From f15cb8b37b9b70056bc82a8eb7151f75fad27a71 Mon Sep 17 00:00:00 2001
From: Cody Rose <webmaster@codyro.se>
Date: Wed, 5 Nov 2025 09:24:37 -0500
Subject: [PATCH] Add source type label to chunk size metric (#4543)

We're tracking chunk size - but not by source type, which means we can't use the metric to locate potential chunking anomalies by source type. It's really easy to add a source type label, so I did.
---
 pkg/engine/engine.go  | 2 +-
 pkg/engine/metrics.go | 6 ++++--
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/pkg/engine/engine.go b/pkg/engine/engine.go
index f5de6cb3b..ffc69e91b 100644
--- a/pkg/engine/engine.go
+++ b/pkg/engine/engine.go
@@ -832,7 +832,7 @@ func (e *Engine) scannerWorker(ctx context.Context) {
 
 		dataSize := float64(len(chunk.Data))
 
-		scanBytesPerChunk.Observe(dataSize)
+		scanBytesPerChunk.WithLabelValues(chunk.SourceType.String()).Observe(dataSize)
 		jobBytesScanned.WithLabelValues(
 			chunk.SourceType.String(),
 			chunk.SourceName,
diff --git a/pkg/engine/metrics.go b/pkg/engine/metrics.go
index 8cded174c..212061ca9 100644
--- a/pkg/engine/metrics.go
+++ b/pkg/engine/metrics.go
@@ -55,13 +55,15 @@ var (
 		[]string{"source_type", "source_name"},
 	)
 
-	scanBytesPerChunk = promauto.NewHistogram(prometheus.HistogramOpts{
+	scanBytesPerChunk = promauto.NewHistogramVec(prometheus.HistogramOpts{
 		Namespace: common.MetricsNamespace,
 		Subsystem: common.MetricsSubsystem,
 		Name:      "scan_bytes_per_chunk",
 		Help:      "Total number of bytes in a chunk.",
 		Buckets:   prometheus.ExponentialBuckets(1, 2, 18),
-	})
+	},
+		[]string{"source_type"},
+	)
 
 	jobChunksScanned = promauto.NewCounterVec(prometheus.CounterOpts{
 		Namespace: common.MetricsNamespace,