Easy things should be easy. Adding monitoring to your application has never been as easy as now.
Does your application exposes prometheus metrics? then adding telegraf.influxdata.com/port: "8080" annotation to the pod is the only thing you need to add telegraf scraping to it
No one likes monitoring/observability, everybody wants to deploy applications but the burden of adding monitoring, fixing it, maintaining it should not weight that much.
Releasing docker images at: Quay
An up to date version of telegraf-operator can be installed by using the InfluxData Helm Repository.
Simply run:
helm repo add influxdata https://helm.influxdata.com/
helm upgrade --install telegraf-operator influxdata/telegraf-operatorTo change one or more settings, please use the --set option - such as:
helm upgrade --install telegraf-operator influxdata/telegraf-operator \
--set certManager.enable=trueThe certManager.enable setting will use cert-manager CRDs to generate TLS certificates for the webhook admission controller used by telegraf-operator. Please note that this requires cert-manager to be installed in the cluster to work.
It is recommended to use a values file instead of setting name-values.
It's also recommended to configure the classes.data values, which specify the telegraf-operator classes and how gathered data should be stored or persisted. Classes are described in more details in Global configuration - classes section.
For example:
classes:
data:
default: |
[[outputs.file]]
files = ["stdout"]This will cause telegraf for default class of monitored workloads to write their data to standard output of the telegraf container.
All of the available settings can be found in the values.yaml file bundled with the Helm chart.
Information about the Helm chart can also be found at https://artifacthub.io/packages/helm/influxdata/telegraf-operator.
An up to date version of telegraf-operator is also available from OperatorHub.io.
Please follow instructions at https://operatorhub.io/operator/telegraf-operator for installing telegraf-operator.
In order for telegraf-operator to monitor a workload, one or more annotations need to be added to the pod. The telegraf.influxdata.com/class annotation specifies which class of workload it is. It also needs information on how to scrape data. For prometheus metrics the annotation is telegraf.influxdata.com/ports, which specifies port or ports to scrape at. The default path is /metrics and can be changed.
By default telegraf-operator comes with an example default class configured to write to an in-cluster instance of InfluxDB.
For Deployment, StatefulSet and most other Kubernetes objects, this should be added to .spec.template.metadata.annotations section - such as:
apiVersion: apps/v1
kind: Deployment
# ...
spec:
# ...
template:
metadata:
annotations:
telegraf.influxdata.com/class: "default"
telegraf.influxdata.com/ports: "8080"
spec:
# ...Please see Pod-level annotations for more details on all annotations telegraf-operator supports.
For development purposes, the repository provides a development version that can be installed by running:
kubectl apply -f https://raw.githubusercontent.com/influxdata/telegraf-operator/master/deploy/dev.yml The command above deploys telegraf-operator, using a separate telegraf-operator namespace and registering webhooks that will inject a telegraf sidecar to all newly created pods.
In order to use telegraf-operator, what's also needed is to define where metrics should be sent.
The examples/classes.yml file provides a set of classes that can be used to get started.
To create sample set of classes, simply run:
kubectl apply -f https://raw.githubusercontent.com/influxdata/telegraf-operator/master/examples/classes.ymlIn order to see the data, you can also deploy InfluxDB v1 in your cluster, which also comes with Chronograf, providing a web UI for InfluxDB v1.
To set it up in your cluster, simply run:
kubectl apply -f https://raw.githubusercontent.com/influxdata/telegraf-operator/master/deploy/influxdb.yml After that, every new pod (created directly or by creating a deployment or statefulset) in your cluster will have include telegraf container for retrieving data.
You can try it by running one of our samples - such as a redis server. Simply do:
kubectl apply -f https://raw.githubusercontent.com/influxdata/telegraf-operator/master/examples/redis.ymlYou can verify the telegraf container is present by doing:
kubectl describe pod -n redis redis-0The output should include a telegraf container.
In order to see the results in InfluxDB and Chronograf, you will need to set up port-forwarding and then access Chronograf from your browser:
kubectl port-forward --namespace=influxdb svc/influxdb 8888:8888Next, go to http://localhost:8888 and continue to Explore section to see your data
Telegraf-operator consists of the following:
- Global configuration - definition of where the metrics should be sent and other auxiliary configuration, specified as classes
- Pod-level configuration - definition of how a pod can be monitored, such as ports for Prometheus scraping and additional configurations
Telegraf-operator is based on concepts of globally defined classes. Each class is a subset of Telegraf configuration and usually defines where Telegraf should be sending its outputs, along with other settings such as global tags.
Usually classes are defined as a secret - such as in classes.yml file - and each class maps to a key in a secret. For example:
stringData:
basic: |+
[[outputs.influxdb]]
urls = ["http://influxdb.influxdb:8086"]
[[outputs.file]]
files = ["stdout"]
[global_tags]
hostname = "$HOSTNAME"
nodename = "$NODENAME"
type = "app"
The above defines that any pod whose Telegraf class is basic will have its metrics sent to a specific URL, which in this case is an InfluxDB v1 instance deployed in same cluster. Its metrics will also be logged by telegraf container for convenience. The data will also have hostname, nodename and type tags added for all metrics.
As of version 1.3.0, telegraf-operator supports detecting when the classes configuration has changed and update telegraf configuration for affected pods.
This functionality requires telegraf version 1.19, which is the first version that supports the new --watch-config option required for this feature.
The development deployment example has hot reload enabled. For Helm chart, version 1.3.0 or newer has to be used and hotReload should be set to true. It is set to false by default to avoid issues when using a version of telegraf prior to 1.19.0.
If deploying telegraf-operator in a different way, telegraf-operator should be run with --telegraf-watch-config=inotify option. The args section of the telegraf-operator Deployment should be added or modified and include the said options - such as:
args:
- --enable-default-internal-plugin=true
- --telegraf-default-class=basic
- --telegraf-classes-directory=/config/classes
- --enable-istio-injection=true
Each pod (either standalone or as part of deployment as well as statefulset) may also specify how it should be monitored using metadata.
The redis.yml example adds annotation that enables the Redis plugin so that Telegraf will automatically retrieve metrics related to it.
apiVersion: apps/v1
kind: StatefulSet
# ...
spec:
template:
metadata:
annotations:
telegraf.influxdata.com/inputs: |+
[[inputs.redis]]
servers = ["tcp://localhost:6379"]
telegraf.influxdata.com/class: basic
# ...
spec:
containers:
- name: redis
image: redis:alpine
Please see redis input plugin documentation for more details on how the plugin can be configured.
The telegraf.influxdata.com/class specifies that the basic class above should be used.
Users can configure the inputs.prometheus plugin by setting the following annotations. Below is an example configuration, and the expected output.
telegraf.influxdata.com/port: is used to configure which port telegraf should scrapetelegraf.influxdata.com/ports: is used to configure which port telegraf should scrape, comma separated list of ports to scrapetelegraf.influxdata.com/path: is used to configure at which path to configure scraping to (a port must be configured also), will apply to all ports if multiple are configuredtelegraf.influxdata.com/scheme: is used to configure at the scheme for the metrics to scrape, will apply to all ports if multiple are configured ( onlyhttporhttpsare allowed as values)telegraf.influxdata.com/interval: is used to configure interval for telegraf scraping (Go style duration, e.g 5s, 30s, 2m .. )telegraf.influxdata.com/metric-version: is used to configure which metrics parsing version to use (1, 2)telegraf.influxdata.com/namepass: is used to configure scraped metrics to preserve configuration fortelegraf, being a TOML value to add to telegraf configuration; all metrics are passed if not specified
NOTE: all annotations should be formatted as strings - for example telegraf.influxdata.com/port: "8080", telegraf.influxdata.com/metric-version: "2" or telegraf.influxdata.com/namepass: "['metric1','metric2']".
apiVersion: apps/v1
kind: StatefulSet
# ...
spec:
template:
metadata:
annotations:
telegraf.influxdata.com/class: influxdb # User defined output class
telegraf.influxdata.com/interval: 30s
telegraf.influxdata.com/path: /metrics
telegraf.influxdata.com/port: "8086"
telegraf.influxdata.com/scheme: http
telegraf.influxdata.com/metric-version: "2"
# ...
spec:
containers:
- name: influxdb
image: quay.io/influxdb/influxdb:v2.0.4
[[inputs.prometheus]]
urls = ["http://127.0.0.1:8086/metrics"]
interval = "30s"
metric_version = 2
[[inputs.internal]]
Additional pod annotations that can be used to configure the Telegraf sidecar:
telegraf.influxdata.com/inputs: is used to configure custom inputs for telegraftelegraf.influxdata.com/internal: is used to enable telegraf "internal" input plugins fortelegraf.influxdata.com/image: is used to configure telegraf image to be used for thetelegrafsidecar containertelegraf.influxdata.com/class: configures which kind of class to use (classes are configured on the operator)telegraf.influxdata.com/secret-env: allows adding secrets to the telegraf sidecar in the form of environment variablestelegraf.influxdata.com/env-configmapkeyref-<VARIABLE_NAME>: allows adding configmap key references to the telegraf sidecar in the form of an environment variabletelegraf.influxdata.com/env-fieldref-<VARIABLE_NAME>: allows adding fieldref references to the telegraf sidecar in the form of an environment variabletelegraf.influxdata.com/env-literal-<VARIABLE_NAME>: allows adding a literal to the telegraf sidecar in the form of an environment variabletelegraf.influxdata.com/env-secretkeyref-<VARIABLE_NAME>: allows adding secret key references to the telegraf sidecar in the form of an environment variabletelegraf.influxdata.com/requests-cpu: allows specifying resource requests for CPUtelegraf.influxdata.com/requests-memory: allows specifying resource requests for memorytelegraf.influxdata.com/limits-cpu: allows specifying resource limits for CPUtelegraf.influxdata.com/limits-memory: allows specifying resource limits for memorytelegraf.influxdata.com/istio-requests-cpu: allows specifying resource requests for CPU for istio sidecartelegraf.influxdata.com/istio-requests-memory: allows specifying resource requests for memory for istio sidecartelegraf.influxdata.com/istio-limits-cpu: allows specifying resource limits for CPU for istio sidecartelegraf.influxdata.com/istio-limits-memory: allows specifying resource limits for memory for istio sidecartelegraf.influxdata.com/volume-mounts: allows specifying extra volumes mount into the telegraf sidecar, the value should be json formatted, eg: {"volumeName": "mountPath"}
apiVersion: apps/v1
kind: StatefulSet
# ...
spec:
template:
metadata:
labels:
app: redis
annotations:
telegraf.influxdata.com/env-fieldref-NAMESPACE: metadata.namespace
telegraf.influxdata.com/env-fieldref-APP: metadata.labels['app']
telegraf.influxdata.com/env-configmapkeyref-REDIS_SERVER: configmap-name.redis.url
telegraf.influxdata.com/env-secretkeyref-PASSWORD: app-secret.redis.password
telegraf.influxdata.com/env-literal-VERSION: "1.0"
telegraf.influxdata.com/volume-mounts: {"xxx-3080bfa7-log":"/opt/xxx/log"}
telegraf.influxdata.com/inputs: |+
[[inputs.redis]]
servers = ["$REDIS_SERVER"]
password = "$PASSWORD"
# ...
spec:
containers:
# ...
These annotations result in additional environment variables available for the telegraf container, which can be used in for example the tags.
And they can be used in the additional input configuration provided in the annotation as shown above.
stringData:
basic: |+
[global_tags]
hostname = "$HOSTNAME"
nodename = "$NODENAME"
namespace = "$NAMESPACE"
app = "$APP"
version = "$VERSION"
This operator is community supported. InfluxData provides no official support for their use.
Pull requests and issues are the responsibility of the project's moderator(s) which may include vetted individuals outside of the InfluxData organization. All issues should be reported and managed via GitHub (not via InfluxData's standard support process).
Please read the CONTRIBUTING file for more details on how to get started with contributing to to telegraf-operator.
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent