Hello, sir and madam.
This problem started to occur around April 29th. At that time, I opened Chrome on the Proxmox virtual machine and visited the website, and it reported that the connection timed out. It was still able to connect at the same time yesterday.
I then started a series of network troubleshooting actions, including reconfiguring the global network configuration and reinitializing the DHCP server (DHCPv4 for the PVE virtual machine, IPv6 was manually set). However, the connection timeout was still clearly displayed in front of me.
After such actions, I roughly found the root cause of the problem. On proxmox, curl accessed websites with only ipv4 addresses and they were all able to connect and obtain them normally; on the virtual machine, curl also accessed, and it fell into an endless black hole, trying to connect to this website, and it had exceeded the set maximum connection attempt time, but if you visit google.com, it can be obtained normally, because curl first discovered the ipv6 address and connected first. My idea is that the process of IPv4 external access of the virtual machine has been trapped in an endless black hole for some unknown reason, while IPv6 has no effect at all. The connection between the host and the virtual machine is normal (Telnet 10.0.0.1 8006), and the virtual machine wants to connect to the public network through the host's bridge for http access, but this idea is strongly "denied" by unknown reasons, resulting in inability to access the public network.
Because the computer room where the host is located has its own firewall and works normally, Proxmox does not need to turn on the firewall, and there is only a single public IPv4, so NAT is used to build the intranet, so iptables and isc-dhcp-server are used. (For example, 10.0.0.5 is the mailbox service, and 10.0.0.6 is the web service.)
Does anyone have suggestions? This weird failure is too torturous.
This problem started to occur around April 29th. At that time, I opened Chrome on the Proxmox virtual machine and visited the website, and it reported that the connection timed out. It was still able to connect at the same time yesterday.
I then started a series of network troubleshooting actions, including reconfiguring the global network configuration and reinitializing the DHCP server (DHCPv4 for the PVE virtual machine, IPv6 was manually set). However, the connection timeout was still clearly displayed in front of me.
After such actions, I roughly found the root cause of the problem. On proxmox, curl accessed websites with only ipv4 addresses and they were all able to connect and obtain them normally; on the virtual machine, curl also accessed, and it fell into an endless black hole, trying to connect to this website, and it had exceeded the set maximum connection attempt time, but if you visit google.com, it can be obtained normally, because curl first discovered the ipv6 address and connected first. My idea is that the process of IPv4 external access of the virtual machine has been trapped in an endless black hole for some unknown reason, while IPv6 has no effect at all. The connection between the host and the virtual machine is normal (Telnet 10.0.0.1 8006), and the virtual machine wants to connect to the public network through the host's bridge for http access, but this idea is strongly "denied" by unknown reasons, resulting in inability to access the public network.
Because the computer room where the host is located has its own firewall and works normally, Proxmox does not need to turn on the firewall, and there is only a single public IPv4, so NAT is used to build the intranet, so iptables and isc-dhcp-server are used. (For example, 10.0.0.5 is the mailbox service, and 10.0.0.6 is the web service.)
Does anyone have suggestions? This weird failure is too torturous.
Last edited: