cloud-init-nonet causes waits for 120 seconds on reboots

Bug #1597369 reported by Alex Schultz
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Invalid
Medium
Fuel Sustaining
8.0.x
Confirmed
Medium
Fuel Sustaining
Mitaka
Invalid
Medium
Fuel Sustaining

Bug Description

We addressed this as part of Bug 1442572 but the purged option causes issue with unit testing in Bug 1493985. We switch the cloud-init from purged to absent which is removing the package but unfortunately leaving the scripts in place so the 120 second boot delay is still there. We need to properly cleanup cloud-init after the installation of the node.

On 7.0 it was
/var/log/apt/history.log
Start-Date: 2015-10-05 22:24:32
Commandline: /usr/bin/apt-get -y -q remove --purge cloud-init
Purge: cloud-init:amd64 (0.7.5-0ubuntu1.12)
End-Date: 2015-10-05 22:24:32

On 8.0
/var/log/apt/history.log
Start-Date: 2016-03-16 13:59:51
Commandline: /usr/bin/apt-get -y -q remove cloud-init
Remove: cloud-init:amd64 (0.7.5-0ubuntu1.17)
End-Date: 2016-03-16 13:59:52

So on 8.0 cloud-init is actually present:
dpkg -L cloud-init
/etc
/etc/rsyslog.d
/etc/rsyslog.d/21-cloudinit.conf
/etc/profile.d
/etc/profile.d/Z99-cloud-locale-test.sh
/etc/init
/etc/init/cloud-init-local.conf
/etc/init/cloud-final.conf
/etc/init/cloud-init.conf
/etc/init/cloud-log-shutdown.conf
/etc/init/cloud-init-nonet.conf
/etc/init/cloud-config.conf
/etc/init/cloud-init-container.conf
...

And nodes are waiting for it on reboot... And if there is a metadata server in the network it may response with the script and, e.g. wipe disks on the node)

description: updated
Revision history for this message
Alexander Bozhenko (alexbozhenko) wrote :
Revision history for this message
Alexander Bozhenko (alexbozhenko) wrote :

root@node-2:~# grep 'start on' /etc/init/cloud*
/etc/init/cloud-config.conf:start on (filesystem and started rsyslog)
/etc/init/cloud-final.conf:start on (stopped rc RUNLEVEL=[2345] and stopped cloud-config)
/etc/init/cloud-init.conf:start on mounted MOUNTPOINT=/ and stopped cloud-init-nonet
/etc/init/cloud-init-container.conf:start on container
/etc/init/cloud-init-local.conf:start on mounted MOUNTPOINT=/ and mounted MOUNTPOINT=/run
/etc/init/cloud-init-nonet.conf:start on mounted MOUNTPOINT=/ and stopped cloud-init-local
/etc/init/cloud-log-shutdown.conf:start on runlevel PREVLEVEL=2

root@node-2:~# rm /etc/init/cloud-*
Solved the problem for me.

Revision history for this message
Alex Schultz (alex-schultz) wrote :

Was actually fixed in 9 as part of https://github.com/openstack/fuel-library/commit/b0c89aa6c66af35f6bd8337a66fc82ba554096d7 where we explicitly disable the cloud-init services. So this only affects 8.0

Changed in fuel:
status: Confirmed → Invalid
summary: - cloud-init configurations need to be purged
+ cloud-init-nonet causes waits for 120 seconds on reboots
tags: added: support
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.