[lxc-users] LXD: network connectivity dies when doing lxc stop / lxc start

[Tomasz Chmielewski]

Here is a weird one, and most likely not LXD fault, but some issue with 
bridged networking.

I'm using bridge networking for all container. The problem is that if I 
stop and start one container, some other containers loose connectivity. 
They loose connectivity for 10, 20 secs, sometimes up to a minute.


For example:

# ping 8.8.8.8
(...)
64 bytes from 8.8.8.8: icmp_seq=16 ttl=48 time=15.0 ms
64 bytes from 8.8.8.8: icmp_seq=17 ttl=48 time=15.0 ms
64 bytes from 8.8.8.8: icmp_seq=18 ttl=48 time=15.1 ms
64 bytes from 8.8.8.8: icmp_seq=19 ttl=48 time=15.0 ms
64 bytes from 8.8.8.8: icmp_seq=20 ttl=48 time=15.0 ms

...another container stopped/started...
...40 seconds of broken connectivity...

64 bytes from 8.8.8.8: icmp_seq=60 ttl=48 time=15.1 ms
64 bytes from 8.8.8.8: icmp_seq=61 ttl=48 time=15.1 ms
64 bytes from 8.8.8.8: icmp_seq=62 ttl=48 time=15.1 ms
64 bytes from 8.8.8.8: icmp_seq=63 ttl=48 time=15.0 ms


Pinging the gateway dies in a similar way.

The networking is as follows:

containers - eth0, private addressing (192.168.0.x)
host - "ctbr0" - private address (192.168.0.1), plus NAT into the world


auto ctbr0
iface ctbr0 inet static
         address 192.168.0.1
         netmask 255.255.255.0
         bridge_ports none
         bridge_stp off
         bridge_fd 0


The only workaround seems to be arpinging the gateway from the container 
all the time, for example:

# arping 192.168.0.1

This way, the container doesn't loose connectivity when other containers 
are stopped/started.

But of course I don't like this kind of fix.

Is anyone else seeing this too? Any better workaround that constant 
arping from all affected containers?


Tomasz Chmielewski
https://lxadm.com