emit step timings as perfherder metrics
Categories
(Release Engineering :: Applications: MozharnessCore, enhancement, P2)
Tracking
(firefox147 fixed)
| Tracking | Status | |
|---|---|---|
| firefox147 | --- | fixed |
People
(Reporter: bhearsum, Assigned: bhearsum)
Details
Attachments
(3 files)
We have an ongoing effort to improve try push turnaround times. One area where we have a blind spot at the moment is the breakdown of time spent on different parts of tasks. For the majority of the tasks that we care about (builds and tests), having timings for the different mozharness steps they run would be a sufficient starting point. In particular, this would let us see how much time is spent in set-up vs. actually making builds or running tests.
| Assignee | ||
Comment 1•8 months ago
|
||
This paves the way for recording how long each action takes, while allowing us to easily ignore skipped actions.
This causes no change in behaviour as there's only one caller of run_action, and the logic being moved now runs directly before the call.
| Assignee | ||
Comment 2•8 months ago
|
||
This gives us a consistent environment variable to look for to find the right place to put artifacts. This will make a subsequent patch that is writing a perfherder artifact deep in the bowels of mozharness to easily write to the correct place.
| Assignee | ||
Comment 3•8 months ago
|
||
We have an ongoing effort to improve try push turnaround times. One area where we have a blind spot at the moment is the breakdown of time spent on different parts of tasks. For the majority of the tasks that we care about (builds and tests), having timings for the different mozharness steps they run would be a sufficient starting point. In particular, this would let us see how much time is spent in set-up vs. actually making builds or running tests.
Comment 5•8 months ago
|
||
| bugherder | ||
Description
•