Reuse of IP address for a new container fail to ping gateway

[slamy]

Hi all,

I am in a strange situation. My setup is 3 PVE 6.3 with VLANs and bridges
Network configuration on nodes is as simple as giving a VLAN number to each physical interfaces when needed and create bridges based on these VLANs interfaces.

Then I create a container, like test-debian, based on official proxmox debian 10 template and configure on it one network interface (hold by the first proxmox bridge).
I configure the network interface with IP, netmask and gateway, it works!

But I realize that I have to redeploy this container, so I remove it and I create a new one, with exactly same hostname and same network configuration.

Unfortunately this new container is unable to ping its gateway...

With tcpdump I confirm that ICMP are received by the gateway but the container never receive the reply...

And, what is strange (or not I would prefer), is that if I ping some other IP, like 8.8.8.8, it works! And now the ping to the gateway is working too.

Anybody already had this behaviour? Anything special to do when a container is reinstalled with same network configuration?

Thanks everybody for the good work so far!

 

[Alwin]

From the top of my head. Try to move the CT to a different bridge and back. Re-check netmask, VLAN ID.

 

[slamy]

Well, thanks Alwin I did what you asked. Nothing happens, same unreachable container between itself and its gateway.
But I wonder if I setup VLANs the way it should be.
First I read the network configuration on official proxmox wiki.
So I setup VLANs only on the physical interfaces, with numbered interface to create VLANs, like eno1.2020
Then I setup the bridge, only to mention the bridge-ports as eno1.2020
Finaly I attach my container to this bridge with no more information about VLANs or physical interfaces used.

Is this setup right? Should I define my VLANs differently?

As I said, it only happens when I destroy a container, which is reachable with no problem and few minutes after I create a new one with the exact same network configuration. This new container is unable to reach the exact same gateway.

For the moment I will force same MAC address to the new container and see if it works...but not the most reliable setup



Thanks

 

[Alwin]

Is this setup right? Should I define my VLANs differently?

This can be done, and will limit the traffic to that VLAN. But the other network components need to know about the VLAN as well. And routing needs to be done between VLAN separated networks.

As I said, it only happens when I destroy a container, which is reachable with no problem and few minutes after I create a new one with the exact same network configuration. This new container is unable to reach the exact same gateway.

But this sounds like the MAC address might be cached somewhere and traffic is not passed back to that MAC.

 

[slamy]

Yes thank you, definitely the MAC address cache is causing the issue. Not Proxmox or LXC related but our switches keep this cache.
After a certain amount of time my new container is reachable with no hack!

You can close this ticket, thank you