From 0d9799729b48f03083b69552906316126cd798f8 Mon Sep 17 00:00:00 2001
From: Robert Gildein <robert.gildein@canonical.com>
Date: Thu, 15 Aug 2024 12:25:59 +0200
Subject: [PATCH] Add alert rules to seldon-core based on the KF093 spec

---
 .../KubeflowSeldonCoreServices.rules          | 24 +++++++++++++++++++
 1 file changed, 24 insertions(+)
 create mode 100644 src/prometheus_alert_rules/KubeflowSeldonCoreServices.rules

diff --git a/src/prometheus_alert_rules/KubeflowSeldonCoreServices.rules b/src/prometheus_alert_rules/KubeflowSeldonCoreServices.rules
new file mode 100644
index 0000000..fbcb587
--- /dev/null
+++ b/src/prometheus_alert_rules/KubeflowSeldonCoreServices.rules
@@ -0,0 +1,24 @@
+groups:
+- name: KubeflowSeldonCoreServices
+  rules:
+  - alert: KubeflowServiceDown
+    expr: up{} < 1
+    for: 5m
+    labels:
+      severity: critical
+    annotations:
+      summary: "{{ $labels.juju_charm }} service is Down ({{ $labels.juju_model }}/{{ $labels.juju_unit }})"
+      description: |
+       One or more targets of {{ $labels.juju_charm }} charm are down on unit {{ $labels.juju_model }}/{{ $labels.juju_unit }}.
+       LABELS = {{ $labels }}
+
+  - alert: KubeflowServiceIsNotStable
+    expr: avg_over_time(up{}[10m]) < 0.5
+    for: 0m
+    labels:
+      severity: warning
+    annotations:
+      summary: "{{ $labels.juju_charm }} service is not stable ({{ $labels.juju_model }}/{{ $labels.juju_unit }})"
+      description: |
+        {{ $labels.juju_charm }} unit {{ $labels.juju_model }}/{{ $labels.juju_unit }} has been unreachable at least 50% of the time over the last 10 minutes.
+        LABELS = {{ $labels }}