Author: pcc
Revision Date: 2017-06-04 22:11:28 UTC

irgen: Create functions instead of global variables for builtin hash and equal algorithms.

These are in fact functions, and have function type, so it is invalid
to represent them using global variables.

Author: Andrew Wilkins
Revision Date: 2016-08-23 23:42:58 UTC

Merge pull request #56 from coryschwartz/master

python3 absolute import

Author: Andrew Wilkins
Revision Date: 2016-07-25 12:25:58 UTC

Use qualified model name in grant/revoke test

If you want to refer to another user's model,
you need to qualify the model with the model
owner's name. e.g. if I'm billie, and I want
to see cath's model foo, then I need to refer
to it as "cath/foo". Plain old "foo" is short
for "billie/foo".

Author: Andrew Wilkins
Revision Date: 2016-02-02 09:18:44 UTC

Initial commit

Author: Andrew Wilkins
Revision Date: 2015-11-24 02:20:32 UTC

Update

Author: Andrew Wilkins
Revision Date: 2015-11-23 09:13:18 UTC

Add osd-devices and osd-journal block storage

Add the "osd-devices" block-type storage, with
minimum of 0 and no maximum. Volumes assigned to
a unit will be added as OSDs.

Also, add the "osd-journal" block-type storage,
with minimum of 0 and maximum of 1. The osd-journal
storage, if supplied, will be used in favour of
osd-journal configuration if supplied.

For now, handling of osd-journal is static, just
as the configuration method was before. Removing
or adding the journal after deployment is not
currently supported. This is possible with further
changes, but requires stopping Ceph and migrating
the existing journal, and so out of scope here.

Author: Andrew Wilkins
Revision Date: 2015-09-25 03:56:18 UTC

Add osd-devices block storage

Add the "osd-devices" block-type storage,
with minimum of 0 and no maximum. Volumes
assigned to a unit will be added as OSDs.

Author: Andrew Wilkins
Revision Date: 2015-09-25 03:27:29 UTC

Add support for "storage" feature

This will cause a filesystem to be mounted
at "/var/lib/mysql" before installation.
If storage is unspecified, then rootfs will
be used, which would be comparable to existing
deployments.

The storage subordinate relation is retained
for backwards-compatibility.

Author: Andrew Wilkins
Revision Date: 2015-04-26 10:23:13 UTC

Don't create upstart config

Author: Andrew Wilkins
Revision Date: 2015-01-28 03:58:00 UTC

Add support for Juju storage

Author: Andrew Wilkins
Revision Date: 2015-01-05 01:15:37 UTC

Update README.md

Author: Andrew Wilkins
Revision Date: 2014-06-17 05:09:34 UTC

Preliminary relation addresses work

Author: Andrew Wilkins
Revision Date: 2014-05-29 08:05:45 UTC

Add Machine to uniter API

This change adds the Machine entity
to the uniter API, to support coming
changes to the uniter. The uniter
will become responsible for watching
machine addresses and updating
relations.

Also, drive-by consolidation of Watch
methods into api/common.

Author: Andrew Wilkins
Revision Date: 2014-05-22 04:00:33 UTC

provider/openstack: fix tests under gccgo

SameContents again.

Fixes lp:1321492

Author: Andrew Wilkins
Revision Date: 2014-04-22 11:07:23 UTC

environs/manual: SHELL=/bin/bash for new user

The ubuntu user should have its shell set to bash.

Fixes lp:1303195

Author: Andrew Wilkins
Revision Date: 2014-03-14 04:40:53 UTC

provider/azure: implement UnitAssigner policy

The Azure provider grows a new configuration
attribute, availability-sets-enabled, which
defaults to true but can be explicitly set to
false at (and only at) bootstrap time.

If the availability-sets-enabled is true, unit
placement is prohibited. This is necessary to
avoid various issues such as incorrectly load-
balancing ports across heterogenous services,
and makes it possible to unambiguously determine
which Cloud Service to assign units to for
high availability.

Author: Andrew Wilkins
Revision Date: 2014-03-14 02:40:04 UTC

Add/fix comments

Author: Andrew Wilkins
Revision Date: 2014-02-27 09:49:56 UTC

Merge trunk

Author: Andrew Wilkins
Revision Date: 2014-01-07 06:20:53 UTC

Fixes, finishing touches

Author: Andrew Wilkins
Revision Date: 2013-12-11 04:19:53 UTC

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

Author: Andrew Wilkins
Revision Date: 2013-11-25 07:10:02 UTC

Implement juju-provision plugin

This plugin provides a way of provisioning
a new machine instance with the prerequisites
for running a Juju agent, but without installing
the agent itself.

Author: Andrew Wilkins
Revision Date: 2013-11-14 03:39:24 UTC

provider/null: Validate must allow invalid old config

The EnvironmentSet is used to push secrets from the CLI
into state; before this, the config in state is invalid.

EnvironmentSet always calls the provider's Validate
method on the new and old config before setting the
config in state. Thus, the Validate method must cater
for the old config being invalid.

Author: Andrew Wilkins
Revision Date: 2013-11-14 02:18:31 UTC

Fix EnvironmentSet API: allow invalid env initially

EnvironmentSet does not work if the existing config
is invalid, as the implementation does an environs.New
with the existing config to obtain the Provider to do
validation. This is always going to fail for a freshly
bootstrapped environment.

This is fixed by only attempting to construct an
Environ with the new config, which is required to be
valid.

Author: Andrew Wilkins
Revision Date: 2013-11-13 07:20:54 UTC

Remove mongodb and data-dir on machine-agent termination

This CL is about cleaning up manual environment state
servers. When a manual environment state server is torn
down, it should remove its juju-db service as well as
its own (i.e. the machine-agent upstart service), and
finally remove the data-dir.

This is implemented by adding an UNINSTALL_SCRIPT entry
into the machine-agent's agent.conf. The machine agent
will extract this when it terminates, copy it out to a
file, and execute it. environs/cloudinit takes care of
generating the script based on what gets installed on
the machine.

Author: Andrew Wilkins
Revision Date: 2013-11-07 04:09:36 UTC

Add tests for PushEnvironmentSecrets (and reenable code)

Also, fix a bug in state.State.SetEnvironmentConfig,
so deleted attributes are removed from Mongo. Also,
modify the dummy provider to not set defaults in config
when calling SecretAttrs.

Fixes lp:1248809

Author: Andrew Wilkins
Revision Date: 2013-11-01 03:09:33 UTC

provider/null: first half of Environ.Destroy()

This is CL 1 of 2: destroy-environment will work
by first destroying all units, and non-environment
manager machines; then, the environment-manager
machine(s) will be destroyed by sshing to them
and killing jujud. jujud will catch the signal,
and clean itself up.

The next CL will implement the manager machine
killing bit.

Author: Andrew Wilkins
Revision Date: 2013-10-18 03:39:49 UTC

environs: warn if .jenv/env.yaml are inconsistent

Changes in environments.yaml do not take effect
for an environment that has already been prepared.
In some cases, it might not be obvious to a user
that the environment is prepared, and that changes
will not take effect. e.g.:
 - User called sync-tools, then modified env.yaml,
   then bootstrapped.
 - User bootstrapped, bootstrap failed, then modified
   env.yaml and tried again. We could remove the
   .jenv file on failure, but we can't guarantee
   juju won't die before doing so.
 - User is bootstrapping the null provider; since
   destroy-environment is not implemented, the user
   must remove the .jenv file manually. It's easy to
   forget to do that.

We now log a warning if any values have been
changed in environments.yaml after preparing,
and provide some direction on what to do.

Also: don't call Prepare for config retrieved from
the config store (i.e. already prepared config).

Fixes #1239550

Author: Andrew Wilkins
Revision Date: 2013-10-09 09:25:18 UTC

Address review comments

- Remove unnecessary admin-secret/ca-private-key copy.
- Log/add bug number to TODO in cmd/jujud.

Author: Andrew Wilkins
Revision Date: 2013-10-07 09:09:57 UTC

provider/null: add comment about ssh lifetime

Author: Andrew Wilkins
Revision Date: 2013-08-15 04:01:53 UTC

worker/deployer: add flag to control syslog fwd

- Add flag to NewSimpleContext to enable syslog forwarding
- Add util function in api/agent, HasJobs(*Entity, params.MachineJob)
- Update jujud to make use of these

Author: Andrew Wilkins
Revision Date: 2013-08-14 07:50:11 UTC

Don't forward logs if deploying on state server

Previously, whenever a unit was deployed a configuration
file was created in /etc/rsyslog.d to forward logs to
the state servers. This introduced a feedback loop when
the the unit was deployed on a state server machine.

Now, check for the existence of /etc/rsyslog.d/25-juju.conf
before creating an additional configuration.

Fixes bug #1211147

Author: Andrew Wilkins
Revision Date: 2013-08-06 09:06:11 UTC

destroy-unit: respond to dying state in hook-failure

If the user invokes destroy-unit while a unit
is in a hook-error state, the uniter should
respond to the dying state as if it were in
steady state.

If the unit hasn't started yet (e.g. the install
hook failed), then the uniter should short-circuit
the termination handling.

Author: Andrew Wilkins
Revision Date: 2013-07-31 10:16:57 UTC

cmd/juju: confirm destroy-environment

Author: Andrew Wilkins
Revision Date: 2013-07-24 09:09:30 UTC

cmd: Enable logging to stderr by default

No change to levels: WARNING and up by default.

This change does lead to some redundant output,
particularly where errors occur. In those cases,
you'll typically see an ERROR log message and
finally some error message (often the same one).

Author: Andrew Wilkins
Revision Date: 2011-09-17 05:58:28 UTC

Update Sphinx docs with some better examples and specifications of address formats

Author: Andrew Wilkins
Revision Date: 2011-07-11 11:20:37 UTC

Merging fix for marshal.dump usage (lp:808117)

Author: Andrew Wilkins
Revision Date: 2011-03-05 07:04:21 UTC

Add smb transport package to setup.py

Author: Andrew Wilkins
Revision Date: 2010-06-25 02:56:26 UTC

Fixing Java unittests (workaround for Python 2.6 defect to detect Windows 7)

Author: Andrew Wilkins
Revision Date: 2009-05-02 12:10:44 UTC

Merging from trunk, to freeze 0.2 changes