feat(reconcile): materialize intermediate DataFrames during reconcile#2535
Open
BesikiML wants to merge 5 commits into
Open
feat(reconcile): materialize intermediate DataFrames during reconcile#2535BesikiML wants to merge 5 commits into
BesikiML wants to merge 5 commits into
Conversation
Add write_and_read_df_with_volumes checkpoints for threshold comparison, aggregate per-rule outputs, and mismatch capture to improve backpressure. Includes unit tests verifying persistence is invoked in each path. Fixes #2257
1 task
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #2535 +/- ##
==========================================
+ Coverage 69.10% 69.18% +0.07%
==========================================
Files 105 105
Lines 9482 9503 +21
Branches 1050 1052 +2
==========================================
+ Hits 6553 6575 +22
+ Misses 2735 2731 -4
- Partials 194 197 +3 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Move imports to module scope, mock target.read_data via constructor, and satisfy mypy/pylint for the threshold comparison test.
Remove the private-method threshold test instead of suppressing protected-access; aggregate and capture tests already verify the intermediate persist wiring pattern.
|
✅ 169/169 passed, 9 flaky, 2 skipped, 52m4s total Flaky tests:
Running from acceptance #4946 |
m-abulazm
requested changes
Jun 26, 2026
| source: DataFrame, | ||
| target: DataFrame, | ||
| key_columns: list[str], | ||
| persistence: AbstractReconIntermediatePersist | None = None, |
Contributor
There was a problem hiding this comment.
persistence is never None so no needs to handle that or allow that in the first place
| @@ -0,0 +1,106 @@ | |||
| import tempfile | |||
Contributor
There was a problem hiding this comment.
we dont really need to unit test this. this is a performance optimization addition that testing it goes beyond simple unit tests. we have e2e tests for that.
can you also attempt a test run and post a screenshot of the result
Make persistence a required parameter in capture_mismatch_data_and_columns since callers always provide it. Remove the dedicated unit test file in favor of existing integration/e2e coverage.
m-abulazm
approved these changes
Jun 26, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
write_and_read_df_with_volumescheckpoints in reconcile computation paths that were still missing materialization (marked with TODOs in [FEATURE]: Tighten writes to delta during reconcile #2257).count()/ sampling work.Motivation
Large Spark plans in reconcile can retain long lineages and cause backpressure. The reconcile runtime already materializes some intermediate results (e.g. after source/target join); this PR fills the remaining gaps called out in #2257.
Changes
reconciliation._compute_threshold_comparisonmismatched_dfbeforecount()compare.reconcile_agg_data_per_rulepersistenceand materializemismatch,missing_in_src,missing_in_tgtcompare.capture_mismatch_data_and_columnspersistence; wired fromReconciliation._get_mismatch_dataFixed issue: #2257
Note: These are temporary intermediate writes via
ReconIntermediatePersist(UC volume / parquet), not changes to finalReconCaptureMAIN/METRICS/DETAILS output tables. Temp data is still cleaned up intrigger_recon/trigger_recon_aggregates.Test plan
test_compare.py,test_aggregates_reconcile.py,test_execute.py) — covers aggregate per-rule, mismatch capture, and threshold paths via existing e2e coveragemetadata_config.volume