https://github.com/coder/observability
https://github.com/coder/observability
Last synced: 15 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/coder/observability
- Owner: coder
- License: cc0-1.0
- Created: 2024-04-07T13:24:04.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-01-16T08:26:41.000Z (3 months ago)
- Last Synced: 2025-04-02T22:35:29.887Z (23 days ago)
- Language: Go
- Size: 564 KB
- Stars: 16
- Watchers: 2
- Forks: 4
- Open Issues: 17
-
Metadata Files:
- Readme: README.gotmpl
- Changelog: CHANGELOG.md
- License: LICENSE
Awesome Lists containing this project
README
# Coder Observability Chart
> [!NOTE]
> This Helm chart is in BETA; use with caution
## Overview
This chart contains a highly opinionated set of integrations between Grafana, Loki, Prometheus, Alertmanager, and
Grafana Agent.
Dashboards, alerts, and runbooks are preconfigured for monitoring [Coder](https://coder.com/) installations.
Out of the box:
Metrics will be scraped from all pods which have a `prometheus.io/scrape=true` annotation.
Logs will be scraped from all pods in the Kubernetes cluster.
## Installation
```bash
helm repo add coder-observability https://helm.coder.com/observability
helm upgrade --install coder-observability coder-observability/coder-observability --version 0.1.1 --namespace coder-observability --create-namespace
```
## Requirements
### General
- Helm 3.7+
### Coder
Kubernetes-based deployments
If your installation is not in a namespace named `coder`, you will need to modify:
```yaml
global:
coder:
controlPlaneNamespace:
externalProvisionersNamespace:
```
Non-Kubernetes deployments (click to expand)
Ensure your Coder installation is accessible to the resources created by this chart.
Set `global.coder.scrapeMetrics` such that the metrics can be scraped from your installation, e.g.:
```yaml
global:
coder:
scrapeMetrics:
hostname: your.coder.host
port: 2112
scrapeInterval: 15s
additionalLabels:
job: coder
```
If you would like your logs scraped from a process outside Kubernetes, you need to mount the log file(s) in and
configure Grafana Agent to scrape them; here's an example configuration:
```yaml
grafana-agent:
agent:
mounts:
extra:
- mountPath: /var/log
name: logs
readOnly: true
controller:
volumes:
extra:
- hostPath:
path: /var/log
name: logs
extraBlocks: |-
loki.source.file "coder_log" {
targets = [
{__path__ = "/var/log/coder.log", job="coder"},
]
forward_to = [loki.write.loki.receiver]
}
```
Ensure these environment variables are set in your Coder deployment:
- `CODER_PROMETHEUS_ENABLE=true`
- `CODER_PROMETHEUS_COLLECT_AGENT_STATS=true`
- `CODER_LOGGING_HUMAN=/dev/stderr` (only `human` log format is supported
currently; [issue](https://github.com/coder/observability/issues/8))
Ensure these labels exist on your Coder & provisioner deployments:
- `prometheus.io/scrape=true`
- `prometheus.io/port=2112` (ensure this matches the port defined by `CODER_PROMETHEUS_ADDRESS`)
If you use the [`coder/coder` helm chart](https://github.com/coder/coder/tree/main/helm), you can use the
following:
```yaml
coder:
podAnnotations:
prometheus.io/scrape: "true"
prometheus.io/port: "2112"
```
For more details, see
the [coder documentation on exposing Prometheus metrics](https://coder.com/docs/v2/latest/admin/prometheus).
### Postgres
You may configure the Helm chart to monitor your Coder deployment's Postgres server. Ensure that the resources created
by this Helm chart can access your Postgres server.
Create a secret with your Postgres password and reference it as follows, along with the other connection details:
```yaml
global:
postgres:
hostname:
port:
database:
username:
mountSecret:
```
The secret should be in the form of `PGPASSWORD=`, as this secret will be used to create an environment
variable.
```yaml
apiVersion: v1
kind: Secret
metadata:
name: pg-secret
namespace: coder-observability
data:
PGPASSWORD:
```
Postgres metrics (click to expand)
A tool called [`postgres-exporter`](https://github.com/prometheus-community/postgres_exporter) is used to scrape metrics
from your Postgres server, and you can see the metrics it is exposing as follows:
```bash
kubectl -n coder-observability port-forward statefulset/postgres-exporter 9187
curl http://localhost:9187/metrics
```
### Grafana
To access Grafana, run:
```bash
kubectl -n coder-observability port-forward svc/grafana 3000:80
```
And open your web browser to http://localhost:3000/.
By default, Grafana is configured to allow anonymous access; if you want password authentication, define this in
your `values.yaml`:
```yaml
grafana:
admin:
existingSecret: grafana-admin
userKey: username
passwordKey: password
grafana.ini:
auth.anonymous:
enabled: false
```
You will also need to define a secret as follows:
```yaml
apiVersion: v1
kind: Secret
metadata:
name: grafana-admin # this matches the "existingSecret" field above
stringData:
username: "" # this matches the "userKey" field above
password: "" # this matches the "passwordKey" field above
```
## Subcharts
{{ template "chart.requirementsTable" . }}
Each subchart can be disabled by setting the `enabled` field to `false`.
| Subchart | Setting |
|-----------------|-------------------------|
| `grafana` | `grafana.enabled` |
| `grafana-agent` | `grafana-agent.enabled` |
| `loki` | `loki.enabled` |
| `prometheus` | `prometheus.enabled` |
## Values
The `global` values are the values which pertain to this chart, while the rest pertain to the subcharts.
These values represent only the values _set_ in this chart. For the full list of available values, please see each
subchart.
For example, the `grafana.replicas` value is set by this chart by default, and is one of hundreds of available
values which are defined [here](https://github.com/grafana/helm-charts/tree/main/charts/grafana#configuration).
{{ template "chart.valuesTable" . }}
{{ template "helm-docs.versionFooter" . }}