Repository navigation
Conversation
One or more co-authors of this pull request were not found. You must specify co-authors in commit message trailer via: Supported
Alternatively, if the co-author should not be included, remove the Please update your commit message(s) by doing |
|
Thanks for the fix. Please apply the following patch (from CI logs) to fix format errors: Please sign the EasyCLA, so this fix can be merged. |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #4718 +/- ##
==========================================
+ Coverage 86.51% 86.52% +0.01%
==========================================
Files 530 530
Lines 21138 21138
==========================================
+ Hits 18286 18288 +2
+ Misses 2852 2850 -2 🚀 New features to boost your workflow:
|
Some unit tests share resources, which causes them to fail if executed in parallel. * curl_http_test uses port 19000 * prometheus exporter_test uses port 9464, bound by the prometheus-cpp Exposer that the exporter constructor starts RESOURCE_LOCK is a named mutex between tests: cases holding the same lock never run concurrently. Using two distinct lock names keeps the two groups parallel with each other and with everything else. See https://cmake.org/cmake/help/v4.5/prop_test/RESOURCE_LOCK.html Signed-off-by: Nicolas Benes <nbenes.gh@xandea.de> Co-authored-by: Jean-François Roche <jfroche@pyxel.be> Co-authored-by: Nicolas Benes <nbenes.gh@xandea.de>
86e8e0a to
a67224a
Compare
|
/easycla |
1 similar comment
|
/easycla |
mateenali66
left a comment
There was a problem hiding this comment.
Format still fails, now on the prometheus file. The job log has the patch.
On main 374145e, ctest -j fails both groups: prometheus exporter_test in 41 of 100 runs on the 9464 bind, curl 7 to 11 of 43 cases in each of 5 runs. With this PR json-v1 shows the lock on all 44 curl and 3 prometheus entries, and every parallel run passed.
Heads-up on #4730, which moves to gtest_discover_tests. Its test list only exists inside ctest, so merged with it ${curl_http_test} and ${exporter_test} are empty at configure time, configure stays quiet and the lock lands on 0 tests. The prometheus hunk merges with no conflict. gtest_discover_tests(... PROPERTIES RESOURCE_LOCK ...) put it back on all of them.
#4730 also splits the OTLP HTTP retry TEST_P (port 4318) into 40 entries, and 37 to 39 of 55 failed under -j8 there. Should otelcpp_port_4318 go in here, or with #4730?
Some unit tests share resources, which causes them to fail if executed in parallel.
RESOURCE_LOCK is a named mutex between tests: cases holding the same lock never run concurrently. Using two distinct lock names keeps the two groups parallel with each other and with everything else.
See https://cmake.org/cmake/help/v4.5/prop_test/RESOURCE_LOCK.html
Changes
Please provide a brief description of the changes here.
For significant contributions please make sure you have completed the following items:
CHANGELOG.mdupdated for non-trivial changes