Here's an excerpt from syslog:

2016-12-11T21:13:30.135787+01:00 WARNING session_manager[740]: [WARNING:liveness_checker_impl.cc(68)] Browser hang detected!
2016-12-11T21:13:30.135852+01:00 WARNING session_manager[740]: [WARNING:liveness_checker_impl.cc(72)] Aborting browser process.
2016-12-11T21:13:30.139484+01:00 INFO session_manager[740]: [INFO:browser_job.cc(124)] Terminating process: Browser did not respond to DBus liveness check.
2016-12-11T21:13:30.139549+01:00 INFO session_manager[740]: [INFO:system_utils_impl.cc(49)] Sending 6 to 4392 as 1000
2016-12-11T21:13:33.144285+01:00 WARNING session_manager[740]: [WARNING:browser_job.cc(132)] Aborting child process 4392's process group 3 seconds after sending signal
2016-12-11T21:13:33.144319+01:00 INFO session_manager[740]: [INFO:browser_job.cc(116)] Terminating process group: Browser took more than 3 seconds to exit after signal.
2016-12-11T21:13:33.144331+01:00 INFO session_manager[740]: [INFO:system_utils_impl.cc(49)] Sending 6 to -4392 as 1000
2016-12-11T21:13:34.842624+01:00 WARNING crash_reporter[929]: [user] Received crash notification for chrome[4404] sig 6, user 1000 (ignoring call by kernel - chrome crash; waiting for chrome to call us directly)
2016-12-11T21:13:38.544001+01:00 WARNING crash_reporter[935]: Received crash notification for chrome[4392] user 1000 (called directly)
2016-12-11T21:13:38.908966+01:00 WARNING crash_reporter[954]: [user] Received crash notification for chrome[4392] sig 6, user 1000 (ignoring call by kernel - chrome crash; waiting for chrome to call us directly)
2016-12-11T21:13:38.987525+01:00 WARNING crash_reporter[955]: Received crash notification for chrome-crash-unknown-process[4511] user 1000 (called directly)
2016-12-11T21:13:39.026632+01:00 WARNING crash_reporter[956]: Received crash notification for chrome-crash-unknown-process[15277] user 1000 (called directly)
2016-12-11T21:13:39.132178+01:00 WARNING crash_reporter[965]: [user] Received crash notification for chrome[4429] sig 6, user 1000 (ignoring call by kernel - chrome crash; waiting for chrome to call us directly)
2016-12-11T21:13:39.142406+01:00 INFO session_manager[740]: [INFO:child_exit_handler.cc(77)] Handling 4392 exit.
2016-12-11T21:13:39.142652+01:00 ERR session_manager[740]: [ERROR:child_exit_handler.cc(85)]   Exited with signal 6
2016-12-11T21:13:39.143121+01:00 INFO session_manager[740]: [INFO:session_manager_service.cc(293)] Exiting process is chrome.
2016-12-11T21:13:39.143295+01:00 INFO session_manager[740]: [INFO:browser_job.cc(116)] Terminating process group: Ensuring browser processes are gone.
2016-12-11T21:13:39.143440+01:00 INFO session_manager[740]: [INFO:system_utils_impl.cc(49)] Sending 9 to -4392 as 1000

So session_manager is killing Chrome after the former gets stuck and doesn't respond on DBus. No chrome logs around that point in time.

I'm going to tentatively assign this to abodenha@ in the hope that he's seen this before and may find the additional data point and logs useful. In case this isn't on anyone's radar yet, it should be since it happens every time the device sees usage.

Also, do we have UMA tracking for this type of event anywhere? We might be blind for this given that it doesn't generate useful crash reports (if any at all, I couldn't find listed on chrome://crashes on the device in question).

There's ChildProcess.HangRendererType but I don't think this sort of hang would trigger that.

I'd expect the crash reporter to pick up a termination like this, especially since crash_reporter is showing up in the log.  Is anything at all showing in chrome://crashes?

The odd thing was that when I checked after the crash, there was no crash report visible in chrome://crashes, and the problem had been occurring on and off for a couple of weeks already.

Re-checking now, there are crash reports listed from this week that might correlate with further occurrences of the issue:

http://crash.corp.google.com/bab6a35880000000
http://crash.corp.google.com/7f6202e300000000
http://crash.corp.google.com/3c63a35880000000
http://crash.corp.google.com/9ba30be300000000
http://crash.corp.google.com/714ee40080000000
http://crash.corp.google.com/5f6215d880000000
http://crash.corp.google.com/416be5d880000000
http://crash.corp.google.com/8b0303e300000000
http://crash.corp.google.com/597e49e300000000
http://crash.corp.google.com/c89249e300000000
http://crash.corp.google.com/6870faff00000000
http://crash.corp.google.com/cbcf89e300000000

Quite a few crash reports from just this week! The first and the last one look suspicious, and these seem to be what's already tracked as issue 661306. Feel free to dupe into that if you agree.

You're getting bit by the file descriptor issues that are plaguing us.  Disabling drive file sync seems to improve things for the victims of this one. We're working on it.