New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 662086 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Apr 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 1
Type: Bug

Blocking:
issue 663373



Sign in to add a comment

Device level targeting for triggered jobs in swarming

Project Member Reported by eyaich@chromium.org, Nov 3 2016

Issue description

For swarming the perf waterfall, we need to be able to consistently run a test on the same hardware for our results to be usable.  For desktop, we target the id of the bot and consistently use that id every time the test is triggered.

For android, we typically have 7 devices hooked up to one host, so need to be able to consistently run a test on the same device.  

One proposed solution is to implement a new dimension for device serial id but this rather brittle.  Ideally we want to be able to implement this based on the device connection (1-7) and not the device that is connected.

This is a bug to track the decision for this and the implementation as it is needed to swarm the chrome perf waterfall.
 
If you are going to run a test on a single device out of the 7 connected devices, why not hav a 1:1 mapping for perf? Since you don't need fleet scalability, it seems like the most sensible solution.
Blocking: 663373

Comment 3 by eyaich@chromium.org, Nov 15 2016

Can you elaborate on what you mean by 1:1 mapping for perf?  Is that different from swarming adding a dimension for device serial id? 

We are blocked on this implementation now and I want to determine what the next steps are in implementation.

Comment 4 by mar...@chromium.org, Nov 15 2016

No, it's reconstructing the fleet the same Skia did. I'll send internal links.
Cc: jbudorick@chromium.org
We're not planning on reconstructing the fleet the way skia did; we want to get our existing hosts and devices on swarming, keeping the 7 devices per host.

Does a dimension for device 1-7 seem like the right way to go?
Cc: mikec...@chromium.org
Ping - please provide an update to your high priority bug. This bug is stale. Is it really P-1?

Comment 8 by stip@chromium.org, Feb 10 2017

Cc: -stip@chromium.org

Comment 9 by eyaich@chromium.org, Apr 10 2018

Status: Fixed (was: Assigned)

Sign in to add a comment