testing: use a valid fake authorized key

testing: use a valid fake authorized key This stops the jujud tests thrashing endlessly because the authorized key worker cannot parse the keys. I fixed this before, but the parsing code was changed to be stricter, so I've added a check that the key does indeed parse correctly. I've also increased the worker restart time somewhat - on some tests we were seeing the workers restarted 200 times before eventual success. https://code.launchpad.net/~rogpeppe/juju-core/505-testing-valid-fake-key/+merge/206779 (do not edit description out of merge proposal)

[gz, 2014-02-17 20:03:24]

LGTM. https://codereview.appspot.com/65160043/diff/1/testing/environ.go File testing/environ.go (right): https://codereview.appspot.com/65160043/diff/1/testing/environ.go#newcode22 testing/environ.go:22: const FakeAuthKeys = `ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAAAYQDP8fPSAMFm2PQGoVUks/FENVUMww1QTK6m++Y2qX9NGHm43kwEzxfoWR77wo6fhBhgFHsQ6ogE/cYLx77hOvjTchMEP74EVxSce0qtDjI7SwYbOpAButRId3g/Ef4STz8= jo@bloggs.com` This was freshly generated by you I take it? While at it, I'd change the bloggs.com to example.com or testing.invalid too.

[jameinel, 2014-02-18 08:29:40]

https://codereview.appspot.com/65160043/diff/1/testing/environ.go File testing/environ.go (right): https://codereview.appspot.com/65160043/diff/1/testing/environ.go#newcode22 testing/environ.go:22: const FakeAuthKeys = `ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAAAYQDP8fPSAMFm2PQGoVUks/FENVUMww1QTK6m++Y2qX9NGHm43kwEzxfoWR77wo6fhBhgFHsQ6ogE/cYLx77hOvjTchMEP74EVxSce0qtDjI7SwYbOpAButRId3g/Ef4STz8= jo@bloggs.com` On 2014/02/17 20:03:24, gz wrote: > This was freshly generated by you I take it? While at it, I'd change the > http://bloggs.com to http://example.com or testing.invalid too. I believe Roger Peppe said that on his setup he ends up having 'testing.invalid' actually route to an IP address. *Why* I can't say for sure. As such, we've taken to use 0.* addresses. They fail with a different error code, but are guaranteed not to be routable. (something like INVALID rather than NOTFOUND.) I guess for the contents of an SSH key it probably doesn't matter.

[rog, 2014-02-18 17:16:34]

Please take a look. https://codereview.appspot.com/65160043/diff/1/testing/environ.go File testing/environ.go (right): https://codereview.appspot.com/65160043/diff/1/testing/environ.go#newcode22 testing/environ.go:22: const FakeAuthKeys = `ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAAAYQDP8fPSAMFm2PQGoVUks/FENVUMww1QTK6m++Y2qX9NGHm43kwEzxfoWR77wo6fhBhgFHsQ6ogE/cYLx77hOvjTchMEP74EVxSce0qtDjI7SwYbOpAButRId3g/Ef4STz8= jo@bloggs.com` On 2014/02/18 08:29:40, jameinel wrote: > On 2014/02/17 20:03:24, gz wrote: > > This was freshly generated by you I take it? While at it, I'd change the > > http://bloggs.com to http://example.com or testing.invalid too. > > I believe Roger Peppe said that on his setup he ends up having 'testing.invalid' > actually route to an IP address. *Why* I can't say for sure. > > As such, we've taken to use 0.* addresses. They fail with a different error > code, but are guaranteed not to be routable. (something like INVALID rather than > NOTFOUND.) > > I guess for the contents of an SSH key it probably doesn't matter. It was freshly generated by me. Changed to joe@0.1.2.3 although I strongly suspect that nothing will ever attempt to resolve the MX address here.