Packet loss on some guests

Jan 31, 2014
113
10
38
Recently installed Proxmox VE 6.3. Two guests work OK, two have > 50 % packet loss so not really useful.
I can not understand what is wrong with the two problem hosts. Both of them I moved today from Proxmox 5.4-15, dumped and restored on this newer proxmox. Of the two working hosts one I created as KVM guest, the other was transferred from that older proxmox two days ago. Guess are Debian 9, 10 or Devuan 2.2.

Attached ping.txt shows packet loss, ping every 1 seconds. For about 10 seconds it works, then for about 20 seconds it does not work and this repeats.
ping2.txt shows pinging another VM on same subnet works, packet loss is 0%.

How to fix this so all the guests also access hosts outside the local subnet?

I have already tried:
  • deleted the network device and added it back to force MAC change
  • changed network device from VirtIO to Intel E1000
  • compared settings between the working guests and not working guests
Network configuration on host:
cat /etc/network/interfaces auto lo iface lo inet loopback iface enp195s0 inet manual auto vmbr0 iface vmbr0 inet static address 33.44.55.94/26 gateway 33.44.55.65 bridge_ports enp195s0 bridge_stp off bridge_fd 0 iface enp5s0f3u2u2c2 inet manual auto vmbr1 iface vmbr1 inet static address 66.77.8896 netmask 255.255.255.224 bridge_ports none bridge_stp off bridge_fd 0
 

Attachments

  • ping2.txt
    6.2 KB · Views: 9
  • ping.txt
    2.2 KB · Views: 12
I did more testing. The guests I had created previously either by creating VM or copying dump from Proxmox 5.4 do work. It is only the two guests I copied from Proxmox 5.4 and restored yesterday that have big packet loss.
I created a new VM just now, Debian 10. It does not have packet loss.
It looks to me the problem is inside Proxmox, when pinging the host IP that acts as gateway the packet loss is > 50 %. So traffic is not disrupted in providers (which is Hetzner) network.
 
mhmm... maybe a problem in the network? duplicate mac/ips can also have such an effect..
 
I upgraded the devuan guest from 2.1 to 3.0, kernel changed from 4.9 to 4.19. No packet loss so far with this kernel. Tested with ping -c 100 and mtr -s 1000 -r -c 1000.
I'll try installing Debian 10 kernel on the Debian 9 problem guests next.
 
Do you have ifupdown2 package installed?
No.
Tested Debian Stretch with stock kernel, mtr shows packet loss 52%.
Installed kernel 4.19 from stretch-backports, after booting with that no packet loss with mtr 1000 run.
It looks to me some kernels on guest do not work on curret Proxmox.
 
mhmm... i cannot reproduce this here, fresh debian9 install (fully upgraded) on a fully upgraded pve 6 host, i have no problem with packet loss whatsoever
 
Hello!
I have the same problem. Proxmox 6.3 installed 3 day ago. VM - new Debian 10, Debian restored from Proxmox 6.2, Ubuntu restored from 6.2. All these virtual machines have lost ping packets. I try to use e1000 and virtio. Always the same :( :(

4 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=100 ttl=116 time=7.58 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=101 ttl=116 time=7.61 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=102 ttl=116 time=7.57 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=103 ttl=116 time=7.56 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=104 ttl=116 time=7.52 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=105 ttl=116 time=8.02 ms
^C
--- onet.pl ping statistics ---
121 packets transmitted, 87 received, 28.0992% packet loss, time 920ms
 
Hello!
I have the same problem. Proxmox 6.3 installed 3 day ago. VM - new Debian 10, Debian restored from Proxmox 6.2, Ubuntu restored from 6.2. All these virtual machines have lost ping packets. I try to use e1000 and virtio. Always the same :( :(

4 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=100 ttl=116 time=7.58 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=101 ttl=116 time=7.61 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=102 ttl=116 time=7.57 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=103 ttl=116 time=7.56 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=104 ttl=116 time=7.52 ms
64 bytes from aafc88a28d9997374.awsglobalaccelerator.com (75.2.92.173): icmp_seq=105 ttl=116 time=8.02 ms
^C
--- onet.pl ping statistics ---
121 packets transmitted, 87 received, 28.0992% packet loss, time 920ms

Hello,
I'm also have this problem with proxmox 6.3-6
Under Debian 10 & Ubuntu 16.04 ping IP get about 10% packet loss.
But If I ping the same IP from Proxmox host, it doesn't have any packet loss.
 
I'd li
mhmm... maybe a problem in the network? duplicate mac/ips can also have such an effect..
Dear dcsapak,

I have had this problem for some time and lost couple of days trying to figure it out.. only to realize that I did indeed restore machine from backup with the same MAC address. THANK YOU. I am an admin for many years but sometimes I get foggy in my brain. Thank you again.
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!