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
{{ message }}
Repository navigation
Sbt dependency cache key is calculated incorrectly, causing slowness & inconsistent run times #477
Description:
Sbt dependency cache key is calculated incorrectly.
Task version:
v3.11.0
Platform:
Ubuntu
macOS
Windows
Runner type:
Hosted
Self-hosted
Repro steps:
Run this workfile on some scala project, later modify some scala or sbt file under project/ and compare hashes.
name: Test Hashfiles
on: push
jobs:
test:
# Detect which parts of the application has changed. Such as API code, db migrations etc.
name: Test
runs-on: ubuntu-22.04
steps:
- uses: actions/checkout@v3
with:
# arbitrary number to fetch enough number of commits in a PR, instead of whole history.
fetch-depth: 1000
- name: Test Before
shell: bash
run: echo "${{ hashFiles('**/*.sbt', '**/project/build.properties', '**/project/**.{scala,sbt}') }}"
- name: Test After
shell: bash
run: echo "${{ hashFiles('**/*.sbt', '**/project/build.properties', '**/project/**.scala', '**/project/**.sbt') }}"
In our case, before modification of api/project/Dependencies.scala:
Test Before: 09f3783957f7496e05c5b7618773c57281bfdf1f33f8cb10e94e88579c3e5308
Test After: 74c085d46b2d61ad084b24ff74f0a1f8252fc91a0b72c5aa6717c6639f1494fc
After modification:
Test Before: 09f3783957f7496e05c5b7618773c57281bfdf1f33f8cb10e94e88579c3e5308
Test After: cf44e926587b9345fe8347c6df516a5d6ef95346eb13731bd98634cedf122a4f
As you can see cache is calculated incorrectly.
Expected behavior:
When an sbt or scala file has changed under
Actual behavior:
All sbt and scala files are ignored under **/project/ and this is causing the cache to act slower / inconsistent. I also realized pruning cache was speeding up our CI runs a lot but they were slowing down by time and they were getting slower and slower almost every day.
Activity
Dogacel commentedon Apr 6, 2023
Opened a PR here: #478
IvanZosimov commentedon Apr 7, 2023
Hi, @Dogacel 👋 Thanks for the issue and PR, we will take a look!
Dogacel commentedon Apr 10, 2023
As PR #478 is merged, I am closing the issue.