-
Notifications
You must be signed in to change notification settings - Fork 56
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
ci: Transition to self-hosted runners #598
ci: Transition to self-hosted runners #598
Conversation
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.
I'm OK with all those tests, except perhaps the removal of ARM linux:
- it may be useful for cloud runners
- it already allowed us to find issues with dependencies
Pinging @porcuquine for an opinion
All else equal, I don't favor reducing test surface — especially if this has actually uncovered errors. There was also this: supranational/blst@4f9d557#diff-78a8613e47fd3a587897556bdbcaf5c3b1bd53400cbd57f01322c98f2ebe1147 — thought that was indeed M1. If we're certain the M1 coverage would catch all the same bugs of this category, and the cost savings warrant it, then removing seems okay. If the cost is relatively minor, I'd generally favor keeping it. I think given the full range of use cases, Lurk on a raspberry pi is not even entirely far-fetched — even if not used for heavy proving… |
And yes, to @huitseeker's point about cloud hardware. Basically, I think explicit support is worthwhile. |
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.
Thanks a lot!
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.
Once more unto the breach ...
arm64
job on CircleCI as we are already testing on ARM via M1 Mac, and I'm not sure we need first-class support for ARM on Linux.linux_release
CircleCI job to self-hosted GH Actions aslinux_ignored
linux_ignored
onmerge_group
only, as it is expensive to run and seems wasteful to trigger on every PR commit. This is a speculative change and fine to revert, but it may cause bottlenecks on the self-hosted runners.linux_exhaustive
CircleCI job to self-hosted GH Actions. The cron job currently runs at midnight UTC time, perhaps it would be better in a timezone like ET?CC=gcc-10 CXX=g++-10
due to the new CUDA-enabled Docker image in Switch to CUDA base image gh-actions-runner#12. This closes ci: GPU runners requireCC=gcc-10 CXX=g++-10
with CUDA 11.5 #597Fixes #504