Problem:
The connection to guest VMs from the host network, apart form the KVM host itself, is painfully slow.
It's only 65 KBits and it also appears to jam the network, as it seems to drop lots of packages.
Details:
![]()
The routing table on the 10.100.200.10 machine (KVM host) is below.
Machines from within 10.100.200.0/24 network can now discover and connect to the VMs in the 10.11.12.0/24 network.
However the connection speed is very slow (about 65 KBits), when a machine from within the 10.100.200.10/24 network initiates the connection and sends data to the VM (10.11.12.20). The problem might also be of existence for the opposite direction but haven't been able to test this yet.
When I do a ping from within the 10.100.200.0/24 network to the NIC 10.11.12.20, I get the "Redirect Host(New nexthop: 10.100.200.10)" message.
I run traceroute from my laptop and it appears that it routes to the expected IPs but shows where the deplays are introduced.
Sometimes it also times out (it displays a * rather than the timings).
Traceroute when it didn't time out:
Traceroute when it timed out:
Question:
As far as I can see, I guess that the dd-wrt router is configured correctly but the KVM host (IP 10.100.200.10) requires some tweaking.
Is there a way of making iptables and route changes on the KVM host (IP 10.100.200.10) to address this problem?
Note:
I am aware of the approach of bridging the NIC and then use the bridge directly with KVM but then I would be bound to use the network within which the KVM host resides in. Network separation is required!
Thanks for your help!
The connection to guest VMs from the host network, apart form the KVM host itself, is painfully slow.
It's only 65 KBits and it also appears to jam the network, as it seems to drop lots of packages.
Details:
- Ubuntu 14.04.1
- KVM/ QEMU
- The pysical network is a Gigabit Ethernet
- The guest VMs are connected to virtual networks with the modus "Route".
- My home office intranet (10.100.200.0/24) has the DD-WRT router as gateway configured. Within this intranet I have got a Linux machine which runs KVM/ QEMU. This machine has a virtual network (10.11.12.0/24) to which all its guest VMs are connected to.
- The KVM host machine has the IP 10.100.200.10.
- To make the VM guests discoverable, I added a static route to the dd-wrt router (Setup > Advanced Routing) with the settings below.
Code:
Destination LAN NET: 10.11.12.0
Subnet Mask: 255.255.255.0
Gateway: 10.100.200.10
Interface: LAN & WLANCode:
$ ip route show
default via 10.100.200.1 dev em1
10.11.12.0/24 dev virbr1 proto kernel scope link src 10.11.12.1
10.100.200.0/21 dev em1 proto kernel scope link src 10.100.200.10However the connection speed is very slow (about 65 KBits), when a machine from within the 10.100.200.10/24 network initiates the connection and sends data to the VM (10.11.12.20). The problem might also be of existence for the opposite direction but haven't been able to test this yet.
When I do a ping from within the 10.100.200.0/24 network to the NIC 10.11.12.20, I get the "Redirect Host(New nexthop: 10.100.200.10)" message.
Code:
$ ping 10.11.12.20
PING 10.11.12.20 (10.11.12.20) 56(84) bytes of data.
64 bytes from 10.11.12.20: icmp_seq=1 ttl=63 time=13.5 ms
From 10.145.24.1: icmp_seq=2 Redirect Host(New nexthop: 10.100.200.10)
64 bytes from 10.11.12.20: icmp_seq=2 ttl=63 time=13.5 ms
From 10.145.24.1: icmp_seq=3 Redirect Host(New nexthop: 10.100.200.10)
64 bytes from 10.11.12.20: icmp_seq=3 ttl=63 time=13.6 ms
From 10.145.24.1: icmp_seq=4 Redirect Host(New nexthop: 10.100.200.10)
64 bytes from 10.11.12.20: icmp_seq=4 ttl=63 time=13.5 ms
From 10.145.24.1: icmp_seq=5 Redirect Host(New nexthop: 10.100.200.10)
64 bytes from 10.11.12.20: icmp_seq=5 ttl=63 time=13.5 msSometimes it also times out (it displays a * rather than the timings).
Traceroute when it didn't time out:
Code:
$ traceroute 10.11.12.20
traceroute to 10.11.12.20 (10.11.12.20), 30 hops max, 60 byte packets
1 10.100.200.1 (10.100.200.1) 0.387 ms 0.706 ms 0.796 ms
2 10.100.200.10 (10.100.200.10) 13.411 ms 25.857 ms 38.277 ms
3 10.11.12.20 (10.11.12.20) 50.970 ms 63.466 ms 75.893 msCode:
$ traceroute 10.11.12.20
traceroute to 10.11.12.20 (10.11.12.20), 30 hops max, 60 byte packets
1 10.100.200.1 (10.100.200.1) 0.359 ms 0.640 ms 0.799 ms
2 10.100.200.10 (10.100.200.10) 13.656 ms 26.051 ms 38.410 ms
3 10.11.12.20 (10.11.12.20) 51.140 ms 63.482 ms *As far as I can see, I guess that the dd-wrt router is configured correctly but the KVM host (IP 10.100.200.10) requires some tweaking.
Is there a way of making iptables and route changes on the KVM host (IP 10.100.200.10) to address this problem?
Note:
I am aware of the approach of bridging the NIC and then use the bridge directly with KVM but then I would be bound to use the network within which the KVM host resides in. Network separation is required!
Thanks for your help!