-
Notifications
You must be signed in to change notification settings - Fork 428
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
[VL][CI] Use cached Velox build binary to accelerate GHA workflow #7474
base: main
Are you sure you want to change the base?
Conversation
Thanks for opening a pull request! Could you open an issue for this pull request on Github Issues? https://github.com/apache/incubator-gluten/issues Then could you also rename commit message and pull request title in the following format?
See also: |
steps: | ||
- uses: actions/checkout@v2 | ||
- name: Generate cache key | ||
run: | | ||
echo ${{ hashFiles('./ep/build-velox/src/**', './dev/**', './cpp/*', './.github/workflows/*') }} > cache-key | ||
echo ${{ hashFiles('./ep/build-velox/src/**', './dev/**', './.github/workflows/velox_backend.yml') }} > cache-key |
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.
./ep/build-velox/src/**
I think we could change our OAP Velox dependency from branches to tags while landing this PR. Branches can be updated without invalidating the new cache.
Edit: Similar issue already existed even with current cache policy. It's fine till now so no need to put to high priority.
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.
@zhztheplayer, yes, this is a potential issue. But before a velox rebase branch is finalized, we may still need oap/velox to hold it as a branch that allows updating (e.g., needs some fixes due to Gluten CI failure). So maybe, we should just have a internal convention to forbid updating oap/velox branch once it is already referenced by Gluten code.
What changes were proposed in this pull request?
In the current main code,
libgluten.so
andlibvelox.so
are cached. If the cache is hit (depending on the hashing result on some directories includinggluten/cpp
), they can be re-used to accelerate workflow. However, any code change forgluten/cpp
in a pr will make the cache missed. Then, needs to re-build Velox code, which is quite time-consuming.This pr makes Velox build binary cached. If cache is hit, directly let the workflow build Gluten CPP, which only requires few minutes. Otherwise, re-build Velox code.
For Gluten CI's static build that build_test/benchmark are not enabled, we only need to cache
libvelox.a
. For Gluten CI's dynamic build that build_test/benchamrk are enabled, building Gluten cpp needs some other velox libs, likelibdbgen.a
, and one extra generated header, so we just cache the whole velox build path.How was this patch tested?
CI.