⁉️ When pipeline is re-run for this user, it doesn't generate cleaned objects

[JGreenlee]

• Using e-mission-server master

• Load the Apr 24 dump for dfc-fermata

• Inspect the cleaned objects for my opcode:

  import emission.storage.timeseries.abstract_timeseries as esta
  import emission.core.wrapper.user as ecwu
  
  opcode = "nrelop_dfc-fermata_test_████"
  keys = [
    "segmentation/raw_trip",
    "analysis/cleaned_trip",
    "analysis/cleaned_place",
    "analysis/cleaned_untracked"
  ]
  
  user_id = ecwu.User.fromEmail(opcode).uuid
  ts = esta.TimeSeries.get_time_series(user_id)
  for key in keys:
    print(key)
    entries_df = esta.TimeSeries.get_time_series(user_id).get_data_df(key)
    print(len(entries_df))

  Output:

  segmentation/raw_trip
  39
  analysis/cleaned_trip
  37
  analysis/cleaned_place
  49
  analysis/cleaned_untracked
  11
  

• Reset pipeline for user
  ./e-mission-py.bash bin/reset_pipeline.py -e nrelop_dfc-fermata_test_████

• Run pipeline for user
  ./e-mission-py.bash bin/debug/intake_single_user.py -e nrelop_dfc-fermata_test_████

  "Cleaning and resampling failed"

  2024-05-06 01:43:14,441:DEBUG:140704544475072:Found 3 sections, need to remove the uncommon ones...
  2024-05-06 01:43:14,507:DEBUG:140704544475072:section counts = [(ObjectId('66386db4b3b578d6ae775005'), 47272), (ObjectId('66386dd9b3b578d6ae7808ae'), 144), (ObjectId('66386ddab3b578d6ae780941'), 120)]
  2024-05-06 01:43:14,507:ERROR:140704544475072:Section counts = [47272, 144, 120], expecting 1
  2024-05-06 01:43:14,509:ERROR:140704544475072:Cleaning and resampling failed for user 4acedfd2-de6a-4623-a968-9d5046a8573b
  Traceback (most recent call last):
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 88, in clean_and_resample
      last_raw_place = save_cleaned_segments_for_ts(user_id, time_query.startTs, time_query.endTs)
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 114, in save_cleaned_segments_for_ts
      return save_cleaned_segments_for_timeline(user_id, tl)
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 140, in save_cleaned_segments_for_timeline
      (last_cleaned_place, filtered_tl) = create_and_link_timeline(tl, user_id, trip_map)
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 971, in create_and_link_timeline
      link_trip_start(ts, curr_cleaned_trip, curr_cleaned_start_place, raw_start_place)
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 1024, in link_trip_start
      _fix_squished_place_mismatch(cleaned_trip.user_id, cleaned_trip.get_id(),
    File "/Users/jgreenle/openpath/e-mission-server/emission/analysis/intake/cleaning/clean_and_resample.py", line 1169, in _fix_squished_place_mismatch
      assert False
  AssertionError
  

• Now we have no cleaned objects (and thus no confirmed or composite objects)

  segmentation/raw_trip
  31
  analysis/cleaned_trip
  0
  analysis/cleaned_place
  0
  analysis/cleaned_untracked
  0
  

[shankari]

The assertion errors are validation or quality control checks on the pipeline. They are turned on by default in dev versions so we can see what is going wrong. In this case, it is almost certainly

2024-05-06 01:43:14,507:ERROR:140704544475072:Section counts = [47272, 144, 120], expecting 1

But if you would like to ignore that for now since it is not a section you worked on, you can turn off the various assertion errors in conf/analysis/debug.conf.json

[JGreenlee]

In that case, it should not affect production right?
This opcode does not have any trips on production either

[shankari]

It doesn't have any trips on production because I upgraded DFC fermata to the version with the BLE matching, and reset the pipeline for all users so that the trips and sections would be re-created and we could generate a snapshot with the correct data format for public dashboard testing.
e-mission/em-public-dashboard#124 (comment)

The trips should be showing up soon, or any errors will be in the intake pipeline logs.

[JGreenlee]

Ah I see, I thought it would have finished processing by now

[JGreenlee]

Circling back to this and confirming that I was able to get my trips to process locally by turning off the assertion errors.

I'd still like to know why my sections were considered "invalid" (I've also seen logs saying "This is messed up segment. Investigate further while processing section, skipping...")
But that might be a hairy topic that we don't have time to dive into now