[chore] make contrib tests more efficient #12490
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #12490 +/- ##
==========================================
- Coverage 92.00% 91.99% -0.01%
==========================================
Files 469 469
Lines 25355 25355
==========================================
- Hits 23327 23325 -2
- Misses 1619 1620 +1
- Partials 409 410 +1 ☔ View full report in Codecov by Sentry. |
atoulme
force-pushed
the
prepare_contrib
branch
2 times, most recently
from
fc7ef2c to
7cae2eb
Compare
atoulme
force-pushed
the
prepare_contrib
branch
from
7cae2eb to
ed455e1
Compare
9861e0a
| contrib_path=/tmp/opentelemetry-collector-contrib | ||
| git clone --depth=1 https://github.com/open-telemetry/opentelemetry-collector-contrib.git $contrib_path | ||
| make CONTRIB_PATH=$contrib_path prepare-contrib | ||
| - uses: actions/upload-artifact@v4 |
There was a problem hiding this comment.
Doesn't it mean that we will be testing against one contrib version for 90 days?
There was a problem hiding this comment.
I'm looking at https://github.com/actions/upload-artifact?tab=readme-ov-file#retention-period. Should we add git sha in the path ?
There was a problem hiding this comment.
ouch. I'll open a new PR to add that
Actually, no, it might not be a problem, since this happens every time and is used inside the same build. But we can certainly do it anyway.
There was a problem hiding this comment.
Why do we need to upload it then? Should we use https://github.com/actions/cache instead?
There was a problem hiding this comment.
I followed the recipe I have seen used most of the time. A cache is a step better I didn't look into because I was afraid of having cache issues. That would shave an additional 4 minutes off the build.
Instead of having each test group edit modules, tidy and generate, do it once and share the output in each subtest to save 5 minutes per branch.