Merge lp://staging/~sinzui/juju-ci-tools/bind-single-space into lp://staging/juju-ci-tools
| Status: | Merged |
|---|---|
| Merged at revision: | 1930 |
| Proposed branch: | lp://staging/~sinzui/juju-ci-tools/bind-single-space |
| Merge into: | lp://staging/juju-ci-tools |
| Diff against target: |
174 lines (+70/-3) 5 files modified
assess_endpoint_bindings.py (+15/-0) jujupy/client.py (+3/-1) jujupy/tests/test_client.py (+6/-0) tests/test_assess_endpoint_bindings.py (+45/-1) tests/test_deploy_stack.py (+1/-1) |
| To merge this branch: | bzr merge lp://staging/~sinzui/juju-ci-tools/bind-single-space |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Christopher Lee (community) | Needs Fixing | ||
|
Review via email:
|
|||
Description of the change
Test global charm bindings.
Bug 1671489 shows that this command fails because the juju is confused by the instruction to bind a space to *all* bindings:
juju deploy mycharm --bind aspace
Juju is wrongly insisting that the user name each binding the charm has. This branch extends the bindings test to deploy the datastore charm to a single space without specifying the exact charm bindings. The test reads the bundle, then sets up the spaces and machines in the maas. We must use the bundle to test this case.
I made secondary changes to help me test
1. The bundle is now saved to artifacts so that we can see what was deployed
2. I extends ModelClient and the fake bootstrap manager to allow me to pass --bind to deploy.
I don't actually use --bind because of how the test is setup :(
IN a future branch, I will explicitly test the command line.
| Curtis Hovey (sinzui) wrote | # |
PS I hacked
http://
to use this branch. develop and 2.1 pass with the fix (and fail without the fix.) When this branch is merged, I will restore the job to use juju-ci-tools.