Closed manics closed 4 years ago
Changes all make sense. Other than the molecule tests, is there a target playbook at the moment?
As for release, I assume this is part of the next milestone. 2.1.0? 3.0.0?
2.1.0
if Python 3 isn't the default, 3.0.0
if it is.
And which do you suggest?
2.1.0
now and advertise to the community they can test a Python 3 deployment by setting a var?
No targets yet. We don't have any servers we can test this single change on, they'll all need web-app versions to be bumped too.
I'm not sure I follow. You mean there are other PRs needed before we're ready? Can you spell out what those would be and a timeline?
I meant you can't just bump the role version in requirements, set the python3 variable in a prod-playbook, and expect it to work. You'll also need to update other versions in the playbook, and potentially change some dependencies for the apps.
Leave you to get this out, but we're still going to need an initial roll-out plan quite soon.
I think we should treat the 5.6 release like any other major release and not make special allowance for it involving a Python migration. We should be able to methodically work through a playbook, bumping versions as we normally would, ... and see what bugs there are in ansible or OMERO.
Presumably we want a test server for release upgrade testing? So, let's pick one that everyone on the team is prepared to use for testing.
That sounds great. I'll leave you to have a plan for the systems to test on and the remaining PRs that are needed. The only addition I'll add is that at least from the point-of-view of the dockers there were a number of breaking changes, so I would plan that into the roll-out even if we ultimately don't need the extra time.
My aim with these latest docker PRs is to get the production containers matching the VMs created by the ansible roles.
Uses https://github.com/ome/omero-py/pull/137
Rest of the commits rearrange the test files, and adds a py2 → py3 OMERO.web upgrade test.