1071995 - diagnostics: Test that we can create a task and that it gets resolved

Status: RESOLVED FIXED

(Taskcluster :: General, defect)

Description

[Jonas Finnemann Jensen (:jonasfj)]

In taskcluster-diagnostics, run-diagnostics should:

 1. createTask for provisionerId: aws-provisioner, workerType: v2
    - Using a standard docker image (like ubuntu:14.04)
    - Run command ["/bin/bash", "-c", "echo 'Hello World'"]
 2. Listen for task-completed message over AMQP
    - Bind to specific taskId that was created in step (1)
 3. Check that the task completed successfully
 4. Check that a log file named something like public/logs/... was uploaded as artifact
    - And that the log file contains "Hello World" :)
    - List artifacts and use queue.buildSignedUrl(queue.getArtifactFromRun, ...)
      to build a URL to fetch artifact from. Then fetch it using a HTTP client that
      supports redirects (superagent-promise is easy to use)
 
These tests will help us check that our production deployment works.
After this we can look at testing artifact extraction, other AMQP messages, task-graphs and many other things. But let's create new bugs blocking bug 1051561, to track these as we move forward.

Remark, the provisionerId and workerType in step (1) should probably be provided through configuration file. They aren't secret, but we might change them someday.

Note, examples in api tests for taskcluster-queue, as well as taskcluster-client documentation from here: https://github.com/taskcluster/taskcluster-client
And API documentation for taskcluster at: docs.taskcluster.net
Will be useful for this bug... (In addition to asking people on #taskcluster in IRC).

Comment 1

[Dustin J. Mitchell [:dustin] (he/him)]

I think the work Chinmay has done covers this?