What makes you say that everything seems to have completed?

There are no apparent failures or tests that timed out. The shard seems to have a completion timestamp ahead of its deadline.

Created	26/09/2018 à 16:51:09 (heure d’été de l’Est)
Started	26/09/2018 à 16:51:10 (heure d’été de l’Est)
Expires	26/09/2018 à 17:51:09 (heure d’été de l’Est)
Completed	26/09/2018 à 17:13:25 (heure d’été de l’Est)
Abandoned	‑‑

I think the timeouts are local to the tests and not swarming-based.

The following appears at the end of the stdout for the first failing step:

 [----------] 4 tests from TimerSequenceTest
 [ RUN ] TimerSequenceTest.OneShotTimerTaskOnPoolSequence
 [ OK ] TimerSequenceTest.OneShotTimerTaskOnPoolSequence (5 ms)
 WARNING:root:collect_cmd had non-zero return code: 255
 WARNING:root:No shard json files found in task_output_dir: '/b/s/w/ir/tmp/rt/0uN6HAy'
 Found ['/b/s/w/ir/tmp/rt/0uN6HAy/summary.json']
 [ RUN ] TimerSequenceTest.OneShotTimerUsedOnPoolSequence
 +------------------------------------------------------------------------+
 | End of shard 0 |
 | Pending: 1.4s Duration: 1387.1s Bot: vm874-m4 Exit: -1 TIMED_OUT |
 +------------------------------------------------------------------------+
 Total duration: 1387.1s
 Running ['/b/s/w/ir/cache/vpython/cbb391/bin/python', '/b/s/w/ir/kitchen-checkout/build/scripts/slave/recipe_modules/swarming/resources/noop_merge.py', '--summary-json', '/b/s/w/ir/tmp/rt/0uN6HAy/summary.json', '--task-output-dir', '/b/s/w/ir/tmp/rt/0uN6HAy', '-o', '/b/s/w/ir/tmp/t/tmpctSJ8P.json'] in None (env: None)
 Command ['/b/s/w/ir/cache/vpython/cbb391/bin/python', '/b/s/w/ir/kitchen-checkout/build/scripts/slave/recipe_modules/swarming/resources/noop_merge.py', '--summary-json', '/b/s/w/ir/tmp/rt/0uN6HAy/summary.json', '--task-output-dir', '/b/s/w/ir/tmp/rt/0uN6HAy', '-o', '/b/s/w/ir/tmp/t/tmpctSJ8P.json'] returned exit code 0
 step returned non-zero exit code: 255

There are not log lines for each of the 4 indicated tests from TimerSequenceTest and there's a clear time out message.

We suspect those are I/O timeouts, which means that no output was produced by the test for over 20 minutes. Is it possible that your change would cause these tests to hang or take a long time on ios?

CL is on hold and looks like flakes were its fault (still not sure how I was supposed to find the logs though...)