fix dbt retry for microbatch models

[aahel]

Resolves #11423

Problem

When a microbatch model fails during dbt run and dbt retry is executed later (e.g., days after), the retry uses the current date instead of the original failure date to compute which batches to run. This results in processing completely different batches than those that originally failed.

Root cause: In retry.py, the batch_map is only populated when a microbatch model had both successful and failed batches (len(result.batch_results.successful) != 0). When all batches fail (or the model was entirely skipped due to an upstream failure), the model is excluded from batch_map, so previous_batch_results is never set on the node. This causes get_batches() in run.py to fall through to normal batch computation, which uses get_invocation_started_at() — the current retry time, not the original run time.

Example from the issue: A model failed on 2025-03-21 with batches 03-18 through 03-21. When retried on 2025-03-25, it processed batches 03-22 through 03-25 instead — completely missing the original failed batches.

Solution

Track the original invocation time from the run_results.json artifact metadata and use it during batch computation on retry.

Changes:

1. core/dbt/task/retry.py: Extract invocation_started_at from the previous run's metadata and pass it to the RunTask as original_invocation_started_at. Also changed self.task_class == RunTask to issubclass(self.task_class, RunTask) so that BuildTask (which extends RunTask) also gets microbatch retry behavior.

2. core/dbt/task/run.py: Added original_invocation_started_at attribute to RunTask. Updated MicrobatchModelRunner.get_microbatch_builder() to use self.parent_task.original_invocation_started_at as default_end_time when available, falling back to get_invocation_started_at() for normal (non-retry) runs.

3. tests/unit/task/test_run.py: Added two tests verifying that the microbatch builder uses the original invocation time during retry and falls back to the current time during normal runs.

Checklist

• I have read the contributing guide and understand what's expected of me.
• I have run this code in development, and it appears to resolve the stated issue.
• This PR includes tests, or tests are not required or relevant for this PR.
• This PR has no interface changes (e.g., macros, CLI, logs, JSON artifacts, config files, adapter interface, etc.) or this PR has already received feedback and approval from Product or DX.
• This PR includes type annotations for new and modified functions.

[github-actions]

Thank you for your pull request! We could not find a changelog entry for this change. For details on how to document a change, see the contributing guide.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.44%. Comparing base (019ebdc) to head (392e6c5).
⚠️ Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #12525      +/-   ##
==========================================
- Coverage   91.45%   91.44%   -0.02%     
==========================================
  Files         203      203              
  Lines       25471    25497      +26     
==========================================
+ Hits        23294    23315      +21     
- Misses       2177     2182       +5     

Flag	Coverage Δ
integration	88.30% <100.00%> (-0.03%)	⬇️
unit	65.44% <50.00%> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
Unit Tests	65.44% <50.00%> (-0.03%)	⬇️
Integration Tests	88.30% <100.00%> (-0.03%)	⬇️

🚀 New features to boost your workflow:

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

[MichelleArk]

Would also be great to include some functional tests for this change.

Example microbatch functional tests testing retry behavior: https://github.com/dbt-labs/dbt-mantle/blob/main/tests/functional/microbatch/test_microbatch.py#L651-L681

Would be great to get a simple retry test for the build command + a microbatch model given that was introduced in this PR as well!