I think this is a hardware or firmware issue, but here is whats happening.

1. MX105 in a pair
2. Links into MDF Stack of (2) MS225 and (3) MS210 switches.
3. MS225 have 10GB Fiber to downstream IDF switch stack.
4. Here were the initial issues
   1. All devices in switch 01 would lose internet multiple times per day.
   2. To fix we had to unplug/replug network cable or move ports.
   3. Meraki replaced switch 01 which fixed issue for a few weeks.
   4. Same issue now (happens on all devices across all switches) where LAN connection stops and to fix you have to unplug and replug ethernet cable.
   5. Some computers are plugged into phones and when this happens the phone is online, the port on the switch shows no issues, but the computer cannot grab an IP and is stuck at 169 IP until we unplug and replug ethernet cable.

Meraki confirmed configuration looks correct, but they do think there is a bug. Issue is their next step is requiring us to packet capture the device having the issue in the moment, so we are now saying onsite until the issues happens. We are thinking there is a glitch either in:

1. Firmware bug
2. Issue with MDF stack (tonight we are tearing down the stack)
3. Issue with stacking cables (tonight going to also remove all stack cables)
4. Issue with HA pair (future plan is to remove MX02 and just run a single MX)

To reiterate, devices going directly into switch 01, switch 02, switch 03 (all part of same stack) have this issue, and same with devices bridging off the phones. When device drops Meraki shows a 169 IP in dashboard and throws ip source/VLAN mis match over and over until we unplug and replug in ethernet cable and then event log shows successful DHCP reservation.

Help?

[IT_Magician]

Forgot to mention tonight after we tear down the stack/remove cables (going to traditional 1GB ethernet link between core switch and other switches) we are also going to factory reset all switches. If that still doesn't work we will remove the MX105 HA pair, and if that still doesn't work well, we are out of options with Meraki support actually telling us what is wrong.

To be fair, with Meraki support they confirmed they say unexpected behavior doing packet captures and we may have found a bug.

[RWelch]

What firmware version are you running or referring to?

[IT_Magician]

Switches are latest 17.2.2

MX is 19.1.9

[Ryan_Miles]

If you want to DM me your case number and I can take a look

[IT_Magician]

I just sent it, thank you

[Brash]

The weirdest thing about reading this is that devices are showing with APIPA addresses.

Even in the event of traffic drops etc, I'd expect devices to keep their DHCP'd IP address.

This symptom seems to indicate to me that either:

- The link to the client is disconnecting and reconnecting forcing it to request a new IP.

- Clients are sending DHCP renews that are not being acknowledged by the DHCP server. The unplugging and plugging of the network cable then triggers a new DHCP request which the server responds to.

If the issue is reproducible, does it occur for clients with static up addresses? Additionally what device is the DHCP server in this scenario?  Do you see anything in the logs on the DHCP server?