1592377 - android-hw/bitbar tcw migration work

Status: RESOLVED FIXED

(Taskcluster :: Workers, task)

Description

[Andrew Erickson [:aerickson]]

Tracking bug for the migration described in https://bugzilla.mozilla.org/show_bug.cgi?id=1591152.

• planning doc: https://docs.google.com/document/d/1T4R9sOONZwQAby2ejLDcWhfOPF0Z4wLhK-nnKIvFSqA/edit?usp=sharing
• deployments overview: https://docs.google.com/document/d/1Tdn98nIl7_k89llPGIfIUl7umkCWUkTOSvj3eBiErcg/edit?usp=sharing

Work to do:

• modify android-tools scripts to use new rootUrl and path
• create a PR/build a mozilla-bitbar-docker image with new rootUrl and path
• create branch/PR for mozilla-bitbar-devicepool to use new rootUrl and path
• create new credentials for workers
• create credentials envvar file for mozilla-bitbar-devicepool
• get testing instance of mozilla-bitbar-devicepool running on backup host
• test new instance by moving some workers and sending jobs
• Sat, Nov 9: land PRs, change devicepool0 to use new envvars, restart, monitor

Comment 1

[Andrew Erickson [:aerickson]]

details on steps for devicepool

...

android-hw/bitbar devicepool TCW migration plan

Before:
step 1. devicepool 0/prod: land PR (https://github.com/bclary/mozilla-bitbar-devicepool/pull/69) to disable test queues, restart.
step 2. devicepool 2: place new env var file, run devicepool from branch/PR (https://github.com/bclary/mozilla-bitbar-devicepool/pull/68), config will only enable test queues
step 3: test new TC deployment

Nov 9:
step 4: devicepool 2: shutdown devicepool.
step 5: devicepool 0/prod: shutdown, land PR (https://github.com/bclary/mozilla-bitbar-devicepool/pull/67), change env var file, update client

Comment 2

[Andrew Erickson [:aerickson]]

Update on devicepool testing: Steps 1-2 completed. Step 3 partially done. The deployment can't accept try push tests yet, but I've been able to create simple tasks via https://firefox-ci-tc.services.mozilla.com/tasks/create that have worked.

payload failure, but reached the g-w: https://firefox-ci-tc.services.mozilla.com/tasks/Nf4hYCmNSumh37tOj5rhIw
successful 10 minute sleep: https://firefox-ci-tc.services.mozilla.com/tasks/OWlSC4cVSIaGJX0thdSFUg

task used for 10 minute sleep:

provisionerId: proj-autophone
workerType: gecko-t-bitbar-gw-test-1
created: '2019-11-01T16:11:39.923Z'
deadline: '2019-11-01T19:11:39.923Z'
payload:
  command:
    - ['/bin/bash', '-c', 'for ((i=1;i<=600;i++)); do echo $i; sleep 1; done']
  maxRunTime: 630
metadata:
  name: Example Task
  description: Markdown description of **what** this task does
  owner: [e-mail]
  source: 'https://firefox-ci-tc.services.mozilla.com/tasks/create'

Comment 3

[Andrew Erickson [:aerickson]]

Migration is complete.

Things went pretty much as expected, but I copied the wrong clients (with bad scopes) over from the legacy deployment. Was fixed pretty quickly.