1301613 - submit a fennec nightly build to a staging Balrog instance via Balrog worker

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Mihai Tabara [:mtabara]⌚️GMT]

This is more or less like a proof-of-concept for me in order to set up and play with Balrogworker and get familiarized with how the flow works.

This work is a pre-requisite for me before picking up bug 1289822 to deploy balrogworker along signingscript in puppet.

Comment 1

[Mihai Tabara [:mtabara]⌚️GMT]

Forgot to add some missing steps here. Got stuck in the Docker-playground as I was using the old boot2docker way. Ended up wiping off Docker entirely from my machine and did a fresh install via 'Docker for Mac' which went on nicer. Used the https://wiki.mozilla.org/Balrog#Hacking as well to install deps before actually running the 'docker-compose up' command.

Comment 2

[Mihai Tabara [:mtabara]⌚️GMT]

Having both Balrog and Balrogworker working on my local machine. Investigating further to actually send valid payload to Balrog via balrogworker.

Comment 3

[Mihai Tabara [:mtabara]⌚️GMT]

Note to self: so I've been trying to claim a dummy task in TC from my local Balrogworker instance today and hit several walls due to my errors. So I'll just enum some of those bits here for future reference:

1. In order to setup the env for Balrog local instance and Balrogworker I've used a slightly modified version of Anthony's work, a forked gist https://gist.github.com/MihaiTabara/d143a2823bde212971585cb7d3fb4d7a

2. Never forget to refresh the docker image every time config.json is touched!

3. Having a worker group different than the default one in https://github.com/mozilla-releng/scriptworker/blob/1d9cbcf60be1cc535611926233e4b01fd77dc724/scriptworker/config.py#L24 may lead to missing scopes. Can be solved by enriching the client with more scopes but also use the existing one since we've got it already:

2016-09-19T16:47:48    DEBUG - Got You do not have sufficient scopes. This request requires you
to have one of the following sets of scopes:
[
  [
    "queue:claim-task",
    "assume:worker-type:test-dummy-provisioner/dummy-worker-mtabara",
    "assume:worker-id:dummy-worker-balrog/dummy-worker-mtabara1"
  ],
  [
    "queue:claim-task:test-dummy-provisioner/dummy-worker-mtabara",
    "queue:worker-id:dummy-worker-balrog/dummy-worker-mtabara1"
  ]
]
 
You only have the scopes:
[
  "assume:project:taskcluster:worker-test-scopes",
  "queue:claim-task:test-dummy-provisioner/dummy-worker-*",
  "queue:define-task:test-dummy-provisioner/dummy-worker-*",
  "queue:poll-task-urls:test-dummy-provisioner/dummy-worker-*",
  "queue:schedule-task:test-dummy-scheduler/*",
  "queue:task-group-id:test-dummy-scheduler/*",
  "queue:worker-id:test-dummy-workers/dummy-worker-*"
]

I was using worker_group: "dummy-worker-balrog" while scopes for "test-dummy-workers" already existed.

Eventually, managed to claim a dummy task with my local balrogworker instance -  https://tools.taskcluster.net/task-inspector/#CbqTL76XTgqxhsZAOnFD1g/0 Got the expected malformed schema validation as I didn't provide the right payload.

Moving forward to send proper data in order to be processed and sent to local Balrog instance.

Comment 4

[Mihai Tabara [:mtabara]⌚️GMT]

We can close this now - we've moved everything to staging, no more local development needed.