lp:~axwalk/juju-core/lp1258240-bootstrap-refresh-dnsname

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #1258240: juju 1.17.0 bootstrap on Hp fails

Critical

Fix Released

Link a bug report

Related blueprints

2131. By Andrew Wilkins on 2013-12-11

provider/common: refresh DNS name during waitSSH

Synchronous bootstrap starts by creating a new
instance, and then attempting to connect to the
instance's SSH port. The method we were using
assumed that the instance would be started with
a public address to begin with, which is not
always the case. We now refresh addresses during
bootstrap.

An alternative approach would be to initially
wait until an actually public address is available,
rather than a "most public" address as returned by
instance.SelectPublicAddress. I considered this to
be a bit more fragile: providers might not know
whether an address is "public" (i.e. all addresses
might be NetworkUnknown), and that approach also
wouldn't cater for changes to public addresses.

Fixes #1258240

2130. By Andrew Wilkins on 2013-12-10

[r=axwalk],[bug=1259180] cloudinit/sshinit: wait for cloud-init to finish

We must wait for cloud-init to finish doing its
thing, as it will make various modifications to
the base OS. One of the modifications is to set
the cloud-local apt mirror, which we'll make use
of during the synchronous bootstrap phase when
apt updating/upgrading.

Fixes #1259180

https://codereview.appspot.com/39940043/

2129. By William Reade on 2013-12-10

[r=fwereade],[bug=1259473] state: forced machine cleanup just sets Dead

fixes lp:1259473

https://codereview.appspot.com/39970043/

2128. By John A Meinel on 2013-12-10

[r=jameinel] Merge the 1.16 branch into trunk to get bug fixes for bug #1256179 and bug #1257481.

This allows upgrading from 1.16.0 => current to properly handle the maas-agent-name field (nil vs ""), and use a safer filter for openstack instance names, so that two similarly named environments won't accidentally kill other instances during 'destroy-environment'.

2127. By Dave Cheney on 2013-12-10

[r=dave-cheney],[bug=1258246] provider/common: always log bootstrap failure

Fixes LP #1258246

2126. By Andrew Wilkins on 2013-12-10

[r=axwalk] Remove secrets-pushing for new bootstraps, API

- environs/BootstrapConfig now no longer removes
  secret attributes from configuration. This is
  on account of configuration not being populated
  into cloud-init config now.
- juju/NewAPIConn now no longer pushes secrets,
  as it is not necessary. This was not present in
  1.16, so it is not a compatibility concern.

If we're okay with disabling new CLI from initialising
an old juju server, then we can also remove secrets
pushing from juju/NewConn. Doing this would allow us
to drop all notion of "secret attributes".

https://codereview.appspot.com/37090043/

2125. By John A Meinel on 2013-12-08

[r=jameinel],[bug=1257427] provider/common/boostrap.go: timeout sensibly

The new synchronous bootstrap tries indefinitely until it is able to
get a DNSName and then connect to the machine that should be starting
up (bug #1257427)

This changes it to use the same timeout that we have for connecting to
the DB or API (10 minutes). It adds some bits that make it possible to
set the timeout to other values (which makes nice tests), and then
changes a few of the types in use into smaller interfaces
(instance.Instance into DNSNamer and os.File into io.Writer), which
also makes it easier to test.

https://codereview.appspot.com/37100043/

2124. By Andrew Wilkins on 2013-12-06

[r=axwalk],[bug=1167616] state: update SetEnvironConfig to take new and old

SetEnvironConfig is updated to take both new and old
configurations, and a delta computed from these. The
only functional change for now is that attributes that
exist in old but not in new are now deleted from state.

To do this right, we should also check that the
settings in state are the same as the "old" config,
so the delta computed is what the user expects. I have
not done this part yet, as it is a fairly big change
and I am not confident I understand all of the
repercussions.

https://codereview.appspot.com/38180043/

2123. By Tim Penhey on 2013-12-06

[r=thumper] Make KVM respect constraints.

The uvtool CLI allows us to specify memory, disk
and cores on the command-line. Constraints often
ask for a subset (or superset) of these.

This branch translates between the juju constraints
format to the command line params that the uvt-kvm
tool wants.

If constraints are supplied that KVM doesn't support,
a log entry is written out at INFO level. It isn't
really a warning, just an informational that we are
ignoring it.

Also, the local provider wasn't passing the constraints
through as the FinishMachineConfig only set the constraints
on the config object if it was a state server node. I am
a little unclear on the implications of this for other
providers.

https://codereview.appspot.com/37610043/

2122. By Tim Penhey on 2013-12-04

[r=thumper] StartContainer returns hardware characteristics.

Initial change to have the hardware characteristics
of containers returned. At this stage just the arch
is filled out.

https://codereview.appspot.com/36980043/