test: mark some tests flaky for now on riscv64 - #65781
Conversation
Signed-off-by: Stewart X Addison <sxa@ibm.com>
|
Review requested:
|
|
Seems to work - https://ci.nodejs.org/job/node-test-commit-rvlinux/nodes=riscv64/9/ shows yellow warning status instead of red with this change. |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #65781 +/- ##
==========================================
+ Coverage 89.96% 90.07% +0.10%
==========================================
Files 757 769 +12
Lines 258129 261396 +3267
Branches 48946 49636 +690
==========================================
+ Hits 232219 235445 +3226
+ Misses 16982 16968 -14
- Partials 8928 8983 +55 🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
There are other flaky WPT tests as well. I think marking only these two as flaky might not be enough.
#64027 lowers the concurrency of WPT tests further and solves the flakiness on SG2042.
There was a problem hiding this comment.
Yep that's fair - we can look at others separately. I'm about to go away on vacation and these five are the ones that were consistently showing problems in my CI job running on the K3 at https://ci.nodejs.org/job/node-test-commit-rvlinux/nodes=riscv64 which I've now scheduled to run daily.
Obviously the preferred solution is to try and make them work so thanks for your efforts in looking at that - I've approved your PR :-)
While RISC-V is an experimental platform I have been testing it in a CI job. The tests in here are currently causing problems, although I have had them all pass at one point from the command line. The reasons for them failing vary, but for now since I know they can pass I'd like to mark them flaky to stop the CI job going red while we try and improve things on this platform. Note that this is NOT currently included as a subjob of the main node-test-commit job so has no impact on Node.js collaborators. I want to be able to use that job to quickly identify regressions going forward and for that I need to be able to have a clean baseline. If anyone fundamentally objects to this approach on this experimental platform let me know.
Failures can be seen in https://ci.nodejs.org/job/node-test-commit-rvlinux/6/testReport
I'll tag @codebytere for awareness just because one of the tests here being the reproducible snapshot one and there probably aren't many people who know that area in enough detail to have any opinion on it :-) I doubt it's anything really obvious so not critical as I'm trying to skip it here but the error is in the log above if anyone wants to look at it.
Tracking issue for problems on RISC-V is at nodejs/build#4099
Cc @nodejs/platform-riscv64