@@ -280,1 +280,4 @@
 
+## Autoescaling runners with KEDA
+
+Another type of configuration for the runners is to create a ScaledJob on KEDA that polls the job endpoints on your Forgejo instance. When it detects a job to be executed, it spawns a runner with the --one-job flag.

@@ -281,0 +284,4 @@
+
+### Registering runners and binding config with autoscalers
+
+To match job endpoints with job execution, we need to pre-register a runner, save the generated JSON file as a ConfigMap, and share it with the autoscaler. On register select the desired scope (user, repository or global) and create the corresponding JSON file. 

@@ -281,0 +324,4 @@
+
+### Create a token on forgejo
+
+The autoscaler will continuously fetch new waiting jobs from Forgejo, so it requires a properly configured token with the appropriate permissions. Navigate to **Forgejo > User settings > Applications > Access Tokens** and generate a new token. This token will be used by the autoscaler

@@ -281,0 +328,4 @@
+
+### Configuring KEDA Autoscaler
+
+Create a new custom resource with the required parameters and custom fields based on the scope chosen during runner registration. For example, set `global="true"` for global runners or specify `repo=xxx` and `owner=xxx` for repository level runners.

@@ -281,0 +359,4 @@
+        containers:
+        - name: runner
+          image: forgejo:5001/runner:latest
+          command: ["sh", "-c", "stackit-git-runner job"]

@@ -281,0 +352,4 @@
+        volumes:
+        - name: runner-data
+          persistentVolumeClaim:
+            claimName: runner-pvc-claim