tools/metrics/histograms/histograms.xml

Index: tools/metrics/histograms/histograms.xml
diff --git a/tools/metrics/histograms/histograms.xml b/tools/metrics/histograms/histograms.xml
index 1043e5631c3993d0e32a18ba3388cdb14f3014aa..80a8be48f8b2211b5dbebeeed102221134ab6af8 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -31248,6 +31248,14 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="Net.HttpJob.PrefilterBytesRead" units="bytes">

[rkaplow, 2016/12/02 20:30:31]

does this need to be bytes, versus like kb for example? Just wondering since the
scale is so large

[jkarlin, 2016/12/05 15:00:25]

It's hard to say. I'd like more accuracy than 1KB. I'm assuming that the vast
majority of resources are less than 50MB. Will adjust and make a v2 metric if
necessary.

+  <owner>jkarlin@chromium.org</owner>
+  <summary>
+    Total prefilter (e.g., before decompression) bytes read for an HttpJob
+    request.
+  </summary>
+</histogram>
+
 <histogram name="Net.HttpJob.TotalTime" units="ms">
   <owner>mmenke@chromium.org</owner>
   <summary>
@@ -108430,6 +108438,12 @@ value.
   <affected-histogram name="HttpCache.PercentBeforeSend"/>
 </histogram_suffixes>
 
+<histogram_suffixes name="HttpJobBytes" separator=".">
+  <suffix name="Cache" label="For requests served from the cache."/>
+  <suffix name="Net" label="For requests served from the network."/>
+  <affected-histogram name="Net.HttpJob.PrefilterBytesRead"/>
+</histogram_suffixes>
+
 <histogram_suffixes name="HttpPipeliningCompatibility">
   <suffix name="disable_test" label="Do nothing"/>
   <suffix name="enable_test" label="Test connection for HTTP pipelining"/>