From c792af3ad0f34309549347835627624a44afe5ba Mon Sep 17 00:00:00 2001
From: lpugoy <44182754+lpugoy@users.noreply.github.com>
Date: Thu, 4 May 2023 22:35:36 +1000
Subject: [PATCH] grafana-mixin: Fix expression for GrafanaRequestsFailing
 alert (#63382)

Fix expression for GrafanaRequestsFailing alert

The intent of the alert is to get the ratio of 5xx to all status codes
[^1]. With the original expression, the left hand side can have more than
one row with the same labels except for the status code. This results in
a promql error because it is doing a many-to-one matching. Doing a sum
on the left hand side first should preserve the intent of the alert and
resolve the issue.

[^1]: https://github.com/grafana/grafana/pull/43116
---
 grafana-mixin/alerts/alerts.libsonnet | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/grafana-mixin/alerts/alerts.libsonnet b/grafana-mixin/alerts/alerts.libsonnet
index fbd957b4fb9..1c92a59bf51 100644
--- a/grafana-mixin/alerts/alerts.libsonnet
+++ b/grafana-mixin/alerts/alerts.libsonnet
@@ -11,8 +11,8 @@
           {
             alert: 'GrafanaRequestsFailing',
             expr: |||
-              100 * namespace_job_handler_statuscode:grafana_http_request_duration_seconds_count:rate5m{handler!~"/api/datasources/proxy/:id.*|/api/ds/query|/api/tsdb/query", status_code=~"5.."}
-              / ignoring (status_code) group_left
+              100 * sum without (status_code) (namespace_job_handler_statuscode:grafana_http_request_duration_seconds_count:rate5m{handler!~"/api/datasources/proxy/:id.*|/api/ds/query|/api/tsdb/query", status_code=~"5.."})
+              /
               sum without (status_code) (namespace_job_handler_statuscode:grafana_http_request_duration_seconds_count:rate5m{handler!~"/api/datasources/proxy/:id.*|/api/ds/query|/api/tsdb/query"})
               > %(grafanaRequestsFailingThresholdPercent)s
             ||| % $._config,