-
Notifications
You must be signed in to change notification settings - Fork 3.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
switch high-perf-docker to runs-on.com runners #13658
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #13658 +/- ##
===========================================
- Coverage 70.8% 58.9% -11.9%
===========================================
Files 2299 817 -1482
Lines 451793 195823 -255970
===========================================
- Hits 319875 115392 -204483
+ Misses 131918 80431 -51487 ☔ View full report in Codecov by Sentry. |
1ebb379
to
064b008
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
5e2a4c1
to
3214c47
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
b46ffc9
to
22b89f2
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
✅ Forge suite
|
✅ Forge suite
|
@@ -1,4 +1,4 @@ | |||
self-hosted-runner: | |||
# Labels of self-hosted runners in array of string | |||
labels: | |||
- high-perf-docker | |||
- runs-on,cpu=64,family=c7,hdd=500,image=ubuntu22-full-x64,run-id=${{ github.run_id }} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is this necessary ? It looks like a label not a runner spec.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is just to tell the actionlint linter not to complain about this in the action yamls.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should build and use a custom AMI soon with the required tools.
This reverts commit 916dba1.
This switches high-perf-docker (c2-standard-60) to roughly equivalent runs-on.com runners. Specifically AWS machines of the c7 family (runs-on has some heuristics to pick the cheapest spot instance in the c7 family that satisfies the CPU requirement of 64 CPUs).
Test Plan
Ran several tests a bunch of times. Compared performance to baseline PR (#13698 ). Performance is roughly similar, usually a bit faster (5-15%), few times a bit slower (less than 5%).
Re spot preemption: Haven't seen a single spot preemption so far (added a log-based chart here https://aptoslabs.grafana.net/d/cb066e70-b378-4de7-aadb-79b43386e664/gha-self-hosted-runners?orgId=1&from=now-30d&to=now ).
Note: Some of the checks below are failing but to the best of my ability they are just broken and also fail in the baseline PR that uses high-perf-docker.
Something to note
Runs-on in some cases actually picks machines with higher CPU count (96+ etc.). This due its heuristics of picking the cheapest instance, (sometimes the bigger machines are cheaper, who knew).
Probably not a problem for most jobs, but may lead to some surprised for perf/benchmarking jobs.
I think its worth the tradeoff for these types of jobs anyways.