[Close #771] fix: set GOMAXPROCS (#772)

The following patch add automatically the environment variable `GOMAXPROCS`,
when a CPU limit is defined. Otherwise CPU throttling may occur.

Reviewed-on: https://gitea.com/gitea/helm-chart/pulls/772
Reviewed-by: techknowlogick <techknowlogick@noreply.gitea.com>
Reviewed-by: justusbunsi <justusbunsi@noreply.gitea.com>
Co-authored-by: Markus Pesch <markus.pesch@cryptic.systems>
Co-committed-by: Markus Pesch <markus.pesch@cryptic.systems>

README.md

@@ -8,6 +8,7 @@
   - [Dependency Versioning](#dependency-versioning)
 - [Installing](#installing)
 - [High Availability](#high-availability)
+- [Limit resources](#limit-resources)
 - [Configuration](#configuration)
   - [Default Configuration](#default-configuration)
     - [Database defaults](#database-defaults)
@@ -149,6 +150,42 @@ Care must be taken for production use as not all implementation details of Gitea
 Deploying a HA-ready Gitea instance requires some effort including using HA-ready dependencies.
 See the [HA Setup](docs/ha-setup.md) document for more details.
 
+## Limit resources
+
+If the application is deployed with a CPU resource limit, Prometheus may throw a CPU throttling warning for the
+application. This has more or less to do with the fact that the application finds the number of CPUs of the host, but
+cannot use the available CPU time to perform computing operations.
+
+The application must be informed that despite several CPUs only a part (limit) of the available computing time is
+available. As this is a Golang application, this can be implemented using `GOMAXPROCS`. The following example is one way
+of defining `GOMAXPROCS` automatically based on the defined CPU limit like `100m`. Please keep in mind, that the CFS
+rate of `100ms` - default on each kubernetes node, is also very important to avoid CPU throttling.
+
+Further information about this topic can be found [here](https://kanishk.io/posts/cpu-throttling-in-containerized-go-apps/).
+
+> [!NOTE]
+> The environment variable `GOMAXPROCS` is set automatically, when a CPU limit is defined. An explicit configuration is
+> not anymore required.
+
+```yaml
+deployment:
+  env:
+  # Will be automatically defined!
+  - name: GOMAXPROCS
+    valueFrom:
+      resourceFieldRef:
+        divisor: "1" # Is required for GitDevOps systems like ArgoCD/Flux. Otherwise throw the system a diff error. (k8s-default=1)
+        resource: limits.cpu
+
+resources:
+  limits:
+    cpu: 100m
+    memory: 512Mi
+  requests:
+    cpu: 100m
+    memory: 512Mi
+```
+
 ## Configuration
 
 Gitea offers lots of configuration options.

templates/gitea/deployment.yaml

@@ -291,6 +291,13 @@ spec:
               value: /data
             - name: GITEA_TEMP
               value: /tmp/gitea
+            {{- if and (hasKey .Values.resources "limits") (hasKey .Values.resources.limits "cpu") }}
+            - name: GOMAXPROCS
+              valueFrom:
+                resourceFieldRef:
+                  divisor: "1"
+                  resource: limits.cpu
+            {{- end }}
             - name: TMPDIR
               value: /tmp/gitea
             {{- if .Values.image.rootless }}

unittests/helm/deployment/basic.yaml

@@ -45,3 +45,31 @@ tests:
           content:
             name: ENV_TO_INI_MOUNT_POINT
             value: /env-to-ini-mounts
+  - it: CPU resources are defined as well as GOMAXPROCS
+    template: templates/gitea/deployment.yaml
+    set:
+      resources:
+        limits:
+          cpu: 200ms
+          memory: 200Mi
+        requests:
+          cpu: 100ms
+          memory: 100Mi
+    asserts:
+      - contains:
+          path: spec.template.spec.containers[0].env
+          content:
+            name: GOMAXPROCS
+            valueFrom:
+              resourceFieldRef:
+                divisor: "1"
+                resource: limits.cpu
+      - equal:
+          path: spec.template.spec.containers[0].resources
+          value:
+            limits:
+              cpu: 200ms
+              memory: 200Mi
+            requests:
+              cpu: 100ms
+              memory: 100Mi