From 916c4c5527e423345e029b6365326f79a1e1c553 Mon Sep 17 00:00:00 2001
From: Paul D'Ambra <paul@posthog.com>
Date: Sat, 22 Jun 2024 14:15:05 +0100
Subject: [PATCH] chore: add even more replay ingestion metrics to the hourly
 task (#23169)

---
 .run/PostHog.run.xml   |  3 +--
 posthog/tasks/tasks.py | 38 +++++++++++++++++++++++++++-----------
 2 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/.run/PostHog.run.xml b/.run/PostHog.run.xml
index 54d34fc6443..77bf3859b14 100644
--- a/.run/PostHog.run.xml
+++ b/.run/PostHog.run.xml
@@ -22,9 +22,8 @@
       <env name="SKIP_SERVICE_VERSION_REQUIREMENTS" value="1" />
     </envs>
     <option name="SDK_HOME" value="$PROJECT_DIR$/env/bin/python" />
-    <option name="SDK_NAME" value="Python 3.11 (posthog)" />
     <option name="WORKING_DIRECTORY" value="" />
-    <option name="IS_MODULE_SDK" value="false" />
+    <option name="IS_MODULE_SDK" value="true" />
     <option name="ADD_CONTENT_ROOTS" value="true" />
     <option name="ADD_SOURCE_ROOTS" value="true" />
     <EXTENSION ID="net.ashald.envfile">
diff --git a/posthog/tasks/tasks.py b/posthog/tasks/tasks.py
index 0964e001717..b609226e05c 100644
--- a/posthog/tasks/tasks.py
+++ b/posthog/tasks/tasks.py
@@ -236,14 +236,16 @@ def invalid_web_replays() -> None:
     query = """
     select
         --team_id,
-        count()
+        count() as all_recordings,
+        countIf(snapshot_source == 'mobile') as mobile_recordings,
+        countIf(snapshot_source == 'web') as web_recordings,
+        countIf(snapshot_source =='web' and first_url is null) as invalid_web_recordings
     from (
         select any(team_id) as team_id, argMinMerge(first_url) as first_url, argMinMerge(snapshot_source) as snapshot_source
         from session_replay_events
-        where min_first_timestamp >= now() - interval 1 hour
-        and min_first_timestamp <= now()
+        where min_first_timestamp >= now() - interval 65 minute
+        and min_first_timestamp <= now() - interval 5 minute
         group by session_id
-        having first_url is null and snapshot_source = 'web'
     )
     --group by team_id
     """
@@ -252,14 +254,28 @@ def invalid_web_replays() -> None:
         results = sync_execute(
             query,
         )
+
+        metrics = [
+            "all_recordings",
+            "mobile_recordings",
+            "web_recordings",
+            "invalid_web_recordings",
+        ]
+        descriptions = [
+            "All recordings that started in the last hour",
+            "Recordings started in the last hour that are from mobile",
+            "Recordings started in the last hour that are from web",
+            "Acts as a proxy for replay sessions which haven't received a full snapshot",
+        ]
         with pushed_metrics_registry("celery_replay_tracking") as registry:
-            gauge = Gauge(
-                "replay_tracking_web_replay_with_missing_first_url",
-                "Acts as a proxy for replay sessions which haven't received a full snapshot",
-                registry=registry,
-            )
-            count = results[0][0]
-            gauge.set(count)
+            for i in range(0, 4):
+                gauge = Gauge(
+                    f"replay_tracking_{metrics[i]}",
+                    descriptions[i],
+                    registry=registry,
+                )
+                count = results[0][i]
+                gauge.set(count)
     except:
         pass