Unreachable devices when not connected to the core switch

[Fabian1]

Hi everyone,

I have an issue, I where I might be blind to see the solution.

We have devices that are completely idle and wait for transaction from the server, otherwise they do not send anything.

They are in a separate VLAN, the routing interface is configured on the meraki core switch.

If the device is connected directly to core switch, the device is available for the server at all time.

If the device is on a different switch behind the core switch, the device will be offline after a while.

If I constantly ping the device, it will also be available at all time.

It looks like, the core switch "forgets" where to send the packet after a while.

Maybe the arp request gets lost or so...

The request from the server to the device comes from outside of the LAN.

We are facing this issue with different vendors and in different networks, so I would guess it is not a problem of the devices, I would guess it might be a configuration issue or firmware problem...

Any idea what could be the problem here?

Best

Fabian

[alemabrahao]

Well, the most logical explanation is that, given this behavior, it seems you don't have the VLAN added to the trunk on the core switch and the access switch.

What is your topology? Can you see the server in the ARP table when it's connected to the other switch?

The problem is probably a small detail you're forgetting.

Please provide more information than you've already verified.

[Fabian1]

The server is on AWS, connected via MPLS to our Network. there is a route to the subnet that points to the vlan interface of the switch..

On the switch I have two IP interfaces, on that is in the same network of the mpls router and the other one is the gateway of the problem device.

The other thing is, when we unplug the device and reconnect it, it works for some minutes again until it gets unreachable. And as long as I ping the terminal, we can keep the connection...

We have this behavior with different vendors of terminals, so it cant be the client.

[cmr]

What switch models and firmware are you running?  I've seen it before where the devices dropped off the client table, but still worked due to to internet usage and there could be a bug for them falling out of the arp table, but we'd need to know more to be sure.

[Fabian1]

We are on MS 18.1.5.1

Switch Models: MS210, MS120

I think there is no option to set a static mac address into the table right? 

DAI is disabled