The only thing I read in the logs is: Exception from TearDownState:
... 
AdbCommandFailedError: (device: 0d88c7fd25995e62) 
    adb shell '( p=org.chromium.chrome;if [[ "$(ps)" = *$p* ]]; then am force-stop $p; fi );echo %$?': failed with exit status 255 and output:
- error: device '0d88c7fd25995e62' not found

Locals:
  args          : ['shell', '( dumpsys battery );echo %$?']
  check_error   : False
  cls           : <class 'devil.android.sdk.adb_wrapper.AdbWrapper'>
  cpu_affinity  : None
  device_serial : '0d88c7fd25995e62'
  output        : "error: device '0d88c7fd25995e62' not found\n"
  retries       : 2
  status        : 255
  timeout       : 30


It's hard for me to assess what's going on. To me this rather seems an infrastructure change.
My CL removed a dangerous path that could lead to stale pointers, the only side-effect besides
correctness might be a slight slow-down due to the additionally GC work.

This looks like an infrastructure failure; you can see the device fail in the middle of https://uberchromegw.corp.google.com/i/chromium.perf/builders/Android%20Nexus5%20Perf%20%281%29/builds/3757, and then completely fail in https://uberchromegw.corp.google.com/i/chromium.perf/builders/Android%20Nexus5%20Perf%20%281%29/builds/3758

How important is this builder?

cc jbudorick: This would be fixed by the new perf runner, correct?

This looks like a dead or hung device.

The bots on chromium.perf are important -- probably somewhere between chromium.android and the main waterfall in importance.

No, this would not be fixed by the new perf runner, since the recovery script doesn't pick it up in subsequent runs.

This bot will get rebooted within a week, roughly; we have automatic ticket filing set up which covers this bot as well. Let me know if you want it back up sooner and I can get it back up.

Didn't see #3, I'll work on fixing this sooner.

Yeah, this needs to be fixed today. I suspect it just needs a manual reboot.

I wonder if this type of device failure would be fixed by a watchdog on the phone. I think I'll swing by the lab sometime in the next hour with a serial cable and take a look.

Looks like the device already got fixed before I had the chance to look at it:
https://uberchromegw.corp.google.com/i/chromium.perf/builders/Android%20Nexus5%20Perf%20%281%29/builds/3776

Not sure who or what got to it, but I think we can close here

Guessing it was the Monday devices ticket.