From: Tom Hughes <tom@compton.nu>
Date: Wed, 3 May 2023 07:31:29 +0000 (+0100)
Subject: Add an alert for postgresql transactions which have been idle too long
X-Git-Url: https://git.openstreetmap.org./chef.git/commitdiff_plain/8db5669627a5b7a14872a452722e7cf91e16c0a8

Add an alert for postgresql transactions which have been idle too long
---

diff --git a/cookbooks/prometheus/templates/default/alert_rules.yml.erb b/cookbooks/prometheus/templates/default/alert_rules.yml.erb
index 5474e7e38..2a7fd1b09 100644
--- a/cookbooks/prometheus/templates/default/alert_rules.yml.erb
+++ b/cookbooks/prometheus/templates/default/alert_rules.yml.erb
@@ -525,6 +525,13 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           queries: "{{ $value }}"
+      - alert: postgresql idle transactions
+        expr: sum(pg_process_idle_seconds_count{state="idle in transaction"}) by (instance, server) > sum(pg_process_idle_seconds_bucket{state="idle in transaction",le="120"}) by (instance, server)
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+        annotations:
+          queries: "{{ $value }}"
   - name: prometheus
     rules:
       - alert: prometheus configuration error