Fix/remove skipped tests

Fix/remove skipped tests

This change addresses four tests
that were previously skipped due
to being unreliable or untestable
on the bot.

The ec2 test is no longer relevant,
as the test it was masking is no
longer in the source tree.

The keymanager test should be an
integration test (it requires
ssh-import-id, thus network access)
so I removed it.

The cmd/juju tests were calling
"juju bootstrap"; I've changed them
to call "juju env" instead, so there
will be no risk of side-effects or
network access.

https://code.launchpad.net/~axwalk/juju-core/unskip-tests/+merge/220899

(do not edit description out of merge proposal)

[axw, 2014-05-26 05:53:25]

Please take a look.

[fwereade, 2014-05-26 07:04:26]

LGTM if you do ssh-import-id tests, either in this branch or an imminent
followup.

https://codereview.appspot.com/100710044/diff/1/cmd/juju/main_test.go
File cmd/juju/main_test.go (left):

https://codereview.appspot.com/100710044/diff/1/cmd/juju/main_test.go#oldcode177
cmd/juju/main_test.go:177: func breakJuju(c *gc.C, environMethod string) (msg
string) {
w000000t! I'd forgotten this existed, but I am *so* happy to see it go.

https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
File state/apiserver/keymanager/keymanager_test.go (left):

https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
state/apiserver/keymanager/keymanager_test.go:284: output, err :=
keymanager.RunSSHImportId("lp:wallyworld")
I'd prefer to see this unit-tested against a mock ssh-import-id, that produces
both good and bad output, rather than just dropping the test. Look at the
upstart tests, search for similar ones, and see if it's at the point where we
can sanely pull out some common code -- I'm sure we've done it 3 or 4 times now.

[axw, 2014-05-26 07:53:27]

Please take a look.

https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
File state/apiserver/keymanager/keymanager_test.go (left):

https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
state/apiserver/keymanager/keymanager_test.go:284: output, err :=
keymanager.RunSSHImportId("lp:wallyworld")
On 2014/05/26 07:04:25, fwereade wrote:
> I'd prefer to see this unit-tested against a mock ssh-import-id, that produces
> both good and bad output, rather than just dropping the test. Look at the
> upstart tests, search for similar ones, and see if it's at the point where we
> can sanely pull out some common code -- I'm sure we've done it 3 or 4 times
now.

Oops, I stupidly assumed it was unit-tested elsewhere. Done.

[axw, 2014-05-26 07:56:02]

On 2014/05/26 07:53:27, axw wrote:
> Please take a look.
> 
>
https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
> File state/apiserver/keymanager/keymanager_test.go (left):
> 
>
https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
> state/apiserver/keymanager/keymanager_test.go:284: output, err :=
> keymanager.RunSSHImportId("lp:wallyworld")
> On 2014/05/26 07:04:25, fwereade wrote:
> > I'd prefer to see this unit-tested against a mock ssh-import-id, that
produces
> > both good and bad output, rather than just dropping the test. Look at the
> > upstart tests, search for similar ones, and see if it's at the point where
we
> > can sanely pull out some common code -- I'm sure we've done it 3 or 4 times
> now.
> 
> Oops, I stupidly assumed it was unit-tested elsewhere. Done.

I added the existing tests. I'm going to add some more tests now, before
landing.

[axw, 2014-05-26 08:06:10]

On 2014/05/26 07:56:02, axw wrote:
> On 2014/05/26 07:53:27, axw wrote:
> > Please take a look.
> > 
> >
>
https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
> > File state/apiserver/keymanager/keymanager_test.go (left):
> > 
> >
>
https://codereview.appspot.com/100710044/diff/1/state/apiserver/keymanager/ke...
> > state/apiserver/keymanager/keymanager_test.go:284: output, err :=
> > keymanager.RunSSHImportId("lp:wallyworld")
> > On 2014/05/26 07:04:25, fwereade wrote:
> > > I'd prefer to see this unit-tested against a mock ssh-import-id, that
> produces
> > > both good and bad output, rather than just dropping the test. Look at the
> > > upstart tests, search for similar ones, and see if it's at the point where
> we
> > > can sanely pull out some common code -- I'm sure we've done it 3 or 4
times
> > now.
> > 
> > Oops, I stupidly assumed it was unit-tested elsewhere. Done.
> 
> I added the existing tests. I'm going to add some more tests now, before
> landing.

On closer inspection, TestImportKeys already tests runSSHKeyImport with a faked
ssh-import-id (with good and bad output). The test I removed was redundant, so
I'm taking it back out.

[axw, 2014-05-26 08:28:20]

Please take a look.