-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Service and profiler tests failing on dartk-linux-release-arm64 and app_jitk-linux-debug-x64 #46746
Comments
While CL 208321 (for 9397b8f) showed all the failures on dartk-linux-debug-ia32 and vm-kernel-optcounter-threshold-linux-release-ia32-try to be fixed (the latter having one test previously approved for a Timeout returning to that state), it also showed all the failures on app_jitk-linux-debug-x64 to be fixed as well but one showed up on the non-trybot, so I imagine there's at least one more similar heisenbug remaining. |
Given that the fix I added was needed even before the original commit landed, I expect it may not be an issue with the commit itself, but just that it exposed already existing issues that were less likely to trigger before. |
…UserTag" This reverts commits 746b8f1 and 9ee2259. Reason for revert: #46746 Original change's description: > [ package:dds ] Add support for caching CPU samples based on UserTag > > DDS can be configured to listen for CPU sample events and cache samples > that were collected while certain UserTags are active. These cached > samples are stored in a ring buffer and are stored until the isolate > shuts down. > > TEST=pkg/dds/test/get_cached_cpu_samples_test.dart > > Change-Id: Ib20770f59f1672c703413486f87795b3bb23f676 > Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/207206 > Commit-Queue: Ben Konyi <bkonyi@google.com> > Reviewed-by: Kenzie Schmoll <kenzieschmoll@google.com> TEST=ci TBR=bkonyi@google.com,rmacnak@google.com,kenzieschmoll@google.com Change-Id: I1b6655ad7e3b10e1145ff545cc90ecf3bc6e092d No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/208341 Commit-Queue: Alexander Markov <alexmarkov@google.com> Reviewed-by: Siva Annamalai <asiva@google.com>
Unfortunately, I can't seem to reproduce these issues on my Linux box... :-( this is going to be a fun one... Also @sstrickl, in the future could you please provide links to logs for a subset of the failures? The builder names and test configuration names don't match up 1:1, so it's a pain to go digging for them in the tree. Having even just one log per failure is a huge help! |
@bkonyi wrote
Should https://dart.googlesource.com/sdk/+/17961606060fa36971fdc40cee8105a13262e97e be reverted while you investigate the failures? |
No, the failures are approved for now and I have a fix for the failures I could reproduce here. The failures are likely due to us running out of sample buffer space, so increasing the number of sample blocks should the failures. |
It looks like https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-nnbd-linux-debug-ia32/286, https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-linux-debug-ia32/7219, https://ci.chromium.org/p/dart/builders/luci.dart.ci.sandbox/cross-vm-linux-release-arm64 all seems to be affected by reland. service/get_allocation_samples_test, service/get_allocation_traces_test, vm/cc/Profiler_BasicSourcePositionOptimized, vm/cc/Profiler_FunctionInline, vm/cc/Profiler_SourcePositionOptimized, vm/cc/Profiler_BinaryOperatorSourcePositionOptimized are the tests that stared to fail with reland on ia32, arm64 bots. |
Main concern with this is that is by approving failures we are potentially disrupting/breaking things downstream from dart sdk(flutter, google). |
I strongly believe that the fix in the CL linked above will resolve most, if not all of the issues. Given that this is next to impossible to reproduce locally on machines with various performance characteristics and we're not actually seeing crashes, just test failures, reverting the reland without landing the potential fix first would be counterproductive. |
Not sure I understand. Revert can be landed quickly as it has no conflicts, the fix can go together with the reland. |
Going to take a different approach for the fix. Revert up here. |
After 746b8f1, there were a lot of service test failurs on a few number of bots. 9397b8f fixes most of the tests, but there are still a handful remaining from runs after it landed:
On dartk-linux-release-arm64 and app_jitk-linux-debug-x64, the following tests are still failing:
On dartk-linux-release-arm64, the following additional tests are still failing:
Given the number of remaining failing tests, I'm going to approve them for now.
cc @bkonyi
The text was updated successfully, but these errors were encountered: