From f484633f2dcbac3b85e90ea6fcbc417b7a41b9f5 Mon Sep 17 00:00:00 2001
From: Gabriel Fontes <hi@m7.rs>
Date: Fri, 28 Apr 2023 11:03:26 -0300
Subject: [PATCH] sitespeedio: cleanup aggregation metrics

---
 hosts/alcyone/services/prometheus/default.nix | 53 ++++++++++++++-----
 hosts/celaeno/services/sitespeedio.nix        | 23 +++-----
 2 files changed, 45 insertions(+), 31 deletions(-)

diff --git a/hosts/alcyone/services/prometheus/default.nix b/hosts/alcyone/services/prometheus/default.nix
index d71fd70b..ea8f6211 100644
--- a/hosts/alcyone/services/prometheus/default.nix
+++ b/hosts/alcyone/services/prometheus/default.nix
@@ -1,22 +1,46 @@
 { config, ... }:
-let
-  mkStatic = job_name: targets: {
-    inherit job_name;
-    scheme = "https";
-    static_configs = [{ inherit targets; }];
-    metrics_path = "/metrics";
-  };
-in {
+{
   services = {
     prometheus = {
       enable = true;
+      globalConfig = {
+        # Scrape a bit more frequently
+        scrape_interval = "30s";
+      };
       scrapeConfigs = [
-        (mkStatic "hydra" ["hydra.m7.rs"])
-        (mkStatic "sitespeed" ["sitespeed.m7.rs"])
-        (mkStatic "nginx" ["alcyone.m7.rs" "celaeno.m7.rs" "merope.m7.rs"])
+        {
+          job_name = "hydra";
+          static_configs = [{
+            targets = [ "hydra.m7.rs" ];
+          }];
+        }
+        {
+          job_name = "nginx";
+          static_configs =[{
+            targets = [ "alcyone.m7.rs" "celaeno.m7.rs" "merope.m7.rs" ];
+          }];
+        }
+        {
+          job_name = "sitespeed";
+          static_configs = [{
+            targets = [ "sitespeed.m7.rs" ];
+          }];
+          metric_relabel_configs = [
+            # Only keep metrics that are not aggregations or are medians
+            {
+              source_labels = [ "aggr_kind" ];
+              regex = "(median|)";
+              action = "keep";
+            }
+            # Then remove the aggregation label
+            {
+              regex = "aggr_kind";
+              action = "labeldrop";
+            }
+          ];
+        }
       ];
-      extraFlags = let
-        prometheus = config.services.prometheus.package;
+      extraFlags = let prometheus = config.services.prometheus.package;
       in [
         # Custom consoles
         "--web.console.templates=${prometheus}/etc/prometheus/consoles"
@@ -27,7 +51,8 @@ in {
       "metrics.m7.rs" = {
         forceSSL = true;
         enableACME = true;
-        locations."/".proxyPass = "http://localhost:${toString config.services.prometheus.port}";
+        locations."/".proxyPass =
+          "http://localhost:${toString config.services.prometheus.port}";
       };
     };
   };
diff --git a/hosts/celaeno/services/sitespeedio.nix b/hosts/celaeno/services/sitespeedio.nix
index 85b99a07..fe89d1b3 100644
--- a/hosts/celaeno/services/sitespeedio.nix
+++ b/hosts/celaeno/services/sitespeedio.nix
@@ -18,33 +18,22 @@
 
     prometheus.exporters.graphite = {
       enable = true;
+      extraFlags = [ "--graphite.mapping-strict-match" ]; # Drop non-matched metrics
       mappingSettings.mappings = [
         # Page summaries
         {
-          match = ''sitespeed_io\.([^.]+)\.([^.]+)\.pageSummary\.([^.]+)\.([^.]+)\.([^.]+)\.([^.]+)\.(.*)'';
+          match = ''^sitespeed_io\.([^.]+)\.([^.]+)\.pageSummary\.([^.]+)\.([^.]+)\.([^.]+)\.([^.]+)\.(.+?)(?:\.(mean|max|median|min|p90|p99|p10|rsd|mdev|stddev|total|values))?$'';
           match_type = "regex";
           name = "sitespeedio_$7";
           labels = {
-            sitespeed_profile = "$1";
-            sitespeed_job = "$2";
+            profile = "$1";
+            site = "$2";
             domain = "$3";
             page = "$4";
             browser = "$5";
             platform = "$6";
-          };
-        }
-        # Site summaries
-        {
-          match = ''sitespeed_io\.([^.]+)\.([^.]+)\.summary\.([^.]+)\.([^.]+)\.([^.]+)\.(.*)'';
-          match_type = "regex";
-          name = "sitespeedio_$6";
-          labels = {
-            sitespeed_profile = "$1";
-            sitespeed_job = "$2";
-            domain = "$3";
-            page = "*";
-            browser = "$4";
-            platform = "$5";
+            # We save this so that we can drop specific aggregate metrics
+            aggr_kind = "$8";
           };
         }
       ];