 Danny Kopping
|
84b7a0364d
|
chore: abstract pg test logic and double runner sizes (#21091)
This PR does two things, both in service of helping to (hopefully!)
speed up CI:
1. abstracts the parallelism logic into a common action and has all
PG-related jobs use it
2. doubles runner sizes from [8->16 CPUs & 32->64GiB
RAM](https://depot.dev/docs/github-actions/runner-types)* and
concomitantly increases parallelism
I only focused on the PG-related jobs since they are generally slowest &
most RAM-intensive.
[<img width="2011" height="460" alt="image"
src="https://github.com/user-attachments/assets/c38ec3fc-dd93-4753-8df3-f22dfe54a3f7"
/>](https://app.datadoghq.com/ci/pipelines/health?query=%40git.repository.id_v2%3Agithub.com%2Fcoder%2Fcoder%20-%40git.is_default_branch%3Atrue&fromUser=true&group=ci-cost&sort=-billableTime&sp=%5B%7B%22p%22%3A%7B%22fingerprint%22%3A%22pVB6pq7htXrn%22%2C%22env%22%3A%22none%22%7D%2C%22i%22%3A%22ci-health-pipeline-panel%22%7D%5D&start=1762245172958&end=1764837172958&paused=false)
_* `test-go-race-pg` doubles from 16->32 CPUs & 64->128GiB RAM and
likewise for the Windows runners; MacOS runners have [only one
size](https://depot.dev/docs/github-actions/runner-types#macos-runners)_
_**NOTE:** don't use the speed of the PG-related jobs in this PR's CI
run as indicative. Tests run outside `main` may use cache, so the speed
may seem artificially low._
---------
Signed-off-by: Danny Kopping <danny@coder.com>
|
2025-12-11 10:12:47 +00:00 |
|