I'm not sure what to do here yet.

It is a timeout, perhaps related to 
  [ERROR:devtools_http_handler.cc(226)] Cannot start http server for devtools. Stop devtools.

Adding Troopers for help parsing the swarming logs to narrow this down.

That error message is misleading, it's tracked in  bug 721714  (which I filed for the same reason).

Picked one green one
https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.android%2Flinux_android_rel_ng%2F306917%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests__with_patch__on_Android%2F0%2Fstdout

and compared with one red one
https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.android%2Flinux_android_rel_ng%2F306866%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests__with_patch__on_Android%2F0%2Fstdout

Looks the test stopped at FrameTreeBrowserTest.FrameTreeAfterCrash.

Another one
https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.android%2Flinux_android_rel_ng%2F306893%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests__with_patch__on_Android%2F0%2Fstdout

This also stopped at FrameTreeBrowserTest.FrameTreeAfterCrash.

https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.android%2Flinux_android_rel_ng%2F306825%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests__with_patch__on_Android%2F0%2Fstdout

This one ended with PASSED.

So, not sure about the culprit, but given 2 occurrences, I'd like to get this taken a look at by the owner of the test.

I wonder if NavigateToURL(shell(), GURL(kChromeUICrashURL)) is flaky on Android?  That might explain issue 585327 as well.

I can try RPH::Shutdown to see if that's any better.

Detected 6 new flakes for test/step "content_browsertests (with patch)". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyLAsSBUZsYWtlIiFjb250ZW50X2Jyb3dzZXJ0ZXN0cyAod2l0aCBwYXRjaCkM. This message was posted automatically by the chromium-try-flakes app.

lukasza@ brings up a good point on the review of https://codereview.chromium.org/2919483003/, which is that FrameTreeAfterCrash is not to blame in many of the failed runs.  It's probably a red herring, in which case we'll need someone else to take a look.

On 2017/06/01 16:10:29, Łukasz A. wrote:
> lgtm, assumming the questions/nits below are answered
> 
> I tried looking at the bug and
> 1) I see that FrameTreeAfterCrash appears in the 2 failed runs pointed out in
> #c3 and #c4, but I failed to see any details in the logs (they don't even say if
> the test timed out).
> 2) When looking at other logs I didn't see FrameTreeAfterCrash.
> 
> Given above, I am not sure how we will be able to evaluate if this speculative
> CL helped or not.

That said, I still think it's worth landing https://codereview.chromium.org/2919483003/ to see if it has any effect (given the overlap with issue 585327).  It's in the CQ now.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/298b9328adbfe81b92c2dadce496dac1e8ddfeb5

commit 298b9328adbfe81b92c2dadce496dac1e8ddfeb5
Author: creis <creis@chromium.org>
Date: Thu Jun 01 18:31:52 2017

Use RenderProcessHost::Shutdown in test to avoid flakiness.

Speculative fix for timeouts on linux_android_rel_ng in the
FrameTreeBrowserTest.FrameTreeAfterCrash test.

BUG= 727725 
TEST=Less flakiness
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2919483003
Cr-Commit-Position: refs/heads/master@{#476362}

[modify] https://crrev.com/298b9328adbfe81b92c2dadce496dac1e8ddfeb5/content/browser/frame_host/frame_tree_browsertest.cc

There are 6 new flakes today after my CL landed, and they're not related to FrameTreeAfterCrash (as lukasza@ expected).  This seems to be a larger issue in content_browsertests, which someone else will need to look into.

Can the current sheriff help find an owner?

The runs now seem to be failing with "Shard 0: not enough capacity" rather than "Shard 0: timed out". The switch happened on this run AFAICT: https://build.chromium.org/p/tryserver.chromium.android/builders/linux_android_rel_ng/builds/309475

However, the CL that triggered that run (https://codereview.chromium.org/2898573002) seem totally unrelated AFAICT.

Assigning to jbudorick@ as someone who might have insight into what's going on here.

 - tasks on linux_android_rel_ng expiring with not enough capacity is being addressed in https://bugs.chromium.org/p/chromium/issues/detail?id=725672. Most of the current flakes appear to be of this variety, so I'm duping this bug into that one.
 - chromium-try-flakes not reporting infra failures to sheriffs is https://bugs.chromium.org/p/chromium/issues/detail?id=726568
 - content_browsertests 90th percentil execution time has crept toward its current hard timeout of 960s over the last month on linux_android_rel_ng. Filed as https://bugs.chromium.org/p/chromium/issues/detail?id=729555