1

I have the following configuration:

  • OpenVZ host running CentOS 6 (2.6.32-042stab049.6), gigabit connection
  • A single container running on that OpenVZ

The host has full gigabit connection for in- and outbound connections. The containers have full gigabit for all outgoing traffic, but incoming traffic seems subject to a default traffic shaping rules where "bursts" are allowed. Downloading large binary files happens at 15MB/s for a few seconds, then it falls back to 50Kb/s.

That very same file can be downloaded on the host at full speed, over and over again.

Traffic shaping isn't enabled:

root@host:~# tc qdisc show dev venet0
root@host:~# tc class show dev venet0
root@host:~# tc filter show dev venet0

IPtables has nothing fancy:

root@host:~# iptables -L
Chain INPUT (policy ACCEPT)
target     prot opt source               destination   
Chain FORWARD (policy ACCEPT)
target     prot opt source               destination   
Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination

I'm actually finding myself a bit stuck and am considering this may just be a bug in the OpenVZ kernel for CentOS 6. I don't have the option to test it on CentOS 5 in the near future. Does anyone have any other ideas on how to troubleshoot? Or does it sound familiar? Given the fact that the host runs just find and only the CT's have the incoming bandwidth issue, I'm looking at OpenVZ as the cause.

Mojah
  • 1,373
  • 2
  • 12
  • 16
  • To fix my own solution: this seems to have been caused by the vmware VMXNET3 driver in combination with the CentOS6 kernel. Switching to the e1000 driver fixed it for now. – Mojah Feb 24 '12 at 14:10

0 Answers0