Skip to content

Commit

Permalink
Add storage for Prometheus using a PersistentVolumeClaim
Browse files Browse the repository at this point in the history
Prior to this change, Prometheus' timeseries data would
be lost whenever the chart was upgraded, the Prometheus Pod
was rescheduled, or restarted due to a configuration change.

Tested with a blank storageClassName with KinD, and the local
path provisioner. After creating load with hey, and restarting
Prometheus, the invocation metrics remained available.

The name of the pvc can also be set, however this is not
documented in values.yaml, this may be required for multiple
installations of OpenFaaS within the same cluster for testing
purposes.

Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
  • Loading branch information
alexellis committed Sep 3, 2024
1 parent fe581ba commit a854829
Show file tree
Hide file tree
Showing 4 changed files with 84 additions and 18 deletions.
8 changes: 6 additions & 2 deletions chart/openfaas/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -680,8 +680,7 @@ For legacy scaling in OpenFaaS Community Edition.
| ----------------------- | ---------------------------------- | ---------------------------------------------------------- |
| `alertmanager.create` | Create the AlertManager component | `true` |
| `alertmanager.image` | Container image used for alertmanager | See [values.yaml](./values.yaml) |
| `alertmanager.resources` | Resource limits and requests for alertmanager pods | See [values.yaml](./values.

| `alertmanager.resources` | Resource limits and requests for alertmanager pods | See [values.yaml](./values.yaml) |

### Prometheus (built-in, for autoscaling and metrics)

Expand All @@ -693,3 +692,8 @@ For legacy scaling in OpenFaaS Community Edition.
| `prometheus.retention.size` | The maximum number of bytes of storage blocks to retain. Units supported: B, KB, MB, GB, TB, PB, EB. 0 meaning disabled. See: [Prometheus storage](https://prometheus.io/docs/prometheus/latest/storage/#operational-aspects)| `0` |
| `prometheus.resources` | Resource limits and requests for prometheus containers | See [values.yaml](./values.yaml) |
| `prometheus.recordingRules` | Custom recording rules for autoscaling. | `[]` |
| `prometheus.pvc` | Persistent volume claim for Prometheus used so that metrics survive restarts of the Pod and upgrades of the chart | `{}` |
| `prometheus.pvc.enabled` | Enable persistent volume claim for Prometheus | `false` |
| `prometheus.pvc.storageClassName` | Storage class for Prometheus PVC, set to `""` for the default/standard class to be picked | `""` |
| `prometheus.pvc.size` | Size of the Prometheus PVC, 60-100Gi may be a better fit for a busy production environment | `10Gi` |
| `prometheus.pvc.name` | Name of the Prometheus PVC, required for multiple installations within the same cluster | `""` |
14 changes: 14 additions & 0 deletions chart/openfaas/templates/prometheus-dep.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,8 @@ metadata:
namespace: {{ .Release.Namespace | quote }}
spec:
replicas: 1
strategy:
type: Recreate
selector:
matchLabels:
app: prometheus
Expand Down Expand Up @@ -119,8 +121,20 @@ spec:
path: prometheus-rules.yml
mode: 0644
{{- end }}

{{- if and .Values.prometheus.pvc.enabled .Values.openfaasPro }}
- name: prom-data
persistentVolumeClaim:
{{- if .Values.prometheus.pvc.name }}
claimName: {{.Values.prometheus.pvc.name}}
{{- else }}
claimName: prometheus-data
{{- end }}
{{- else }}
- name: prom-data
emptyDir: {}
{{- end }}

{{- with .Values.nodeSelector }}
nodeSelector:
{{ toYaml . | indent 8 }}
Expand Down
30 changes: 30 additions & 0 deletions chart/openfaas/templates/prometheus-pvc.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,30 @@
{{- $functionNs := default .Release.Namespace .Values.functionNamespace }}
{{- if .Values.openfaasPro }}
{{- if and .Values.prometheus.create .Values.prometheus.pvc.enabled }}
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
labels:
app: {{ template "openfaas.name" . }}
chart: {{ .Chart.Name }}-{{ .Chart.Version }}
component: prometheus
heritage: {{ .Release.Service }}
release: {{ .Release.Name }}
{{- if .Values.prometheus.pvc.name }}
name: {{.Values.prometheus.pvc.name}}
{{- else }}
name: prometheus-data
{{- end }}
namespace: {{ .Release.Namespace | quote }}
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: {{ .Values.prometheus.pvc.size | quote }}
{{- with .Values.prometheus.pvc.storageClassName }}
storageClassName: {{ . | quote }}
{{- end }}
{{- end }}
{{- end }}
50 changes: 34 additions & 16 deletions chart/openfaas/values.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -359,6 +359,11 @@ iam:
url: https://kubernetes.default.svc.cluster.local
tokenExpiry: 2h

## Prometheus is required for metrics and autoscaling
##
## It is bundled into OpenFaaS to be used only as an internal component
## if you wish to retain the metrics for a longer period, you should
## scrape this instance from an external Prometheus server
prometheus:
image: prom/prometheus:v2.54.0
create: true
Expand All @@ -371,22 +376,19 @@ prometheus:
annotations: {}
recordingRules: []

alertmanager:
image: prom/alertmanager:v0.27.0
create: true
resources:
requests:
memory: "25Mi"
cpu: "50m"
limits:
memory: "50Mi"

stan:
# Image used for the NATS Streaming when using the deprecated
# support in the Community Edition (CE)
image: nats-streaming:0.25.6

# NATS (required for async)
# Set to true to enable persistent storage for the Prometheus Pod
# otherwise, the data will be lost when the Pod is restarted
pvc:
enabled: false
# You may want to set this higher for production, or lower for development/staging.
size: 30Gi
# Leave the storageClassName blank for the default storage class
# using the string "default" does not necessarily mean the default
# storage class
storageClassName:

## NATS is used for OpenFaaS Pro and is required for:
## asynchronous invocations, billing & auditing webhooks
nats:
channel: "faas-request"
# Stream replication is set to 1 by default. This is only recommended for development.
Expand All @@ -405,6 +407,22 @@ nats:
memory: "120Mi"
cpu: "100m"

## alertmanager is only used for OpenFaaS CE
alertmanager:
image: prom/alertmanager:v0.27.0
create: true
resources:
requests:
memory: "25Mi"
cpu: "50m"
limits:
memory: "50Mi"

## stan is only used for OpenFaaS CE and will be removed in
## a fture release, having already been deprecated by the NATS team
stan:
image: nats-streaming:0.25.6

# ingress configuration
ingress:
enabled: false
Expand Down

0 comments on commit a854829

Please sign in to comment.