From f2b7be89184b8d93da394c7972f5c889567be814 Mon Sep 17 00:00:00 2001
From: "Michael J. Sullivan" <sully@msully.net>
Date: Fri, 14 Feb 2025 16:58:08 -0800
Subject: [PATCH] Report the number of times a feature is used as a metric
 (#8342)

Currently we only report how many branches it is used in total.

Also report how many branches have their schemas introspected
currently.  This is helpful here because we only get feature count
info from introspected branches.
---
 edb/server/compiler/ddl.py   |  4 ++--
 edb/server/dbview/dbview.pyx | 19 +++++++++++++++++--
 edb/server/metrics.py        | 13 ++++++++++++-
 3 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/edb/server/compiler/ddl.py b/edb/server/compiler/ddl.py
index 8c99d2d828a..e9d0924a029 100644
--- a/edb/server/compiler/ddl.py
+++ b/edb/server/compiler/ddl.py
@@ -1247,7 +1247,7 @@ def produce_feature_used_metrics(
     features: dict[str, float] = {}
 
     def _track(key: str) -> None:
-        features[key] = 1
+        features[key] = features.get(key, 0) + 1
 
     # TODO(perf): Should we optimize peeking into the innards directly
     # so we can skip creating the proxies?
@@ -1408,7 +1408,7 @@ def administer_repair_schema(
         user_schema=current_tx.get_user_schema_if_updated(),
         global_schema=current_tx.get_global_schema_if_updated(),
         config_ops=config_ops,
-        feature_used_metrics={},
+        feature_used_metrics=None,
     )
 
 
diff --git a/edb/server/dbview/dbview.pyx b/edb/server/dbview/dbview.pyx
index 49574d6dced..7ee48274392 100644
--- a/edb/server/dbview/dbview.pyx
+++ b/edb/server/dbview/dbview.pyx
@@ -359,12 +359,20 @@ cdef class Database:
         tname = self.tenant.get_instance_name()
         keys = self._feature_used_metrics.keys() | feature_used_metrics.keys()
         for key in keys:
+            # Update the count of how many times the feature is used
             metrics.feature_used.inc(
                 feature_used_metrics.get(key, 0.0)
                 - self._feature_used_metrics.get(key, 0.0),
                 tname,
                 key,
             )
+            # Update the count of branches using the feature at all
+            metrics.feature_used_branches.inc(
+                (feature_used_metrics.get(key, 0.0) > 0)
+                - (self._feature_used_metrics.get(key, 0.0) > 0),
+                tname,
+                key,
+            )
 
         self._feature_used_metrics = feature_used_metrics
 
@@ -1873,9 +1881,16 @@ cdef class DatabaseIndex:
     cdef inline set_current_branches(self):
         metrics.current_branches.set(
             sum(
-                1
+                dbname != defines.EDGEDB_SYSTEM_DB
                 for dbname in self._dbs
-                if dbname != defines.EDGEDB_SYSTEM_DB
+            ),
+            self._tenant.get_instance_name(),
+        )
+        metrics.current_introspected_branches.set(
+            sum(
+                dbname != defines.EDGEDB_SYSTEM_DB
+                and db.user_schema_pickle is not None
+                for dbname, db in self._dbs.items()
             ),
             self._tenant.get_instance_name(),
         )
diff --git a/edb/server/metrics.py b/edb/server/metrics.py
index c76b62378e3..033e31fd6b1 100644
--- a/edb/server/metrics.py
+++ b/edb/server/metrics.py
@@ -47,6 +47,12 @@
     labels=('tenant',),
 )
 
+current_introspected_branches = registry.new_labeled_gauge(
+    'branches_introspected_current',
+    'Current number of branches whose schemas are introspected.',
+    labels=('tenant',),
+)
+
 total_backend_connections = registry.new_labeled_counter(
     'backend_connections_total',
     'Total number of backend connections established.',
@@ -211,11 +217,16 @@
     labels=('tenant', 'extension'),
 )
 
-feature_used = registry.new_labeled_gauge(
+feature_used_branches = registry.new_labeled_gauge(
     'feature_used_branch_count_current',
     'How many branches a schema feature is used by.',
     labels=('tenant', 'feature'),
 )
+feature_used = registry.new_labeled_gauge(
+    'feature_used_num_count_current',
+    'How many times a schema feature is used.',
+    labels=('tenant', 'feature'),
+)
 
 auth_successful_logins = registry.new_labeled_counter(
     "auth_successful_logins_total",