Bootstrap failed: DNS/routing misconfigured on maas

[Curtis Hovey]

As seen at
    http://reports.vapour.ws/releases/issue/57291641749a564c0ee11a72

Juju 1.25.5 can bootstrap and deploy to maas 1.8, but juju 1.25.6 cannot. The bootstrap node cannot reach http://archive.ubuntu.com/ for example:

ubuntu@maas-node-3:~$ tracepath 10.125.0.13
 1?: [LOCALHOST] pmtu 1500
 1: 10.0.200.137 2999.078ms !H
     Resume: pmtu 1500
ubuntu@maas-node-3:~$ wget http://archive.ubuntu.com/
--2016-05-03 20:56:43-- http://archive.ubuntu.com/
Resolving archive.ubuntu.com (archive.ubuntu.com)...

Note that code backported from juju 2 may assume a default-gateway is configured in maas, but this was never true for maas 1.8

[James Tunnicliffe (dooferlad)]

Please attach /var/log/cloud-init* and the output of:
ip -d route
ip -d addr

The contents of /etc/resolv.conf and, for each address in resolv.conf 'ip -d route get <address>'.

I have been setting up a MAAS 1.8 today but it has been slow going (http://images.maas.io/ being really slow earlier didn't help. Now back up to being plain slow).

[Curtis Hovey (sinzui)]

Hi James. This the output from the commands. I will attached the logs shortly.

ubuntu@maas-node-2:~$ ip -d route
default via 10.0.200.1 dev eth1
10.0.200.0/24 dev juju-br0 proto kernel scope link src 10.0.200.136

ubuntu@maas-node-2:~$ ip -d addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 52:54:00:35:96:2c brd ff:ff:ff:ff:ff:ff
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master juju-br0 state UP group default qlen 1000
    link/ether 52:54:00:9d:91:bb brd ff:ff:ff:ff:ff:ff
    inet 10.0.200.136/24 brd 10.0.200.255 scope global eth1
       valid_lft forever preferred_lft forever
    inet6 fe80::5054:ff:fe9d:91bb/64 scope link
       valid_lft forever preferred_lft forever
4: juju-br0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 52:54:00:9d:91:bb brd ff:ff:ff:ff:ff:ff
    inet6 fe80::fc4f:b7ff:fef0:d742/64 scope link
       valid_lft forever preferred_lft forever

ubuntu@maas-node-2:~$ cat /etc/resolv.conf
# Dynamic resolv.conf(5) file for glibc resolver(3) generated by resolvconf(8)
# DO NOT EDIT THIS FILE BY HAND -- YOUR CHANGES WILL BE OVERWRITTEN
nameserver 10.125.0.13
search maas

ubuntu@maas-node-2:~$ ip -d route get 10.125.0.13
10.125.0.13 via 10.0.200.1 dev eth1 src 10.0.200.136
    cache

[Curtis Hovey (sinzui)]

Attached are the cloud-init logs from the failed bootstrap.

[James Tunnicliffe (dooferlad)]

Ah, looks like this is to do with recent bridge script changes that were too enthusiastic when moving addresses and routes around.

[James Tunnicliffe (dooferlad)]

Have confirmed with MAAS 1.9.

I have a fix for several issues around bridges and routes that should land soon, which should get this.

[James Tunnicliffe (dooferlad)]

http://reviews.vapour.ws/r/4902 fixes this.

[James Tunnicliffe (dooferlad)]

https://github.com/juju/juju/pull/5482 is running. Should merge for master soon. I am going to be AFK for a few days but will try and get this marked fix committed.