bionic: netplan field bonds not found

[Vidmantas]

During openstack deployment juju gives error on all lxd containers.

Example:

sudo juju status

3/lxd/0 down pending bionic bridge activation error: yaml: unmarshal errors:
  line 7: field bonds not found in struct netplan.

yaml:

openstack-dashboard:
    annotations:
      gui-x: '500'
      gui-y: '-250'
    charm: cs:~openstack-charmers-next/bionic/openstack-dashboard
    num_units: 1
    bindings:
      website: web
    options:
      openstack-origin: distro-proposed
    to:
    - lxd:3

All servers have one default nic for native vlan and 2 nics in bond.

 I ssh to sever that holds 3/lxd/0
brctl show

bridge name bridge id STP enabled interfaces
lxdbr0 8000.000000000000 no

lxc list
If this is your first time running LXD on this machine, you should also run: lxd init
To start your first container, try: lxc launch ubuntu:16.04

+------+-------+------+------+------+-----------+
| NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS |
+------+-------+------+------+------+-----------+

I attached yaml from this server.

[John A Meinel (jameinel)]

Is that the only yaml file in /etc/netplan ? because there are no Bonds defined there that we would be processing.

I believe there are other locations where netplan config can exist. I don't remember offhand. I think maybe /var/run and /run ?

[Vidmantas (vidmantasvgtu)]

 sudo juju status
7 started 158.129.209.18 c4yk3b bionic default Deployed
7/lxd/0 down pending bionic bridge activation error: yaml: unmarshal errors:
  line 7: field bonds not found in struct netplan.Network
  line 7: field vlans not found in struct netplan.Network
ssh into 158.129.209.18

attached yaml

This one has bonds in it

[Vidmantas (vidmantasvgtu)]

Any news?

[Vidmantas (vidmantasvgtu)]

Keeping bug alive.

[John A Meinel (jameinel)]

sorry about the delay in responding, were occupied with a management sprint.
This is a high priority for us to fix, and is likely to be fixed for the next 2.3 stable release.

[Vidmantas (vidmantasvgtu)]

Why milestone changed to none?

[John A Meinel (jameinel)]

2.4-beta2 has been released, so any bugs targeted at clearly won't actually
be addressed in that release.
Ideally we will still target the next 2.4 release, but the bot doesn't know
that.

On Sat, May 12, 2018 at 12:36 AM, Vidmantas <email address hidden>
wrote:

> Why milestone changed to none?
>
> --
> You received this bug notification because you are subscribed to juju.
> Matching subscriptions: juju bugs
> https://bugs.launchpad.net/bugs/1764735
>
> Title:
> bionic: netplan field bonds not found
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/juju/+bug/1764735/+subscriptions
>

[John A Meinel (jameinel)]

https://github.com/juju/juju/pull/8697

[John A Meinel (jameinel)]

https://github.com/juju/juju/pull/8708 merges 2.3 into develop

[Vidmantas (vidmantasvgtu)]

Cool. I am new to juju. I installed it with snap install juju --classic. I have version 2.3.7 stable.

How much on average I would have to wait for stable release with this fix? I assume I could try it with development release? using snap install juju --edge --classic.

[John A Meinel (jameinel)]

It should be available as --candidate, so you should be able to:
 snap refresh juju --classic --candidate

(We automatically build candidate stable to 'candidate', and we build the develop branch into --edge)

$ snap info juju
name: juju
summary: juju client
publisher: canonical
contact: http://jujucharms.com
license: Other Open Source
description: |
  Through the use of charms, juju provides you with shareable, re-usable, and repeatable expressions of
  devops best practices.
snap-id: e2CPHpB1fUxcKtCyJTsm5t3hN9axJ0yj
channels:
  stable: 2.3.7 (4152) 54MB classic
  candidate: 2.3.8+2.3-6caee25 (4311) 55MB classic
  beta: 2.4-beta2 (4242) 55MB classic
  edge: 2.4-rc1+develop-192d958 (4315) 63MB classic

If you used --edge, then you would get the 2.4 pre release.

[John A Meinel (jameinel)]

We would love to have feedback on whether it works for you before we release, but if you want to wait for a 2.3.8 stable release we expect to release it later this week after we get feedback from others.

[Vidmantas (vidmantasvgtu)]

Ok, thank you. I will try to deploy 2.3.8 and see how it goes.

[Vidmantas (vidmantasvgtu)]

Now I get error:
7/lxd/0 down pending bionic failed to start machine 7/lxd/0 (can't get info for image 'juju/bionic/amd64': not found), retrying in 10s (8 more attempts)

juju has access to internet and can reach MAAS.

juju version 2.3.8-artful-amd64

[Vidmantas (vidmantasvgtu)]

After deployment server has in netplan folder:
50-cloud-init.yaml.bak.1526541370 98-juju.yaml.bak.1526541394 98-juju.yaml.bak.1526541420 99-juju.yaml 99-juju.yaml.bak.1526541382 99-juju.yaml.bak.1526541407

lxc list
If this is your first time running LXD on this machine, you should also run: lxd init
To start your first container, try: lxc launch ubuntu:16.04

+------+-------+------+------+------+-----------+
| NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS |
+------+-------+------+------+------+-----------+

 no containers created

[Vidmantas (vidmantasvgtu)]

2018-05-17 07:20:33 DEBUG juju.cloudconfig.containerinit container_userdata.go:162 generated network config:

auto lo {eth00_16_3e_78_08_a5}

iface lo inet loopback
  dns-nameservers 10.200.0.2 8.8.8.8
  dns-search maas

iface {eth00_16_3e_78_08_a5} inet static
  address 10.202.0.35/16

2018-05-17 07:20:33 INFO juju.cloudconfig.containerinit container_userdata.go:165 generated network config has no ipv4 gateway
2018-05-17 07:20:33 DEBUG juju.cloudconfig.containerinit container_userdata.go:182 generating netplan from container.NetworkConfig{NetworkType:"bridge", Device:"lxdbr0", MTU:0, Interfaces:[]net$
2018-05-17 07:20:33 DEBUG juju.service discovery.go:63 discovered init system "systemd" from series "bionic"
2018-05-17 07:20:33 INFO juju.cloudconfig userdatacfg_unix.go:460 Fetching agent: curl -sSfw 'agent binaries from %{url_effective} downloaded: HTTP %{http_code}; time %{time_total}s; size %{siz$
2018-05-17 07:20:33 INFO juju.container.lxd lxd.go:192 instance "juju-ee4066-0-lxd-2" configured with map[eth0:map[name:eth0 parent:br-ens3.402 hwaddr:00:16:3e:78:08:a5 mtu:1500 type:nic nictyp$
2018-05-17 07:20:33 INFO juju.container.lxd lxd.go:203 starting instance "juju-ee4066-0-lxd-2" (image "juju/bionic/amd64")...
2018-05-17 07:20:33 ERROR juju.provisioner provisioner_task.go:932 cannot start instance for machine "0/lxd/2": can't get info for image 'juju/bionic/amd64': not found

[Vidmantas (vidmantasvgtu)]

Server reaches proxy, but cat get image

[Vidmantas (vidmantasvgtu)]

issue is during deployment no gateway is set even if MAAS is configured

[John A Meinel (jameinel)]

Are you setting "http-proxy" and "https-proxy" for Juju? You need to set it
both for MAAS and for Juju.

juju bootstrap --model-default http-proxy=PROXY --model-default
https-proxy=PROXY

John
=:->

On Thu, May 17, 2018 at 1:26 PM, Vidmantas <email address hidden>
wrote:

> issue is during deployment no gateway is set even if MAAS is configured
>
> --
> You received this bug notification because you are a bug assignee.
> Matching subscriptions: juju bugs
> https://bugs.launchpad.net/bugs/1764735
>
> Title:
> bionic: netplan field bonds not found
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/juju/+bug/1764735/+subscriptions
>

[Vidmantas (vidmantasvgtu)]

Yes, thank you. I tested small deployments and vlan now deploys just fine.

[Ryan Finnie (fo0bar)]

Confirmed that 2.3.8 + bionic + bonding + lxd works.