1275272 - Autophone - nexus-6p-8 usb disconnected 2016-05-24

Status: RESOLVED FIXED

(Testing Graveyard :: Autophone, defect)

Description

[Bob Clary [:bc] (inactive)]

nexus-6p-8 disconnected from usb on host autophone-3 about 4:21 AM 2016-05-24 and did not reconnect after a reboot. The previous disconnection for nexus-6p-8 occurred yesterday.

Reviewing the logs it appears to disconnect when Rwv completes on a non mozilla-inbound repo, then runs several rca tests in a row on mozilla-inbound.

As a first step, I'm going to disable rca on nexus-6p-8 on all branches to see if this affects stability. I'll enable it on try to allow us to try to disconnect it on demand using rca. I'll also attempt to reproduce using multiple rca runs on my local 6p.

van, stephend: Can you disconnect/reconnect nexus-6p-8 again?

Comment 1

[Bob Clary [:bc] (inactive)]

Attachment: bug-1275272-v1.patch

Comment 2

[Bob Clary [:bc] (inactive)]

https://github.com/mozilla/autophone/commit/776bb898f5b0c84fec6cea03b5cf8d69028c60a6

deployed 07:30 AM PDT 2016-05-24

also updated linux.

stephend is on pto, so van you're it!

Comment 3

[Van Le [:van]]

cable reseated. forgot to mention i checked and changed the silent settings in yesterday's bug for this phone. i also put back the longer usb > usb c cable to confirm that the applied fix resolves the issue.

fwiw, monoprice is generally very good but my last build, i had 10 bad cat5 cables from a batch of 200.

Comment 4

[Bob Clary [:bc] (inactive)]

fyi, I ran a test locally with nexus-6p-6 with rca. I did 21 mozilla-inbound builds from sat-sun without problems. This was somewhat different from scenario in production since there was a delay in between builds while the new build downloaded. I re-ran the exact same 21 builds from the build cache which loaded them from the local disk rather than trying to download them. Again I saw no problems. The difference between my environment and production is that I'm on Fedora 23 while the production servers are on Ubuntu 14.04.4 LTS.