lxc containers failing dhcp
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
lxc (Ubuntu) |
Confirmed
|
Undecided
|
Unassigned |
Bug Description
I just noticed this about a week ago when I was leaving Hong Kong after the ODS. My older containers (e.g. precise, or lucid) are failing to bring up networking. Checking syslog in the container:
Nov 23 06:39:59 subunit-lucid dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 8
Nov 23 06:39:59 subunit-lucid dhclient: 5 bad udp checksums in 5 packets
Nov 23 06:40:07 subunit-lucid dhclient: No DHCPOFFERS received.
Nov 23 06:40:07 subunit-lucid dhclient: No working leases in persistent database - sleeping.
Nov 23 06:44:03 subunit-lucid dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 4
Which makes me think that the host lxc bridge no longer does checksum fixups, which will break older containers. Without networking updating the containers to do a local fix is uhm, hard - if I'm right there's a bunch of folk out there scratching their heads and recreating containers (with appropriate swear words...)
Indeed - running
sudo iptables -t mangle -A POSTROUTING -o lxcbr0 -p udp -m udp --dport 68 -j CHECKSUM --checksum-fill
In the host permits networking to work correctly.