750987 - Deploy server-aitc 1.1-1

Status: VERIFIED FIXED

(Cloud Services :: Operations: Deployment Requests - DEPRECATED, task)

Description

[Ryan Kelly [:rfkelly]]

This release allows the AITC client to store arbitrary extra fields as part of the JSON app record.  This is necessary because the set of required/allowed fields is still in flux during client development.


Build Commands:

make build CHANNEL=prod RPM_CHANNEL=prod PYPI=http://pypi.build.mtv1.svc.mozilla.com/simple PYPIEXTRAS=http://pypi.build.mtv1.svc.mozilla.com/extras PYPISTRICT=1

make build_rpms CHANNEL=prod RPM_CHANNEL=prod PYPI=http://pypi.build.mtv1.svc.mozilla.com/simple PYPIEXTRAS=http://pypi.build.mtv1.svc.mozilla.com/extras PYPISTRICT=1

(Passing specific tag names like SERVER_AITC=1.1-1 seems to be busted for github projects at the moment, but using CHANNEL=prod will pull in the latest release tags)

Bug fixes:

Bug 750566 - Don't reject unknown app record fields


QA Plan:

We don't have a client-driven test plan for AITC since we don't have a client yet, so I guess the most we can do is verify that the functional tests all pass.

Comment 1

[Ryan Kelly [:rfkelly]]

Objections have been raised to Bug 750566, please hold off until they're sorted.

Comment 2

[Ryan Kelly [:rfkelly]]

Oh, I see telliott gave this the green light in IRC.  I'm also happy with that, please ignore above comment.

Comment 3

[:Atoll]

aitc -> 1.1-1

Gunicorn runs, so presumably it's working. Needs someone to test the API.

Comment 4

[Ryan Kelly [:rfkelly]]

I don't appear to have sudo on the stage webheads, so I couldn't grab the secrets file to run the functional testsuite with the usual procedure of synthesizing auth tokens.  Instead I hacked up my tests to use stage-token, and all tests passed.

Comment 5

[James Bonacci [:jbonacci]]

I can help you with this if you want to repeat your testing with the actual secrets file in place.

Comment 6

[James Bonacci [:jbonacci]]

Comments from Bug 751485 indicate that this method of copying over a secrets file may no longer be needed...
:rfkelly is your plan to keep both methods available, or focus on the method spelled out in Bug 751485?

Comment 7

[Ryan Kelly [:rfkelly]]

I feel like the Bug 751485 method should be sufficient and is easier to run on stage/production.  Both methods will remain available but it shouldn't be necessary to run them both since they ultimately test the same thing.  Interested in your thoughts as well.

Comment 8

[:Atoll]

Production deploy on hold while QA is in progress.

ChangeWindow has been prepared, r? https://services.etherpad.mozilla.org/25 :rfkelly :jbonacci

Comment 9

[James Bonacci [:jbonacci]]

Yea, :rfkelly and I were working through some issues found during load test of Stage...

Comment 10

[Toby Elliott [:telliott]]

This should stay on stage and never hit production. It moves outside our spec, so we don't want to deploy it live, and the presence of it on stage should be sufficient for anant to lock down any client changes he's working on.

So this bug can be closed one it's working on stage.

Comment 11

[:Atoll]

Working on stage, so marking as RESOLVED FIXED. :jbonacci and :rfkelly testing continues.

Comment 12

[James Bonacci [:jbonacci]]

Talked it over with :rfkelly - we are pretty sure the issues we found during loadtest are not OPs related/deployment related.
Thanks.