cloud-init does not bring up all bridge interfaces created with openvswitch

[gadLinux]

Hi,

In my systems (all of them where I have openstack), I found that cloud-init does not perform well. Mainly because the bridges I have were created with ovs.

Since they are not ready at boot time it seems that the system got stalled for a few mins. And after boot not all services recover.

For example rbdmap is not called again after the net is up.

It may be the order where the things are done. Or maybe something worse. The problem is that cloud-init stalls for 2 mins on boot. RBD doesn't run because cloud-init starts in nonet mode. And the whole system becomes broken.

It's curious that ceph-osd are re-run after net is up. So some scripts are relaunched. In fact two times. So they start to work well. But some others no.

root@red-compute:~# ovs-vsctl show
8d08d8e4-49f2-4243-b1db-7641984a8530
    Bridge br-ex
        Port br-ex
            Interface br-ex
                type: internal
    Bridge br-int
        Port "eth1"
            Interface "eth1"
        Port br-int
            Interface br-int
                type: internal
    ovs_version: "2.0.1"

root@red-compute:~# cat /etc/network/interfaces
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

# The primary network interface
auto eth0
iface eth0 inet manual
 up ifconfig $IFACE 0.0.0.0 up
        up ip link set $IFACE promisc on
        down ip link set $IFACE promisc off
        down ifconfig $IFACE down

auto eth1
iface eth1 inet manual
 up ifconfig $IFACE 0.0.0.0 up
        down ifconfig $IFACE down

auto br-int
iface br-int inet dhcp

And when I boot up, cloudinit does not find the interfaces so no network is up.

I have to take an screenshot of the issue. The problem is that rbd storage is not boot up because cloudinit does not find the interfaces to connect to the intranet (because it's a bridge), and all the system becomes a nightmare.

[gadLinux (gad-aguilardelgado)]

This is a problem when installing metal with mass.

This is related to this
https://bugs.launchpad.net/juju-core/+bug/1271144

but not resolved.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[gadLinux (gad-aguilardelgado)]

It seems it's related to this but also:

https://bugs.launchpad.net/cloud-init/+bug/1345433

[James Page (james-page)]

Marking this bug as Fix Released - alot of effort has been invested into first class support for OVS based networking across MAAS, cloud-init and netplan.io over the last two years and I now believe this issue to be resolved.

Please re-test and re-open the issue if its still a problem.