Integrate remote cache server for Bazel #21
Conversation
|
Is the |
|
@chickenandpork It's included intentionally, so I could test the branch in GHA directly. This however is not intended to be used directly as it's not really working stable. Are you interested in taking over and finishing this PR? |
OK, I understand; it looked like an error of duplicate code!
I don't think my javascript is sufficient. I only worked javascript (and coffeescript!) briefly around 2020. Sorry :( Personally I use a local buchgr cache and I set up the same in ephemeral build clusters. |
I'm interested in picking this up. What's not stable? What else is outstanding? |
|
@JohnAmican The summary is in #18 (comment), the stability is the main issue as on large projects the REST API calls to the GHA cache get rate limited. |
Related to #18
This pull request introduces a significant enhancement to the setup-bazel action by implementing a remote cache server for Bazel build outputs, aiming to improve build performance and cache utilization.
remote-cache-server.jsthat creates an HTTP server compatible with Bazel's remote caching API. This server handlesGETandPUTrequests to store and retrieve cache objects, leveraging@actions/cachefor underlying storage.index.jsto start the remote cache server as a detached process during the action's execution. The server's URL is then added to.bazelrcto enable Bazel to use it for caching.config.jsto include settings for the remote cache server's URL and log file path, ensuring that these new features are configurable.post.jsto stop the remote cache server gracefully at the end of the action's execution and logs the server's output for debugging purposes.package.jsonto include the remote cache server in the build process, ensuring that all necessary components are compiled and ready for use.For more details, open the Copilot Workspace session.