Fix itermittent error installing Juju

We originally made the decision to disable apt update and upgrades for
faster machine provisioning, but we've experienced the scenario where
the provisioning of a container will fail because the apt cache in the
lxd image is stale. This can cause the cloud-init used by Juju to get
stuck in a loop trying to install packages that no longer exist in the
Ubuntu Archive.

This patch enables the run of `apt-get update` when Juju provisions a
container. This will add some time to the provisioning process, but
ensures greater reliability in the process.

Signed-off-by: Adam Israel <adam.israel@canonical.com>

diff --git a/jenkins/VCA/start_build b/jenkins/VCA/start_build
index 6cc5e30..fc91aad 100755 (executable)
--- a/jenkins/VCA/start_build
+++ b/jenkins/VCA/start_build
@@ -81,7 +81,7 @@ cp $HERE/update-lxd-image.sh /etc/cron.daily
 INFO "Bootstrapping VCA"
 juju bootstrap localhost osm \
 --config default-series=xenial \
---config enable-os-refresh-update=false \
+--config enable-os-refresh-update=true \
 --config enable-os-upgrade=false \
 --config apt-http-proxy=http://10.44.127.1:8000