Hi,
i have some kind of legacy setup, which i created with Proxmox 5: I have a bunch of LXC containers which are privileged. Inside the LXC containers, i have installed a docker, and some docker containers. The networking of all this is managed by a VM with OPNsense. This worked fine with Proxmox 5, but after the upgrade to 6 beta, i have a networking issue:
When the hypervisor has an uptime of around 2 days, the networking for the docker containers fails. The networking of the encapsulating LXC containers seems to continue just fine. However, the docker containers aren't able to produce any outbound traffic, neither ICMP or DNS via UDP. The errors of the ping command and the nslookup command shows an error "invalid argument" when creating the socket.
At the same time, the dmesg output of the hypervisor is full of arp table overflow messages. However, each LXC and the hypervisor has only 1-2 neighbors, so i dont think that the arp table really has an overflow.
At the same time, the hypervisor itself is working as expected. Proxmox Web Frontend is accessible, LXC containers are start- and stoppable, only the nested docker containers don't work.
Only solution for me right now is to do a full reboot of the hypervisor. However, this is merely a workaround, and i really want to solve this problem.
What can i do?
i have some kind of legacy setup, which i created with Proxmox 5: I have a bunch of LXC containers which are privileged. Inside the LXC containers, i have installed a docker, and some docker containers. The networking of all this is managed by a VM with OPNsense. This worked fine with Proxmox 5, but after the upgrade to 6 beta, i have a networking issue:
When the hypervisor has an uptime of around 2 days, the networking for the docker containers fails. The networking of the encapsulating LXC containers seems to continue just fine. However, the docker containers aren't able to produce any outbound traffic, neither ICMP or DNS via UDP. The errors of the ping command and the nslookup command shows an error "invalid argument" when creating the socket.
At the same time, the dmesg output of the hypervisor is full of arp table overflow messages. However, each LXC and the hypervisor has only 1-2 neighbors, so i dont think that the arp table really has an overflow.
At the same time, the hypervisor itself is working as expected. Proxmox Web Frontend is accessible, LXC containers are start- and stoppable, only the nested docker containers don't work.
Only solution for me right now is to do a full reboot of the hypervisor. However, this is merely a workaround, and i really want to solve this problem.
What can i do?