This reverts commit a253e50609017e40c53c29acc3c0231163689c14.
This is a prime suspect in our gate being terrible lately - newer CoreOS
does much more on boot, as well as runs more services. The timeline
roughly matches up.
Change-Id: I58e6ba9d7ac360df98d74746d47b3a76da6f99db
This fetches the latest CoreOS stable version and builds the image using
that version.
Co-Authored-By: Arun S A G <sagarun@gmail.com>
Depends-On: Idfdc54210e33c71719c7fd0c905d0b802809e173
Depends-On: I6a572bdd791841add913d33c5b79b59033c3f237
Change-Id: I7201450b21304fa29f22437c05da9e04daa4b9f8
Bring ironic-python-agent in line with the other ironic projects.
Stop ignoring all E12* errors except E129
Stop ignoring E711
Change-Id: Icb9bc198473d1b5e807c20869eb2af7f4d7ac360
The current script used to inject the IPA container into CoreOS is
not pep8 compliant, has requirements that are not in
global-requirements, and never has tests run. Now requirement on
plumbum is removed, script is pep8 compliant, flake8 runs against the
script when tests are run.
Closes-bug: #1337551
Change-Id: I14d3e46e60c4e072f34f80dc50d7214079953f49
Two reasons:
1) Use a build that is marked stable
2) Fixes an issue where boot hangs for minutes on updating CA
things, which enables much faster deploy times.
Change-Id: Ic7fca96f0dfe87065e2df5feaf2c29efc662a0e1
http://coreos.com/docs/cluster-management/setup/cloudinit-cloud-config/
describes how to use coreos-cloudinit to configure a node. Migrate to
using this instead of a single BASH script (which is deprecated).
This also means we need the latest version of CoreOS, so I've updated
coreos-oem-inject.py to pull from the most recent copies, and this
includes moving to the amd64-usr builds.
Change-Id: I620b8fec1bfc5d58b560434d17b4f617734fadb8
Using systemd-nspawn gives us more flexibility in what hardware the
agent can see. With this setup, the network is not virtualized, and the
/dev inside the container looks nearly identical to the one in the host.
Also added RestartSec=30s to the unit file because without this, systemd
will respawn the agent every 100ms if it dies (say, because DHCP hasn't
completed yet). Since it retries a max of 5 times per 60s, unless the restart value is
set higher it will be marked 'failed' forever and not retry. Adding this prevents that.
Change-Id: Ic64ab4b0824595683e5428464512594dd90eed83
Using docker export gives us only the top filesystem layer and not all
the layers used to build. Resulting image is about 20% smaller. Further
optimizations will be made in future commits via removing build time
deps and temporary files in the Dockerfile.
Change-Id: I0f7ad6d00c6f47836dcd71b05907076701bb2789
(cherry picked from commit 99dba22eaa3961234abde1294e55026fd2cc9d4d)