Enabled service to have multiple ports, and the serviceMonitor to mon… #988

tmakamure · 2023-08-25T17:53:53Z

Enabled service to have multiple ports, and the serviceMonitor to monitor multiple endpoints on the service, added relevant configuration examples as comments to values.yaml and updated the README file for the new chart fields.

Issue

#987

Description of changes

Updated the serviceMonitor.yaml template to allow for more service endpoints to be monitored
Updated the service.yaml template to allow for more port to be exposed on the fluent-bit service
Updated the values.yaml file to add the new config keys, and comments on how to configure them
Updated the README file of the chart to add configuration notes for the new config fields

Checklist

Added/modified documentation as required (such as the README.md for modified charts)
Incremented the chart version in Chart.yaml for the modified chart(s)
Manually tested. Describe what testing was done in the testing section below
Make sure the title of the PR is a good description that can go into the release notes

Testing

After making the changes to the chart, I ran helm template . in the chart directory and inspected the output to ensure it is producing the desired effect. An excerpt of the output is as shown below:

---
# Source: aws-for-fluent-bit/templates/service.yaml
apiVersion: v1
kind: Service
metadata:
  labels:
    helm.sh/chart: aws-for-fluent-bit-0.1.28
    app.kubernetes.io/name: aws-for-fluent-bit
    app.kubernetes.io/instance: release-name-aws-for-fluent-bit
    app.kubernetes.io/version: "2.31.11"
    app.kubernetes.io/managed-by: Helm
  name: release-name-aws-for-fluent-bit
  namespace: default
spec:
  ports:
  - name: monitor-agent
    port: 2020
    protocol: TCP
    targetPort: 2020
  - name: metrics
    targetPort: 2021
    protocol: TCP
    port: 2021
  selector:
    app.kubernetes.io/name: aws-for-fluent-bit
    app.kubernetes.io/instance: release-name-aws-for-fluent-bit
  sessionAffinity: None
  type: ClusterIP
---
# Source: aws-for-fluent-bit/templates/servicemonitor.yaml
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: release-name-aws-for-fluent-bit
  namespace: default
spec:
  endpoints:
  - port: monitor-agent
    scheme: http
    interval: 30s
    path: /api/v1/metrics/prometheus
    scrapeTimeout: 10s
  - interval: 10s
    path: /metrics
    port: metrics
    scheme: http
    scrapeTimeout: 10s
  jobLabel: app.kubernetes.io/instance
  namespaceSelector:
    matchNames:
    - default
  selector:
    matchLabels:
      app.kubernetes.io/name: aws-for-fluent-bit
      app.kubernetes.io/instance: release-name-aws-for-fluent-bit

Pre-Requisites to Testing Procedure

Changes in Values

Before testing the chart, the values.yaml file needs to be updated as below:

NB: (Note that we remove the [] on lines 387 and 371 and set the service monitor to enabled

serviceMonitor:
  service: 
    type: ClusterIP
    port: 2020
    targetPort: 2020
    extraPorts: # Removing the empty array placeholder []  because we are now pupulating the array
      - port: 2021
        targetPort: 2021
        protocol: TCP
        name: metrics
  ## When set true then use a ServiceMonitor to configure scraping
  enabled: true # Enabling the servicMonitor
  interval: 30s
  telemetryPath: /api/v1/metrics/prometheus
  labels:
    # app: example-application
    # teamname: example
  timeout: 10s
  relabelings: []
  targetLabels: []
  metricRelabelings: []
  extraEndpoints:  # removing the empty array placeholder [] because now we are populating the array
    - port: metrics
      path: /metrics
      interval: 30s
      scrapeTimeout: 10s
      scheme: http

In addition, the serviceMonitor object is a CRD created by the promehteus community. If your are doing helm template locally using helm 3.x please see below:

Templating locally

Helm does not automatically recognise CRDs, to be able to output the template including CRDs we need to:

Download the serviceMonitor CRD from the prometheus operator Repo: https://github.com/prometheus-operator/prometheus-operator/blob/main/example/prometheus-operator-crd-full/monitoring.coreos.com_servicemonitors.yaml
We add this YAML file to our chart directory either in the root together with Chart.yamland values.yaml or under the templates directory. This will help helm to understand the CRD resource.
Next in the serviceMonitor.yaml template on line two, there is a condition: {{- if and ( .Capabilities.APIVersions.Has "monitoring.coreos.com/v1" ) ( .Values.serviceMonitor.enabled ) }} - which checks if the cluster API version has the "monitoring.coreos.com/v1" capabillity. For a dry run, since we are not installing this on an actual cluster with the relevant API version installed, we may temporarily disable that check by changing the and to an or. (If you are test installing on actual cluster - please see the next heading.)
Once all this is setup - you can then run helm template . and inspect the output for kind: Service and the kind: ServiceMonitor
Revert the condition in serviceMonitor.yaml template back to and AND delete the serviceMonitor CRD file from the chart directory
In values.yaml - comment out lines [372-375] amnd [388 - 392], place back the the empty array place holders [] on lines 371 and 387, and set the enabled: false for the serviceMonitor such that theconfig looks like below:

serviceMonitor:
  service: 
    type: ClusterIP
    port: 2020
    targetPort: 2020
    extraPorts: []
      # - port: 2021
      #   targetPort: 2021
      #   protocol: TCP
      #   name: metrics
  ## When set true then use a ServiceMonitor to configure scraping
  enabled: false
  interval: 30s
  telemetryPath: /api/v1/metrics/prometheus
  labels:
    # app: example-application
    # teamname: example
  timeout: 10s
  relabelings: []
  targetLabels: []
  metricRelabelings: []
  extraEndpoints: []
    # - port: metrics
    #   path: /metrics
    #   interval: 30s
    #   scrapeTimeout: 10s
    #   scheme: http

Installing on test cluster

If you are testing by installing the helm chart on a cluster - you need to ensure that you first install the prometheus-operator CRDs on the cluster - especially the serviceMonitor in our case.

Example using minikube cluster

Download the serviceMonitor CRD from the prometheus operator repo: https://github.com/prometheus-operator/prometheus-operator/blob/main/example/prometheus-operator-crd-full/monitoring.coreos.com_servicemonitors.yaml
Run kubectl apply -f </path/to/serviceMonitor.yaml> to install the CRD on the cluster (this is so that the condition in serviceMonitor.yaml template passes)
In the values.yaml of our chart:
- Enable the serviceMonitor
- Uncomment the lines [372-375] and lines [388-392]
- remove the empty array placeholders [] on lines 387 and 371
The config for service Monitor should now look like:

serviceMonitor:
  service: 
    type: ClusterIP
    port: 2020
    targetPort: 2020
    extraPorts:
      - port: 2021
        targetPort: 2021
        protocol: TCP
        name: metrics
  ## When set true then use a ServiceMonitor to configure scraping
  enabled: true
  interval: 30s
  telemetryPath: /api/v1/metrics/prometheus
  labels:
    # app: example-application
    # teamname: example
  timeout: 10s
  relabelings: []
  targetLabels: []
  metricRelabelings: []
  extraEndpoints: 
    - port: metrics
      path: /metrics
      interval: 30s
      scrapeTimeout: 10s
      scheme: http

Run helm install aws-for-fluentbit . from the chart directory. Helm will install the chart.
To get the service monitor run kubectl get serviceMonitor -n <namespace where chart is deployed> and you will see the service monitor.
Run kubectl describe serviceMonitor <serviceMonitor name from previous output> -n <the namespace where it is installed> and inspect the out put.
To see the yaml output run: kubectl get serviceMonitor <serviceMonitor name> -n <namespace of serviceMonitor> -o yaml and inspect the output
Do the same for the Service resource

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

…itor multiple endpoints, added relevant configuration examples as comments to values.yaml

…es for new chart fields

PettitWesley · 2023-08-25T21:53:04Z

stable/aws-for-fluent-bit/README.md

 | `serviceMonitor.interval`| Set how frequently Prometheus should scrape |`30s`|
 | `serviceMonitor.telemetryPath`| Set path to scrape metrics from |`/api/v1/metrics/prometheus`|
 | `serviceMonitor.labels`| Set labels for the ServiceMonitor, use this to define your scrape label for Prometheus Operator |`[]`|
 | `serviceMonitor.timeout`| Set timeout for scrape |`10s`|
 | `serviceMonitor.relabelings`| Set relabel_configs as per [details](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#relabel_config) |`[]`|
 | `serviceMonitor.targetLabels`| Set of labels to transfer on the Kubernetes Service onto the target. |`[]`|
 | `serviceMonitor.metricRelabelings`| MetricRelabelConfigs to apply to samples before ingestion. |`[]`|
+| `serviceMonitor.extraEndpoints`| Extra endpoints on the fluen-bit service for the serviceMonitor to monitor |`[]`|


the fluen-bit service

Typo

Good spot - fixing that now

PettitWesley

LGTM. Thanks!

PettitWesley · 2023-08-28T22:01:43Z

@tmakamure need to update chart version to get CI to pass:

❌ Chart aws-for-fluent-bit/ has the same Chart version as the last release 0.1.29

PettitWesley · 2023-08-28T22:11:42Z

stable/aws-for-fluent-bit/templates/servicemonitor.yaml

+{{- end }}
+{{- with .Values.serviceMonitor.extraEndpoints }}
+    {{- toYaml . | nindent 2 }}


Is this template correct?

I tried generating it, and it didn't work:

This was in values.yaml

metricRelabelings: [] extraEndpoints: [] - port: metrics path: /metrics interval: 30s scrapeTimeout: 10s scheme: http

And this was the error output from template dry run:

Error: failed to parse stable/aws-for-fluent-bit/values.yaml: error converting YAML to JSON: yaml: line 387: did not find expected key helm.go:84: [debug] error converting YAML to JSON: yaml: line 387: did not find expected key failed to parse stable/aws-for-fluent-bit/values.yaml helm.sh/helm/v3/pkg/cli/values.(*Options).MergeValues helm.sh/helm/v3/pkg/cli/values/options.go:55 main.runInstall helm.sh/helm/v3/cmd/helm/install.go:212 main.newTemplateCmd.func2 helm.sh/helm/v3/cmd/helm/template.go:82 github.com/spf13/cobra.(*Command).execute github.com/spf13/cobra@v1.5.0/command.go:872 github.com/spf13/cobra.(*Command).ExecuteC github.com/spf13/cobra@v1.5.0/command.go:990 github.com/spf13/cobra.(*Command).Execute github.com/spf13/cobra@v1.5.0/command.go:918 main.main helm.sh/helm/v3/cmd/helm/helm.go:83 runtime.main runtime/proc.go:250 runtime.goexit runtime/asm_amd64.s:1594

@PettitWesley I have ammended the test intsrctions see below: the template is correct

@PettitWesley I have also updated the chart version

PettitWesley · 2023-08-28T22:12:25Z

stable/aws-for-fluent-bit/templates/service.yaml

@@ -11,6 +11,14 @@ spec:
    port: {{ .Values.serviceMonitor.service.port }}
    protocol: TCP
    targetPort: {{ .Values.serviceMonitor.service.targetPort }}
+{{- if .Values.serviceMonitor.service.extraPorts }}


I think the template for the extraEndpoints below needs to be similar to this, explicitly model each field.

@PettitWesley the template is correct I double checked now. I have made the ammendmends on the instrustions on how to do the the test run (Should have included that in PR message too):

Changes in Values

Before testing the chart, the values.yaml file needs to be updated as below:

NB: (Note that we remove the [] on lines 387 and 371 and set the service monitor to enabled

serviceMonitor: service: type: ClusterIP port: 2020 targetPort: 2020 extraPorts: # Removing the empty array placeholder [] because we are now pupulating the array - port: 2021 targetPort: 2021 protocol: TCP name: metrics ## When set true then use a ServiceMonitor to configure scraping enabled: true # Enabling the servicMonitor interval: 30s telemetryPath: /api/v1/metrics/prometheus labels: # app: example-application # teamname: example timeout: 10s relabelings: [] targetLabels: [] metricRelabelings: [] extraEndpoints: # removing the empty array placeholder [] because now we are populating the array - port: metrics path: /metrics interval: 30s scrapeTimeout: 10s scheme: http

In addition, the serviceMonitor object is a CRD created by the promehteus community. If your are doing helm template locally using helm 3.x please see below:

Templating locally

Helm does not automatically recognise CRDs, to be able to output the template including CRDs we need to:

Download the serviceMonitor CRD from the prometheus operator Repo: https://github.com/prometheus-operator/prometheus-operator/blob/main/example/prometheus-operator-crd-full/monitoring.coreos.com_servicemonitors.yaml

We add this YAML file to our chart directory either in the root together with Chart.yamland values.yaml or under the templates directory. This will help helm to understand the CRD resource.

Next in the serviceMonitor.yaml template on line two, there is a condition: {{- if and ( .Capabilities.APIVersions.Has "monitoring.coreos.com/v1" ) ( .Values.serviceMonitor.enabled ) }} - which checks if the cluster API version has the "monitoring.coreos.com/v1" capabillity. For a dry run, since we are not installing this on an actual cluster with the relevant API version installed, we may temporarily disable that check by changing the and to an or. (If you are test installing on actual cluster - please see the next heading.)

Once all this is setup - you can then run helm template . and inspect the output for kind: Service and the kind: ServiceMonitor

Revert the condition in serviceMonitor.yaml template back to and AND delete the serviceMonitor CRD file from the chart directory

In values.yaml - comment out lines [372-375] amnd [388 - 392], place back the the empty array place holders [] on lines 371 and 387, and set the enabled: false for the serviceMonitor such that theconfig looks like below:

serviceMonitor: service: type: ClusterIP port: 2020 targetPort: 2020 extraPorts: [] # - port: 2021 # targetPort: 2021 # protocol: TCP # name: metrics ## When set true then use a ServiceMonitor to configure scraping enabled: false interval: 30s telemetryPath: /api/v1/metrics/prometheus labels: # app: example-application # teamname: example timeout: 10s relabelings: [] targetLabels: [] metricRelabelings: [] extraEndpoints: [] # - port: metrics # path: /metrics # interval: 30s # scrapeTimeout: 10s # scheme: http

Installing on test cluster

If you are testing by installing the helm chart on a cluster - you need to ensure that you first install the prometheus-operator CRDs on the cluster - especially the serviceMonitor in our case.

Example using minikube cluster

Download the serviceMonitor CRD from the prometheus operator repo: https://github.com/prometheus-operator/prometheus-operator/blob/main/example/prometheus-operator-crd-full/monitoring.coreos.com_servicemonitors.yaml

Run kubectl apply -f </path/to/serviceMonitor.yaml> to install the CRD on the cluster (this is so that the condition in serviceMonitor.yaml template passes)

In the values.yaml of our chart:

Enable the serviceMonitor

Uncomment the lines [372-375] and lines [388-392]

remove the empty array placeholders [] on lines 387 and 371

The config for service Monitor should now look like:

serviceMonitor: service: type: ClusterIP port: 2020 targetPort: 2020 extraPorts: - port: 2021 targetPort: 2021 protocol: TCP name: metrics ## When set true then use a ServiceMonitor to configure scraping enabled: true interval: 30s telemetryPath: /api/v1/metrics/prometheus labels: # app: example-application # teamname: example timeout: 10s relabelings: [] targetLabels: [] metricRelabelings: [] extraEndpoints: - port: metrics path: /metrics interval: 30s scrapeTimeout: 10s scheme: http

Run helm install aws-for-fluentbit . from the chart directory. Helm will install the chart.

To get the service monitor run kubectl get serviceMonitor -n <namespace where chart is deployed> and you will see the service monitor.

Run kubectl describe serviceMonitor <serviceMonitor name from previous output> -n <the namespace where it is installed> and inspect the out put.

To see the yaml output run: kubectl get serviceMonitor <serviceMonitor name> -n <namespace of serviceMonitor> -o yaml and inspect the output

Do the same for the Service resource

PettitWesley

Thanks!

PettitWesley · 2023-09-01T22:45:01Z

I ran the dry run again without setting any of the new values, which shows the current state of the values.yaml and template won't break anyone who doesn't use these new fields (no regressions).

tmakamure added 3 commits August 25, 2023 19:15

Enabled service to have multiple ports, and the serviceMonitor to mon…

0dcef6f

…itor multiple endpoints, added relevant configuration examples as comments to values.yaml

Updating the README for chart configuration, and setting default valu…

0866bc0

…es for new chart fields

Updating the char version

899c65b

tmakamure requested review from PettitWesley and hossain-rayhan as code owners August 25, 2023 17:53

PettitWesley reviewed Aug 25, 2023

View reviewed changes

PettitWesley previously approved these changes Aug 25, 2023

View reviewed changes

Correcting Typo in README file

8eb3a41

tmakamure dismissed PettitWesley’s stale review via 8eb3a41 August 25, 2023 23:33

tmakamure requested a review from PettitWesley August 28, 2023 17:59

PettitWesley approved these changes Aug 28, 2023

View reviewed changes

PettitWesley suggested changes Aug 28, 2023

View reviewed changes

tmakamure added 2 commits August 29, 2023 09:47

Updating chart version

a3fc79c

Fixing merge conflict on chart version

e1de577

tmakamure requested a review from PettitWesley August 29, 2023 09:14

PettitWesley approved these changes Sep 1, 2023

View reviewed changes

PettitWesley merged commit 463b7a4 into aws:master Sep 1, 2023
1 check passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Enabled service to have multiple ports, and the serviceMonitor to mon… #988

Enabled service to have multiple ports, and the serviceMonitor to mon… #988

tmakamure commented Aug 25, 2023 •

edited

Loading

PettitWesley Aug 25, 2023

tmakamure Aug 25, 2023

PettitWesley left a comment

PettitWesley commented Aug 28, 2023

PettitWesley Aug 28, 2023

tmakamure Aug 29, 2023

tmakamure Aug 29, 2023

PettitWesley Aug 28, 2023

tmakamure Aug 29, 2023

PettitWesley left a comment

PettitWesley commented Sep 1, 2023

Enabled service to have multiple ports, and the serviceMonitor to mon… #988

Enabled service to have multiple ports, and the serviceMonitor to mon… #988

Conversation

tmakamure commented Aug 25, 2023 • edited Loading

Issue

Description of changes

Checklist

Testing

Pre-Requisites to Testing Procedure

Changes in Values

Templating locally

Installing on test cluster

Example using minikube cluster

PettitWesley Aug 25, 2023

Choose a reason for hiding this comment

tmakamure Aug 25, 2023

Choose a reason for hiding this comment

PettitWesley left a comment

Choose a reason for hiding this comment

PettitWesley commented Aug 28, 2023

PettitWesley Aug 28, 2023

Choose a reason for hiding this comment

tmakamure Aug 29, 2023

Choose a reason for hiding this comment

tmakamure Aug 29, 2023

Choose a reason for hiding this comment

PettitWesley Aug 28, 2023

Choose a reason for hiding this comment

tmakamure Aug 29, 2023

Choose a reason for hiding this comment

Changes in Values

Templating locally

Installing on test cluster

Example using minikube cluster

PettitWesley left a comment

Choose a reason for hiding this comment

PettitWesley commented Sep 1, 2023

tmakamure commented Aug 25, 2023 •

edited

Loading