New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 664161 link

Starred by 1 user

Issue metadata

Status: Archived
Owner:
Closed: Aug 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 1
Type: Bug-Regression



Sign in to add a comment

storage.indexeddb_endure_tracing failure on chromium.perf.fyi on Windows 10 Low End tests

Project Member Reported by eyaich@chromium.org, Nov 10 2016

Issue description

Most recent failing builder: 
https://uberchromegw.corp.google.com/i/chromium.perf.fyi/builders/Win%2010%20Low-End%20Perf%20Tests/builds/1125

Link to failing log: https://chromium-swarm.appspot.com/user/task/3266cfda3566a810

Seems to be hitting the maxium recursion depth, See trucated log below.  

Ethan Ben do you have any insight as to why this might be failing on a windows bot or what condition it is looking for to exit the recursion?

WARNING:root:Please specify the TBMv1 metrics you are interested in explicitly. This implicit functionality will be removed on July 17, 2016.
WARNING:root:No importer found for TraceDataPart("telemetry")
INFO:root:*************** BROWSER STANDARD OUTPUT ***************
INFO:root:
INFO:root:*********** END OF BROWSER STANDARD OUTPUT ************
INFO:root:********************* BROWSER LOG *********************
INFO:root:No log file
INFO:root:***************** END OF BROWSER LOG ******************
Traceback (most recent call last):
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\internal\story_runner.py", line 89, in _RunStoryAndProcessErrorIfNeeded
    test.Measure(state.platform, results)
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\web_perf\timeline_based_measurement.py", line 301, in Measure
    self._ComputeLegacyTimelineBasedMetrics(results, trace_result)
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\web_perf\timeline_based_measurement.py", line 329, in _ComputeLegacyTimelineBasedMetrics
    model = model_module.TimelineModel(trace_result)
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\timeline\model.py", line 75, in __init__
    self.ImportTraces(trace_data, shift_world_to_zero=shift_world_to_zero)
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\timeline\model.py", line 158, in ImportTraces
    self.FinalizeImport(shift_world_to_zero, importers)
  File "c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\timeline\model.py", line 173, in FinalizeImport
    process.FinalizeImport()
............
............
.........
    if self._AddSliceIfBounds(root.sub_slices[-1], child):
  _AddSliceIfBounds at c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\timeline\thread.py:263
    if self._AddSliceIfBounds(root.sub_slices[-1], child):
  _AddSliceIfBounds at c:\b\s\w\irkucyq7\third_party\catapult\telemetry\telemetry\timeline\thread.py:259
    child_end_micros = round(child.end * 1000)
RuntimeError: maximum recursion depth exceeded

 

Comment 1 by zh...@chromium.org, Nov 15 2016

ping
Apologies Zhen, Emily asked me about this bug on an email thread and I replied there. My response for this bug was:

All I can say about this one is that it's hitting a bug in the Python trace importer code. Zhen or Ned might be able to shed some light here. One of the major technical reasons we were big on switching to TBMv2 is that the canonical JS importer/model/etc code have proved much easier and more worthwhile to maintain properly than the Python versions. Along those lines, one thing I can suggest to debug here is that if you can somehow get the trace data that is being produced, try using trace-viewer to import it and see if it blows up; if it does you then have devtools to investigate why the trace might be malformed.
Cc: nedngu...@google.com zh...@chromium.org
+ned +zhen since the last comment mentioned they might be able to shed some light. Emily is out on leave, who would be an appropriate new owner for this?
Status: Archived (was: Untriaged)
This issue was created > 6 months ago. The perf waterfall has changed significantly since then. If this bug is still relevant, please re-open.

Sign in to add a comment