[Findit] Flake Analyzer - NextCommitPositionPipeline is suggesting to run a commit position that has already been analyzed |
|||
Issue descriptionhttps://findit-for-me.appspot.com/waterfall/flake?key=ag9zfmZpbmRpdC1mb3ItbWVypgELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCJwY2hyb21pdW0ubWVtb3J5L0xpbnV4IE1TYW4gVGVzdHMvOTM5Mi9icm93c2VyX3Rlc3RzL1UyVnpjMmx2YmxKbGMzUnZjbVZVWlhOMExsSmxjM1J2Y21WUWFXNXVaV1JUWld4bFkzUmxaRlJoWWc9PQwLEhNNYXN0ZXJGbGFrZUFuYWx5c2lzGAEM Generator pipelines.flake_failure.determine_approximate_pass_rate_pipeline.DetermineApproximatePassRatePipeline(*(u'ENCODED:boHH6HlAZEbA0jDg7NHrcEFIN',), **{'analysis_urlsafe_key': u'ag9zfmZpbmRpdC1mb3ItbWVypgELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCJwY2hyb21pdW0ubWVtb3J5L0xpbnV4IE1TYW4gVGVzdHMvOTM5Mi9icm93c2VyX3Rlc3RzL1UyVnpjMmx2YmxKbGMzUnZjbVZVWlhOMExsSmxjM... (460 bytes))#f759bfa3881e4654852eab5a0ea5f9ae raised exception. AssertionError: (/base/data/home/apps/s~findit-for-me/waterfall-backend:15111-6a36582.409252435710767791/infra_api_clients/../third_party/pipeline/pipeline.py:2434) Traceback (most recent call last): File "/base/data/home/apps/s~findit-for-me/waterfall-backend:15111-6a36582.409252435710767791/infra_api_clients/../third_party/pipeline/pipeline.py", line 2207, in evaluate yielded = pipeline_iter.send(next_value) File "/base/data/home/apps/s~findit-for-me/waterfall-backend:15111-6a36582.409252435710767791/gae_libs/pipelines.py", line 295, in run sub_pipeline = pipeline_iter.send(next_future) File "/base/data/home/apps/s~findit-for-me/waterfall-backend:15111-6a36582.409252435710767791/pipelines/flake_failure/determine_approximate_pass_rate_pipeline.py", line 61, in RunImpl assert previous_swarming_task_output AssertionError
,
Apr 27 2018
,
May 4 2018
The following revision refers to this bug: https://chromium.googlesource.com/infra/infra/+/cbdbc07ede5760d839736786baaec9895100f3c8 commit cbdbc07ede5760d839736786baaec9895100f3c8 Author: Jeffrey Li <lijeffrey@chromium.org> Date: Fri May 04 20:45:40 2018 [Findit] Flake Analyzer - Fixing running existing data points The root cause is a corner case when exponential search lands exactly on a data point that has already been ran, when the distance is exactly a square number. There is an off-by-1 error checking for when to restart exponential search from 1, causing the existing data point to be suggested to run again, which causes problems later on in the analysis. 1. Fix the off-by-1 error 2. Add assertion checks to ensure existing data points are never suggested again 3. Add assertion checks for when a suspected build is identified to ensure it is indeed a suspect (stable -> flaky). Bug: 836976 Change-Id: I68e0f8014a9efe8e01009bd380a9d255a9806941 Reviewed-on: https://chromium-review.googlesource.com/1043606 Reviewed-by: Chan Li <chanli@chromium.org> Commit-Queue: Jeffrey Li <lijeffrey@chromium.org> [modify] https://crrev.com/cbdbc07ede5760d839736786baaec9895100f3c8/appengine/findit/pipelines/flake_failure/next_commit_position_pipeline.py [modify] https://crrev.com/cbdbc07ede5760d839736786baaec9895100f3c8/appengine/findit/model/flake/master_flake_analysis.py [modify] https://crrev.com/cbdbc07ede5760d839736786baaec9895100f3c8/appengine/findit/services/flake_failure/test/lookback_algorithm_test.py [modify] https://crrev.com/cbdbc07ede5760d839736786baaec9895100f3c8/appengine/findit/model/flake/test/master_flake_analysis_test.py [modify] https://crrev.com/cbdbc07ede5760d839736786baaec9895100f3c8/appengine/findit/services/flake_failure/lookback_algorithm.py
,
Aug 31
|
|||
►
Sign in to add a comment |
|||
Comment 1 by lijeffrey@chromium.org
, Apr 27 2018