Temporarily loss of connection between 2 vm on the same host

[crazy-to-bike]

Hello,

I've got 2 VMs installed on a Proxmox-Server. An IPCop and a Debian server. The internet connection of the debian server is realized through the IPCop. So both have configured vmbr1 to communicate. Both are configured with virtio nics.
vmbr1 is also connected to a real nic in the Proxmox-Server and this nic is connected to a switch. A lot of PCs connect to the virtualized Proxmox server. Internet connections are forwarded to the IPCop.
But sometimes the users get timeouts at their browser, and nagios (installed on the Deban server) reports errors with the IPCop connection:

Notification Type: PROBLEM Service: SYS - host up/ping Host: Firewall: paedML Linux 5.1.0 State: CRITICAL for 0d 0h 1m 14s Address: 10.16.1.254 Info: PING CRITICAL - Packet loss = 12%, RTA = 4752.27 ms

or

Notification Type: PROBLEM Service: WEB - squid Host: Firewall: paedML Linux 5.1.0 State: CRITICAL for 0d 0h 3m 2s Address: 10.16.1.254 Info: Connection refused

Pings from the Debian Server to the IPCop are very inregularly:

ping 10.16.1.254
PING 10.16.1.254 (10.16.1.254) 56(84) bytes of data.
64 bytes from 10.16.1.254: icmp_seq=1 ttl=64 time=0.000 ms
64 bytes from 10.16.1.254: icmp_seq=2 ttl=64 time=0.000 ms
64 bytes from 10.16.1.254: icmp_seq=3 ttl=64 time=0.455 ms
64 bytes from 10.16.1.254: icmp_seq=4 ttl=64 time=0.028 ms
64 bytes from 10.16.1.254: icmp_seq=5 ttl=64 time=0.553 ms
64 bytes from 10.16.1.254: icmp_seq=6 ttl=64 time=0.415 ms
64 bytes from 10.16.1.254: icmp_seq=7 ttl=64 time=0.621 ms
64 bytes from 10.16.1.254: icmp_seq=8 ttl=64 time=7.07 ms
64 bytes from 10.16.1.254: icmp_seq=9 ttl=64 time=0.039 ms
64 bytes from 10.16.1.254: icmp_seq=10 ttl=64 time=7.27 ms
64 bytes from 10.16.1.254: icmp_seq=11 ttl=64 time=0.044 ms
64 bytes from 10.16.1.254: icmp_seq=12 ttl=64 time=0.525 ms
64 bytes from 10.16.1.254: icmp_seq=13 ttl=64 time=7.42 ms
64 bytes from 10.16.1.254: icmp_seq=14 ttl=64 time=0.047 ms
64 bytes from 10.16.1.254: icmp_seq=15 ttl=64 time=0.206 ms
64 bytes from 10.16.1.254: icmp_seq=16 ttl=64 time=0.640 ms
64 bytes from 10.16.1.254: icmp_seq=17 ttl=64 time=0.147 ms
64 bytes from 10.16.1.254: icmp_seq=18 ttl=64 time=0.418 ms
64 bytes from 10.16.1.254: icmp_seq=19 ttl=64 time=7.33 ms
64 bytes from 10.16.1.254: icmp_seq=20 ttl=64 time=0.047 ms

A user of the same system virtualized with Proxmox reported the same problems and that he was forced to install the IPCop on a real machine.
It would be a pitty if I really had to do that to have a stable connection between the Debian server and the IPCop.

What else could I do?

 

[crazy-to-bike]

Both are configured with virtio nics.

correction: The IPCop is configured with Intel E1000 because virtio doesn't work.

So I matter if it would be better to configure the Debian VM also with Intel E1000. But I don't think that this would make a difference.