You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The caching mechanism in unit-test.yaml isn't working as I would expect in our component. The first time the workflow runs on a PR, there is a cache miss, HOOMD builds, and is successfully stored, e.g.,:
like I would expect. However, after I push another commit to the PR branch, a cache miss is still reported so HOOMD gets built again, but now a warning is raised because the cache key already exists!
It seemed like the intent of caching was that the base build of HOOMD would persist within a PR if not across all workflow runs. Right now, we are spending ~30 minutes rebuilding HOOMD on every PR commit... any ideas what's going on here? Are you able to reproduce the same issue in your repo?
The text was updated successfully, but these errors were encountered:
It seemed like the intent of caching was that the base build of HOOMD would persist within a PR if not across all workflow runs.
Yes. Practically this is limited to a single PR due to cache scoping and a short expiration time.
Right now, we are spending ~30 minutes rebuilding HOOMD on every PR commit... any ideas what's going on here? Are you able to reproduce the same issue in your repo?
The cache was working when I originally implemented it, but it no longer functions. Many users of action/cache have the same problem when using actions/cache within a container:
I have attempted many of these workarounds and was still not able to fix the problem (see e.g. #9, #12). GitHub has been silent on the issue. The only commits to action/cache in years have been documentation updates surrounding the removal of v2 (which doesn't have this issue) and small spelling fixes in the docs.
sigh Thanks for the explanation. For now, we will disable building HPMC to save time since we don't need it for our component. Hopefully GitHub implements a fix for this.
The caching mechanism in
unit-test.yaml
isn't working as I would expect in our component. The first time the workflow runs on a PR, there is a cache miss, HOOMD builds, and is successfully stored, e.g.,:https://github.com/mphowardlab/azplugins/actions/runs/12917804952/job/36025002876
like I would expect. However, after I push another commit to the PR branch, a cache miss is still reported so HOOMD gets built again, but now a warning is raised because the cache key already exists!
https://github.com/mphowardlab/azplugins/actions/runs/12920560098/job/36033082022
It seemed like the intent of caching was that the base build of HOOMD would persist within a PR if not across all workflow runs. Right now, we are spending ~30 minutes rebuilding HOOMD on every PR commit... any ideas what's going on here? Are you able to reproduce the same issue in your repo?
The text was updated successfully, but these errors were encountered: