feat(dynamic-sampling): per-org transaction rebalancing by constantinius · Pull Request #116475 · getsentry/sentry

constantinius · 2026-05-29T13:47:39Z

Add calculations for transaction rebalancing to the per-org pipeline
Compare calculation outputs with the legacy pipeline and log results

Closes TET-2403

… rebalancing

linear-code · 2026-05-29T13:47:45Z

shellmayr · 2026-05-29T14:16:56Z

        compare_rebalanced_projects_with_cache(config, rebalanced_projects, cached_sample_rates)

-    if not get_eap_transaction_volumes(config):
+    transaction_volumes = get_eap_transaction_volumes(config)


The default interval on this query is 5 minutes, the one in the legacy pipeline is 1 hour (here vs here) - if we are never going to use the 5 minutes, we should probably change the default.

shellmayr · 2026-05-29T14:23:10Z

+) -> dict[int, tuple[list[RebalancedItem], float]]:
+    intensity = options.get("dynamic-sampling.prioritise_transactions.rebalance_intensity")
+    sample_rates = _project_sample_rates(config)
+    return {


Are we not clamping the sample rates yet? If no, should we record information about it?

Co-authored-by: Simon Hellmayr <shellmayr@users.noreply.github.com>

…-rebalancing

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit c30e3ed. Configure here.}

…anced projects ones

…ple_rates

sentry · 2026-05-29T16:44:16Z

+            TransactionsRebalancingInput(
+                classes=[
+                    RebalancedItem(id=transaction_name, count=count)
+                    for transaction_name, count in project_data["transaction_counts"]
+                ],
+                sample_rate=sample_rate,
+                total_num_classes=project_data.get("total_num_classes"),
+                total=project_data.get("total_num_transactions"),
+                intensity=intensity,
+            )
+        )


Bug: The call to TransactionsRebalancingModel().run() lacks error handling. An empty transaction_counts list will raise an unhandled InvalidModelInputError, crashing the background task.
_{Severity: MEDIUM}

Suggested Fix

Before calling the model, add a guard to check if transaction_counts is empty and return early if it is. This mimics the safe handling present in the legacy code. Alternatively, wrap the call to TransactionsRebalancingModel().run() in a try...except InvalidModelInputError block to handle the validation failure gracefully without crashing the task.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI agent. Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not valid. Location: src/sentry/dynamic_sampling/per_org/calculations.py#L131-L142 Potential issue: In `run_transaction_balancing`, the call to `TransactionsRebalancingModel().run()` is not wrapped in any error handling. The model's `run` method will raise an `InvalidModelInputError` if its input validation fails, which occurs if the `transaction_counts` list is empty. This is a realistic edge case for projects that have no transactions passing the volume filter. The unhandled exception propagates up, causing the `run_calculations_per_org_task` to fail and preventing dynamic sampling calculations from completing for that organization.

feat(dynamic-sampling): initial implementation of per-org transaction…

df2acee

… rebalancing

constantinius requested a review from a team as a code owner May 29, 2026 13:47

github-actions Bot added the Scope: Backend Automatically applied to PRs that change backend components label May 29, 2026

constantinius requested a review from shellmayr May 29, 2026 13:48

sentry-warden Bot reviewed May 29, 2026

View reviewed changes

Comment thread src/sentry/dynamic_sampling/per_org/calculations.py

cursor Bot reviewed May 29, 2026

View reviewed changes