Recall functions are uncovered but truly executed.

[nedbat]

Originally reported by rebecca (Bitbucket: liulrebecca, GitHub: Unknown)

---

in one python file called vault.py, line 148 is executed, but function eh_evt_2_events and prepare_eh_evt_4_vault which are recalled are not marked as covered:
line 148: df = evt_hub_rdd.flatMap(lambda evt: eh_evt_2_events(evt)).map(lambda evt: prepare_eh_evt_4_vault(evt, batch_date, batch_key)).map(lambda evt_dict: Row(**evt_dict)).toDF()

This file is to process data and save to vault: Uploaded data are processed and saved in vault as expected when testing.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/657
• This issue had attachments: vault_py.html. See the original issue for details.

[nedbat]

@liulrebecca I don't understand all that's going on here because I don't have all the code. But send_queue_message sounds like the work is happening in a different process? You should read up on how to measure subprocesses (http://coverage.readthedocs.io/en/latest/subprocess.html) and see if that helps.

[nedbat]

Original comment by rebecca (Bitbucket: liulrebecca, GitHub: Unknown)

---

@nedbat Thanks for your help. I will read up the articles.
BTW, the codes are running in a spark cluster, we found that some codes are executed in worker nodes, but coverage.py is running in the master node, maybe it's the root cause why some codes are not marked as covered. I will try again.

[nedbat]

Issue #658 is also about PySpark, but I don't know enough to know if they are related.

[nedbat]

Original comment by Abdeali Kothari (Bitbucket: AbdealiJK, GitHub: AbdealiJK)

---

The #658 is running on a single machine (local-mode) of spark. So, it may not be (but not sure myself)

[nedbat]

Do these help?

• https://github.com/nvembar/spark-coverage-example
• https://stackoverflow.com/a/67184714/14343