From e9d60394bfb4e3f65d36cddf796b1394d29503a4 Mon Sep 17 00:00:00 2001
From: Gregory Mierzwinski <gmierz1@live.ca>
Date: Thu, 23 Apr 2026 21:21:40 -0400
Subject: [PATCH] Bug 2034671 - Ingest suite-level replicates from performance
 data.

---
 tests/etl/test_perf_data_load.py | 19 +++++++++++++++++++
 treeherder/etl/perf.py           | 17 +++++++++++++++++
 2 files changed, 36 insertions(+)

diff --git a/tests/etl/test_perf_data_load.py b/tests/etl/test_perf_data_load.py
index 11259cde567..9e7d3e61b24 100644
--- a/tests/etl/test_perf_data_load.py
+++ b/tests/etl/test_perf_data_load.py
@@ -16,6 +16,7 @@
 from treeherder.perf.models import (
     MultiCommitDatum,
     PerformanceDatum,
+    PerformanceDatumReplicate,
     PerformanceFramework,
     PerformanceSignature,
 )
@@ -523,3 +524,21 @@ def test_multi_commit_data_is_removed_by_dedicated_management_script(
     assert (
         PerformanceDatum.objects.all().count() == DATA_PER_ARTIFACT
     )  # data ingested in the old way remains intact
+
+
+def test_suite_replicates_are_ingested(test_repository, perf_job, sample_perf_artifact):
+    suite_replicates = [1.0, 2.0, 3.0]
+    sample_perf_artifact["blob"]["suites"][0]["replicates"] = suite_replicates
+    _, submit_datum = _prepare_test_data(sample_perf_artifact)
+
+    store_performance_artifact(perf_job, submit_datum)
+
+    suite_datum = PerformanceDatum.objects.get(
+        signature=PerformanceSignature.objects.get(suite="youtube-watch", test="")
+    )
+    stored_replicates = list(
+        PerformanceDatumReplicate.objects.filter(performance_datum=suite_datum).values_list(
+            "value", flat=True
+        )
+    )
+    assert sorted(stored_replicates) == sorted(suite_replicates)
diff --git a/treeherder/etl/perf.py b/treeherder/etl/perf.py
index a1d30ff35e9..bf7c88dd52f 100644
--- a/treeherder/etl/perf.py
+++ b/treeherder/etl/perf.py
@@ -227,6 +227,23 @@ def _load_perf_datum(job: Job, perf_datum: dict):
                 # keep a register with all multi commit perf data
                 MultiCommitDatum.objects.create(perf_datum=suite_datum)
 
+            replicates = suite.get("replicates", [])
+            if replicates and len(replicates) > 0:
+                try:
+                    # Add the replicates to the PerformanceDatumReplicate table, and
+                    # catch and ignore any exceptions that are produced here so we don't
+                    # impact the standard workflow
+                    PerformanceDatumReplicate.objects.bulk_create(
+                        [
+                            PerformanceDatumReplicate(
+                                value=replicate, performance_datum=suite_datum
+                            )
+                            for replicate in replicates
+                        ]
+                    )
+                except Exception as e:
+                    logger.info(f"Failed to ingest replicates for suite datum {suite_datum}: {e}")
+
             if _suite_should_alert_based_on(signature, job, datum_created):
                 generate_alerts.apply_async(args=[signature.id], queue="generate_perf_alerts")