1337775 - decision task fails on vendored servo code, I think due to the large volume of files

Status: RESOLVED DUPLICATE of bug 1303904

(Firefox Build System :: Task Configuration, task)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

looking at this:
https://treeherder.mozilla.org/#/jobs?repo=autoland&revision=be030db91f007f59dcbd4b51033caee8b2251a8f&selectedJob=75501908

we get:
https://public-artifacts.taskcluster.net/Hx__YbTARyi9-SQMWPl5kA/0/public/logs/live_backing.log

which shows this error:
[task 2017-02-08T13:59:26.346387Z] Querying version control for metadata: https://hg.mozilla.org/integration/autoland/json-automationrelevance/be030db91f007f59dcbd4b51033caee8b2251a8f
[task 2017-02-08T13:59:26.350294Z] attempt 1/2
[task 2017-02-08T13:59:26.350334Z] retry: calling get_automationrelevance, attempt #1
[task 2017-02-08T13:59:26.350366Z] Starting new HTTPS connection (1): hg.mozilla.org
[task 2017-02-08T13:59:31.690299Z] retry: Caught exception: 
[task 2017-02-08T13:59:31.690469Z] sleeping for 10.00s (attempt 1/2)
[task 2017-02-08T13:59:41.700824Z] attempt 2/2
[task 2017-02-08T13:59:41.700953Z] retry: calling get_automationrelevance, attempt #2
[task 2017-02-08T13:59:41.703171Z] Starting new HTTPS connection (1): hg.mozilla.org
[task 2017-02-08T13:59:46.945563Z] retry: Caught exception: 
[task 2017-02-08T13:59:46.945607Z] retry: Giving up on get_automationrelevance
[task 2017-02-08T13:59:46.945913Z] Traceback (most recent call last):
[task 2017-02-08T13:59:46.945945Z]   File "/home/worker/checkouts/gecko/taskcluster/mach_commands.py", line 165, in taskgraph_decision
[task 2017-02-08T13:59:46.946945Z]     return taskgraph.decision.taskgraph_decision(options)
[task 2017-02-08T13:59:46.946980Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/decision.py", line 102, in taskgraph_decision
[task 2017-02-08T13:59:46.947076Z]     write_artifact('task-graph.json', tgg.optimized_task_graph.to_json())
[task 2017-02-08T13:59:46.947111Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/generator.py", line 142, in optimized_task_graph
[task 2017-02-08T13:59:46.947853Z]     return self._run_until('optimized_task_graph')
[task 2017-02-08T13:59:46.947909Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/generator.py", line 251, in _run_until
[task 2017-02-08T13:59:46.947936Z]     k, v = self._run.next()
[task 2017-02-08T13:59:46.947979Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/generator.py", line 244, in _run
[task 2017-02-08T13:59:46.947999Z]     do_not_optimize)
[task 2017-02-08T13:59:46.948044Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/optimize.py", line 35, in optimize_task_graph
[task 2017-02-08T13:59:46.948067Z]     existing_tasks=existing_tasks)
[task 2017-02-08T13:59:46.948112Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/optimize.py", line 98, in annotate_task_graph
[task 2017-02-08T13:59:46.948141Z]     optimized, replacement_task_id = task.optimize(params)
[task 2017-02-08T13:59:46.948184Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/task/transform.py", line 91, in optimize
[task 2017-02-08T13:59:46.948213Z]     params, self.when['files-changed'])
[task 2017-02-08T13:59:46.948257Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/files_changed.py", line 58, in check
[task 2017-02-08T13:59:46.948289Z]     changed_files = get_changed_files(repository, revision)
[task 2017-02-08T13:59:46.948333Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/files_changed.py", line 33, in get_changed_files
[task 2017-02-08T13:59:46.948368Z]     contents = retry(get_automationrelevance, attempts=2, sleeptime=10)
[task 2017-02-08T13:59:46.948409Z]   File "/home/worker/checkouts/gecko/python/redo/redo/__init__.py", line 162, in retry
[task 2017-02-08T13:59:46.948434Z]     return action(*args, **kwargs)
[task 2017-02-08T13:59:46.948479Z]   File "/home/worker/checkouts/gecko/taskcluster/taskgraph/files_changed.py", line 31, in get_automationrelevance
[task 2017-02-08T13:59:46.948512Z]     response = requests.get(url, timeout=5)
[task 2017-02-08T13:59:46.948549Z]   File "/home/worker/checkouts/gecko/python/requests/requests/api.py", line 67, in get
[task 2017-02-08T13:59:46.948578Z]     return request('get', url, params=params, **kwargs)
[task 2017-02-08T13:59:46.948618Z]   File "/home/worker/checkouts/gecko/python/requests/requests/api.py", line 53, in request
[task 2017-02-08T13:59:46.948649Z]     return session.request(method=method, url=url, **kwargs)
[task 2017-02-08T13:59:46.948689Z]   File "/home/worker/checkouts/gecko/python/requests/requests/sessions.py", line 468, in request
[task 2017-02-08T13:59:46.948713Z]     resp = self.send(prep, **send_kwargs)
[task 2017-02-08T13:59:46.948751Z]   File "/home/worker/checkouts/gecko/python/requests/requests/sessions.py", line 576, in send
[task 2017-02-08T13:59:46.948775Z]     r = adapter.send(request, **kwargs)
[task 2017-02-08T13:59:46.948815Z]   File "/home/worker/checkouts/gecko/python/requests/requests/adapters.py", line 449, in send
[task 2017-02-08T13:59:46.948845Z]     raise ReadTimeout(e, request=request)
[task 2017-02-08T13:59:46.948892Z] ReadTimeout: HTTPSConnectionPool(host='hg.mozilla.org', port=443): Read timed out. (read timeout=5)

we shouldn't block execution of builds/tests based on automationrelevance.

Comment 1

[Gregory Szorc [:gps]]

Dupe.

This is an annoying bug. However, it isn't trivial to fix the performance bottleneck in the pushlog extension due to the way the APIs are being called.