repository·master·Indexed 12 days ago
https://github.com/influxdata/telegrafA highly extensible agent for collecting, processing, and exporting metrics and logs from various sources to different destinations. It features a plugin-based architecture including inputs, outputs, processors, aggregators (such as Histogram, Merge, and Quantile), parsers, serializers, and secret stores.
Telegraf is a plugin-driven agent designed for collecting, processing, aggregating, and writing metrics, logs, and other arbitrary data.
Key features include:
The execd processor plugin runs an external program as a separate, long-running process. Telegraf pipes metrics into the process's stdin and reads the processed metrics back from its stdout. Any output written to stderr is captured in the Telegraf logs.
influx data format is supported because the communication must be serialize-parse symmetrical to avoid losing critical type data.execd process.[[processors.execd]]
command = ["/path/to/your_program", "arg1", "arg2"]
# environment = ["KEY=value"]
# restart_delay = "10s"
# data_format = "influx"The opcua_listener plugin is a service input that receives data from an OPC UA server by subscribing to nodes and events.
Important Note on Service Inputs: As a service input, this plugin starts a service to listen and wait for metrics or events. Because of this behavior:
interval setting may not apply.--test, --test-wait, and --once may not produce output.[[inputs.opcua_listener]]
name = "opcua_listener"
endpoint = "opc.tcp://localhost:4840"The inputs.firehose plugin allows Telegraf to act as an HTTP listener for metrics sent via AWS Data Firehose. It follows the official AWS request-response schema.
Key Requirements:
records.[*].data field in the request body. The data in this field must be base64 encoded and must match one of Telegraf's supported input data formats.interval setting may not apply, and CLI flags like --test or --once may not produce output.[[inputs.firehose]]
service_address = ":8080"
paths = ["/telegraf"]
data_format = "value"
data_type = "string"The Heartbeat Output Plugin sends a heartbeat signal via an HTTP POST request to a specified endpoint at a regular interval. This is primarily used to monitor the presence and health of Telegraf instances within large-scale deployments.
Requirements:
v1.37.0 or higher# Example configuration snippet (conceptual)
[[outputs.heartbeat]]
url = "http://your-endpoint.com/heartbeat"
id = "telegraf-instance-01"
token = "your-auth-token"
interval = "1m"The OAuth2 Secret Store plugin retrieves and maintains secrets (tokens) from OAuth2 services like Auth0, AzureAD, or custom providers. It specifically supports the 2-legged client credentials flow and automatically renews expired tokens to prevent API call failures.
To use a secret retrieved by this store in other Telegraf plugins, reference it using the syntax @{<store-id>:<secret-key>}.
Note: Only specific Telegraf plugins and options support secret store referencing. Check the documentation of the plugin you are using for a Secret store support section to confirm compatibility.
# Example reference syntax
# If your store id is 'secretstore' and your token key is 'mytoken'
option_name = "@{secretstore:mytoken}"The MQTT Producer Output Plugin acts as an MQTT producer, writing Telegraf metrics to an MQTT broker. It supports both MQTT protocol versions 3.1.1 and 5 (defaulting to 3.1.1).
If you are using mosquitto MQTT server version v2.0.12 or later, you must set the keep_alive option to a non-zero value. Failure to do so may result in the server returning an identifier rejected error. A common value used by the eclipse/paho.golang library is 30.
The sqlserver input plugin collects lightweight metrics from Microsoft SQL Server instances using Dynamic Management Views (DMVs).
Supported SQL Server environments:
The VMware vSphere Input Plugin gathers metrics from vSphere servers of a vCenter. It collects data regarding clusters, hosts, resource pools, VMs, datastores, and vSAN information.
Requirements:
The Amazon Timestream output plugin allows Telegraf to write metrics directly to the Amazon Timestream service.
Requirements:
v1.16.0 or higherallcloud, datastoreThe win_perf_counters plugin collects metrics from Windows Performance Counters. This plugin is exclusively for Windows operating systems.
Important Note: If the specific Performance Counter Object being polled does not exist on the system at the time the Telegraf agent starts, the plugin will not gather data for that object.
By default, the plugin provides basic CPU and Disk statistics. You can customize the collection by specifying custom counter paths or using the print_all option to match specific patterns.
The opcua input plugin gathers data from an OPC UA server by subscribing to configured nodes. It supports various authentication methods, security policies, and discovery mechanisms like browse-based discovery.
Requirements:
opc.tcp://localhost:4840)[[inputs.opcua]]
endpoint = "opc.tcp://localhost:4840"
# ... other configuration