As a noob when it comes to Proxmox, I do not now where to look for the cause.
The symptoms/observations
- server with VMs and containers (lcx) worked fine for a couple of months. Lets call this one pve2
-- except for a problem with disk access, misconfiguration on my part, solved
- after a reboot this morning, TCP connection from my windows desktop to VMs get reset after 5~40 seconds
-- connection from my windows laptop last longer, about 2 minutes, then also get reset
- ssh connection to the server is solid, no resets (thanks goodness!)
Wireshark on the Windows machine sees the RST packets originating from the same (virtual) MAC as the TCP-traffic.
tshark on an affected VM does not see the RST packets.
TCP connection to port 22/ssh, port 80/http and other ports are affected in the same way: random resets.
Wait, what? How do I even run tshark on a VM if I can not ssh to the VM?
Well, the ssh to the server is solid and from there I can ssh to the VM. This connection is unaffected.
Quick check with another desktop in the same LAN (Laptop, wireless instead of wired): same effect, but seems less frequent.
I did this check because my Windows desktop is a VM running on top of another Proxmox VE installation (called pve3.
I like this setup. It is well worth the slight performance penalty).
To rule out something being wonky in the pve3+windows setup I used the laptop, which is still windows on bare metal.
I tried rebooting the pve2 server, the symptoms remained.
Where - and how - would I look for the origins of the RST packets resetting my TCP connections?
I am suspecting the virtual switch or firewall on the pve2 server - hence the post in the Proxmox forum.
But how to isolate?
merci
hase
The symptoms/observations
- server with VMs and containers (lcx) worked fine for a couple of months. Lets call this one pve2
-- except for a problem with disk access, misconfiguration on my part, solved
- after a reboot this morning, TCP connection from my windows desktop to VMs get reset after 5~40 seconds
-- connection from my windows laptop last longer, about 2 minutes, then also get reset
- ssh connection to the server is solid, no resets (thanks goodness!)
Wireshark on the Windows machine sees the RST packets originating from the same (virtual) MAC as the TCP-traffic.
tshark on an affected VM does not see the RST packets.
TCP connection to port 22/ssh, port 80/http and other ports are affected in the same way: random resets.
Wait, what? How do I even run tshark on a VM if I can not ssh to the VM?
Well, the ssh to the server is solid and from there I can ssh to the VM. This connection is unaffected.
Quick check with another desktop in the same LAN (Laptop, wireless instead of wired): same effect, but seems less frequent.
I did this check because my Windows desktop is a VM running on top of another Proxmox VE installation (called pve3.
I like this setup. It is well worth the slight performance penalty).
To rule out something being wonky in the pve3+windows setup I used the laptop, which is still windows on bare metal.
I tried rebooting the pve2 server, the symptoms remained.
Where - and how - would I look for the origins of the RST packets resetting my TCP connections?
I am suspecting the virtual switch or firewall on the pve2 server - hence the post in the Proxmox forum.
But how to isolate?
merci
hase