It's also not clear to me what is causing this, the system reboots orderly, not due to a kernel crash.

From a Moma search on 'cheets_StartAndroid "DUT rebooted during the test run"' it seems the issue has been around for > 6 months.

e.g. this is an instance on auron_paine from 02/22/18:

https://stainless.corp.google.com/browse/chromeos-autotest-results/178827513-chromeos-test/

I'm bit confused about the failure on veyron_mighty:

Supposedly the test ran on chromeos4-row6-rack11-host21. I ssh-ed into that device hoping to find some additional info in the local logs, however /var/log/messages doesn't match with the info on stainless (https://stainless.corp.google.com/browse/chromeos-autotest-results/216901379-chromeos-test/). It also reports an uptime > 13h, while the failure was reported about 3h ago (10:52:27):

uptime
 13:48:51 up  3:11,  0 users,  load average: 1.02, 0.86, 0.86

Am I just embarrassing myself looking at the wrong device or is there something weird here?

Deleted: logs.tar.gz 270 KB

logs.tar.gz 270 KB Download

Deleted: dmesg.log 57.3 KB

dmesg.log 57.3 KB View Download

I'm at least in part embarrassing myself, the uptime actually says 3:11 hours ... Will do another attempt to match the logs.

The first part of /var/log/messages actually matches, I got confused trying to match the kernel messages in /sys/fs/pstore/console-ramoops with /var/log/messages, which indeed don't match. Looks like a new log file was started after the mysterious reboot.

Copying current sheriffs

Kicking to current arc constable, though at this point this might be obsolete / wontfixable. This is unlikely to be an infra issue.