[vikramlc-cognite]

Summary

fetch_logs uploads log files into the extractor's data set (when one is set), keeps a new copy on every run, and ends completed_with_errors when nothing was uploaded.

Type of change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that changes existing behavior)
• Refactor (no functional change)
• Documentation update
• Chore / tooling / CI

What changed

• New hook Extractor.get_log_upload_data_set_id(); the default uses the config data set when the config derives from WithDataSetId, otherwise none. A marked TODO(data-domains) placeholder for the future integration data domains field (not implemented).
• _log_upload_action.py: passes data_set_id to upload_bytes only when set; removes overwrite=True; external ID is now extractor-logs-<integration>-<date>-<action external id> (hashed if over 255 characters).
• Nothing uploaded -> completed_with_errors with "No log files were uploaded (0 of N ...)"; partial uploads stay succeeded; invalid parameters stay failed.
• Per-file status for a day with no log file is now missing (was skipped), to match the missing_files counter. Counter keys are unchanged.
• ActionContext.set_result(..., completed_with_errors=...); tests (about 12 existing ones updated); CHANGELOG.

Why it changed

• Uploaded log files had no data set, were silently overwritten by repeat runs, and an empty fetch reported succeeded.
• Related issue: EDG-1037 (parent: EDG-1031)
• Related docs / discussion: PM decisions (extractor's own data set, then none; new copy each run; nothing uploaded -> Completed with Errors). Stacked on EDG-1035. Needs Odin completed_with_errors (cognitedata/odin#324).

What to focus on during review

• The external-ID scheme and its 255-character fallback.
• The default data-set hook: it does one data_sets.retrieve for a configured data set; an exception fails the action.
• The visible change of the per-file status string from skipped to missing.

Test evidence

• tests/test_unstable/test_log_upload_action.py: 9 net new tests (data set passed/omitted, default hook, two runs -> different IDs without overwrite, ID length, partial succeeded, none uploaded completed_with_errors, invalid range failed, cancel wins)
• uv run pytest tests/test_unstable: 309 passed; the same 1 failed + 48 environment errors as master
• ruff, ruff format, mypy, pre-commit run --all-files pass
• 22 tests fail with the three production files reverted

Risks and unknowns

• Open product question: a partial result (some days uploaded, some missing) still ends succeeded.
• Log files accumulate (nothing is cleaned up), by design of "new copy each time".
• The File and DB extractors should override the hook with their own configured data set in their repos (follow-up).
• Needs Odin Add status codes to time series queue #324 deployed first.

Rollout and rollback

• Release order: Odin Add status codes to time series queue #324, then this. Revert the commit to roll back; files already uploaded remain.

Checklist

• Self-reviewed the diff
• Tests added or updated (or N/A with reason)
• Docs updated (or N/A): module docstring, CHANGELOG
• No secrets, credentials, or PII committed
• Breaking changes called out above and communicated to affected teams: new external-ID format, skipped -> missing status string

[gemini-code-assist]

Code Review

This pull request updates the built-in fetch_logs action to upload a new copy of log files on every invocation instead of overwriting them, utilizing unique, deterministic external IDs. It also introduces the Extractor.get_log_upload_data_set_id() hook to resolve the target data set for log uploads and adds a new completed_with_errors action status when no files are uploaded. The review feedback correctly identifies a potential bug in the external ID generation where a negative room value would slice the integration ID from the end rather than truncating it, and provides a robust fix using max(0, room).

[gemini-code-assist]

If room is negative (for example, if MAX_EXTERNAL_ID_LENGTH is configured to be smaller or if suffix is longer in future modifications), integration_external_id[:room] will slice from the end of the string instead of truncating it. This would keep almost the entire string and could cause the external ID to exceed the limit, leading to upload failures. Ensure room is non-negative by using max(0, room).

Suggested change

	room = MAX_EXTERNAL_ID_LENGTH - len("extractor-logs--") - len(suffix)
	return f"extractor-logs-{integration_external_id[:room]}-{suffix}"
	room = max(0, MAX_EXTERNAL_ID_LENGTH - len("extractor-logs--") - len(suffix))
	return f"extractor-logs-{integration_external_id[:room]}-{suffix}"

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.33%. Comparing base (370f43a) to head (6087f48).

Additional details and impacted files

@@                        Coverage Diff                         @@
##           EDG-1035-action-outcome-status     #592      +/-   ##
==================================================================
+ Coverage                           85.25%   85.33%   +0.08%     
==================================================================
  Files                                  46       46              
  Lines                                4882     4910      +28     
==================================================================
+ Hits                                 4162     4190      +28     
  Misses                                720      720              

Files with missing lines	Coverage Δ
...extractorutils/unstable/core/_log_upload_action.py	99.33% <100.00%> (+0.06%)	⬆️
cognite/extractorutils/unstable/core/actions.py	100.00% <100.00%> (ø)
cognite/extractorutils/unstable/core/base.py	89.53% <100.00%> (+0.22%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.