Hi all,the Gerrit master branch is currently broken because of PolyGerrit test failures.The first red build is:and the changes that possibly have broken it are:
On 21 Apr 2020, at 15:14, David Ostrovsky <david.o...@gmail.com> wrote:
Am Dienstag, 21. April 2020 15:52:59 UTC+2 schrieb lucamilanesio:On 21 Apr 2020, at 13:57, David Ostrovsky <david.o...@gmail.com> wrote:
Am Dienstag, 21. April 2020 12:00:13 UTC+2 schrieb lucamilanesio:Hi all,the Gerrit master branch is currently broken because of PolyGerrit test failures.The first red build is:and the changes that possibly have broken it are:Given that for both changes the PG tests were passing on GerritForge CI,can we bisect the problem?Actually, Dimitrii has found the issue and that tells me that when the Bazel rules are modified, we should run the tests on *everything*.I’ll leave Dimitrii describe the details :-) because the Bazel rules are still half of “black magic” to me, I need to study a lot more on it.I think I understand Bazel rules. So let me try to summarize what happened.In this CL: [1] that changed only PG related are of code, PG test ruleswere changed/added. One new test rule wasn't tagged with "manual" label.That means, that wildcard target execution:$ bazel test //...would trigger that test. However, PG test cannot be executed in that way,but must be started from shell script that initializes the environment.
Why this breakage was not detected in: [1]? Because of the CI optimizationmachinery we are skipping server check entirely wenn only PG area of codewas touched. That code is here: [2].
One possible solution would be to always trigger PG and server tests if anyBazel related files changed (Thomas started to work on this in: [3]),
or just removethe whole fuzzy logic with check skipping and always run all tests.
I think that with the switch to RBE and with activated remote caching we shouldbe fine running more test, not less tests.
--
--
To unsubscribe, email repo-discuss...@googlegroups.com
More info at http://groups.google.com/group/repo-discuss?hl=en
---
You received this message because you are subscribed to the Google Groups "Repo and Gerrit Discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to repo-discuss...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/repo-discuss/56f2385f-007c-43fe-b2ce-f35331d2b824%40googlegroups.com.