[Findit] Flake Analyzer - Determine # of trybots needed and set up accordingly |
||||
Issue descriptionDetermine per platform over a specified time period: 1. Average execution time of flake try job 2. Map request times over time 3. Determine spike sizes with window size = findings from 1. 4. Determine average and maximum spike size. 5. Findings from 3, 4 to determine number of trybots needed
,
Feb 16 2017
,
Apr 25 2017
Is this done already?
,
Apr 28 2017
The work for this is not done at the moment - moving down to priority 3. The reason being flake try jobs are run quite sporadically resulting in bots idling the majority of the time, so it's unlikely more than 1 of each is needed at any given time. However once we enable manual triggering try job reruns or increase the number of tests we automatically trigger, this will very likely change and become necessary to avoid long pending times since some try jobs can take upwards of an hour.
,
Jul 31
This hasn't been an issue in a long time, closing. |
||||
►
Sign in to add a comment |
||||
Comment 1 by lijeffrey@chromium.org
, Jan 18 2017