opendev/subunit2sql
Code Issues Proposed changes
60f40c76e5
subunit2sql/doc/source/data_model.rst
Matthew Treinish ee6a359650 Add microsecond columns to the test_runs table 
This commit adds a migration to add 2 columns to the test_runs table to
store microseconds for the start and stop time stamps for each test_run
row. Previously we were relying on the microseconds to be carried over
from the datetime object passed into the sqlalchemy object, however in
certain configurations this data was being lost. To avoid this in the
future this decouples the microseconds from the timestamp so we're no
longer dependent on the underlying db to store this correctly.

As part of this it adds a new config flag to skip the functionally
optional parts of migrations. If there are operations which are
nominally optional in that if they weren't run the functionality
wouldn't be any difference in the result given certain DB configurations
(in this case if the microseconds were already stripped by the db)

Change-Id: Ibaafb7d8fc8a8e8aaf7b96672d5c47f46180e0ca
Story: #2000096
2015-03-05 11:58:55 -05:00

2.9 KiB
Raw Blame History

The subunit2sql Data Model

The subunit2sql data model consists of 3 basic data types: runs, tests, and test_runs. Each of these 3 types have an associated key value pair metadata table to store arbitrary metadata about any specific row.

Runs

Runs represent an individual test run, or in other words, a complete subunit stream. They are used to track how many streams have been stored in the db and high level information about them

Properties:

  • passes: The total number of successful tests in the run.
  • fails: The total number of failed tests during the run.
  • skips: The total number of skipped tests during the run.
  • run_time: The sum of the duration of executed tests during the run. Note, this is not the time it necessarily took for the run to finish. For example, the time for setUpClass and tearDownClass (assuming the stream is from a python unittest run) would not be factored in. (as they aren't stored in the subunit stream) Also, if the tests are being run in parallel since this is just a raw sum this is not factored in.
  • artifacts: An optional link to where the logs or any other artifacts from the run are stored.
  • run_at: The time at which the run was stored in the DB.

Tests

Tests are the higher level grouping of unique tests across all runs. They are used to aggregate the information about an individual tests from all the runs stored in the db.

Properties:

  • test_id: This would be normally be considered the test name, it is the id used in the subunit stream for an individual test
  • success: The total number of times this test has been run successfully
  • failure: The total number of times this test has failed
  • run_count: The total number of times this test has been executed (obviously this excludes skips) it should be the sum of the success and failure columns
  • run_time: The moving average of the total duration of each test execution

Test Runs

Test runs represent the individual execution of a test as part of run. They are used for recording all the information about a single test's run.

Properties:

  • test_id: The uuid representing the test which was run
  • run_id: The uuid representing the run this was part of
  • status: The outcome of the test. The valid values here are: exists, xfail, unxsuccess, success, fail, skip. You can refer to the testtools documentation for the details on each status.
  • start_time: The timestamp when test execution started
  • start_time_microsecond: The microsecond component of the timestamp when

    test execution started

  • stop_time: The timestamp when the test finished executing
  • stop_time_microsecond: The microsecond component of the timestamp when

    test execution finished