Looking...

https://ci.chromium.org/buildbot/tryserver.chromium.linux/fuchsia_x64/38161 seems to be an example where it was a fair chunk less than 20 minutes. LUCI says "18 mins 55 secs". The VM's last log is "1084.017" == 18.07s.

...18.07 minutes?

Aha, looking at https://logs.chromium.org/v/?s=chromium%2Fbb%2Ftryserver.chromium.linux%2Ffuchsia_x64%2F38161%2F%2B%2Frecipes%2Fsteps%2Fcontent_unittests__with_patch_%2F0%2Fstdout

there's some sneaky output:

[01084.017] 03759.03785> Tests took 1071 seconds.
+--------------------------------------------------------------------------------------+
| End of shard 0                                                                       |
|  Pending: 0.6s  Duration: 1112.0s  Bot: gce-trusty-dd4ff4d6-us-west1-c-brqz  Exit: 0 |
+--------------------------------------------------------------------------------------+
Total duration: 1112.0s
Missing or invalid gtest JSON file: /tmp/tmpiZI6Im/0/output.json
ValueError: Unterminated string starting at: line 51722 column 7 (char 4489213)
Task ran but no result was found: shard 0 test output was missing or invalid
some shards did not complete: 0


I downloaded the swarming output:
$ python swarming.py collect -S chromium-swarm.appspot.com --task-output-dir=foo 3a854cf8377a7710

and indeed the file is truncated:

[scottmg:~/work/cr/src/tools/swarming_client ((4bd9152...))]$ tail foo/0/output.json
      },
      "LocalStorageContextMojoTest.InvalidVersion": {
         "file": "../../content/browser/dom_storage/local_storage_context_mojo_unittest.cc",
         "line": 368
      },
      "LocalStorageContextMojoTest.MetaDataClearedOnDelete": {
         "file": "../../content/browser/dom_storage/local_storage_context_mojo_unittest.cc",
         "line": 440
      },
      "Lo[scottmg:~/work/cr/src/tools/swarming_client ((4bd9152...))]$

I think this is a Fuchsia/Chromium problem, not an Infra issue.

(sorry, yeah, 18.07 minutes)

I CQ'd a fix for what I believe is the problem. I'll check on success rate for this bot in the morning PST.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c552dad1797351aa6b54998e2a8fca21d0c07758

commit c552dad1797351aa6b54998e2a8fca21d0c07758
Author: Scott Graham <scottmg@chromium.org>
Date: Tue Dec 19 07:47:06 2017

fuchsia: Move runner shutdown to after flush delay

The delay before VM termination was accidentally moved in
https://chromium-review.googlesource.com/c/chromium/src/+/734128 and so
on fast shutdown the results .json file might not be flushed to the disk
image. This could result in a truncated output.json (see linked bug)
causing bot failures.

Also, unify delay to longer of current values as they're arbitrary
anyway.

TBR=wez@chromium.org

Bug:  796026 
Change-Id: Ie8986ec928e37d251571590c9f05b58109df0f1f
Reviewed-on: https://chromium-review.googlesource.com/833757
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#524966}
[modify] https://crrev.com/c552dad1797351aa6b54998e2a8fca21d0c07758/build/fuchsia/runner_common.py

I believe this is resolved, looking at https://ci.chromium.org/buildbot/tryserver.chromium.linux/fuchsia_x64/?limit=200.

Thanks Scott for fixing this so quickly.