telegraf
/telegraf
Konfiguruje Telegraf jako agenta zbierającego metryki infrastruktury i aplikacji: wtyczki wejściowe dla baz i usług, wyjścia do InfluxDB lub Prometheus oraz własne potoki metryk.
- Autor
- TerminalSkills ↗
- Repozytorium
- skills ↗
- Licencja
- Apache-2.0
- Ostatnia zmiana repozytorium
- 25.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 8
Brak sygnałów. Repozytorium zmienione 25.10.2026, bez odwołań do wycofanych modeli i API.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
Co robi i kiedy
Konfiguruje Telegraf jako agenta zbierającego metryki infrastruktury i aplikacji: wtyczki wejściowe dla baz i usług, wyjścia do InfluxDB lub Prometheus oraz własne potoki metryk.
Oryginalny opis (EN)
Configure Telegraf as a metrics collection agent for infrastructure and application monitoring. Use when a user needs to collect system metrics, set up input plugins for databases and services, configure output to InfluxDB or Prometheus, or build custom metric pipelines.
Jak działa
- Overview
- Instructions
- Examples
- Guidelines
Podgląd SKILL.md
---
name: telegraf
description: >-
Configure Telegraf as a metrics collection agent for infrastructure and
application monitoring. Use when a user needs to collect system metrics,
set up input plugins for databases and services, configure output to
InfluxDB or Prometheus, or build custom metric pipelines.
license: Apache-2.0
compatibility: "Telegraf 1.38+ (checked against 1.40), InfluxDB 2.x or 3.x"
metadata:
author: terminal-skills
version: "1.1.0"
repository: "https://github.com/influxdata/telegraf"
category: devops
tags: "telegraf, metrics, influxdb, monitoring, observability"
---
# Telegraf
## Overview
Set up Telegraf to collect, process, and forward metrics from systems, databases, and applications. Covers input/output plugin configuration, metric filtering, aggregation processors, and deployment patterns.
Two behaviours matter on current releases (1.38 and later):
- **Strict environment-variable handling is the default.** `$VAR` and `${VAR}` are substituted only inside quoted string values; a variable that is not set is **not** reported: `${NAME}` stays as literal text (or the value ends up empty), so write `${NAME:?NAME is required}` for anything mandatory, and a literal `$` must be written `$$`. Start with `--non-strict-env-handling` only as a stop-gap.
- **`fieldpass`/`fielddrop` are deprecated.** Use `fieldinclude`/`fieldexclude`; `namepass`, `namedrop`, `tagpass`, `taginclude` and friends are unchanged.
Install with the vendor's apt/yum repository (package `telegraf`, config in `/etc/telegraf/`) or run the `telegraf` Docker image (see Task F). Validate before every deploy:
```bash
telegraf --config /etc/telegraf/telegraf.conf --config-directory /etc/telegraf/telegraf.d --test # one gather, prints line protocol, writes nothing
```
## Instructions
### Task A: Basic System Metrics Collection
```toml
# /etc/telegraf/telegraf.conf — Collect system metrics and send to InfluxDB
[global_tags]
environment = "production"
region = "us-east-1"
[agent]
interval = "10s"
round_interval = true
metric_batch_size = 1000
metric_buffer_limit = 10000
flush_interval = "10s"
hostname = ""
omit_hostname = false
skip_processors_after_aggregators = true # do not re-run processors on aggregator output
[[outputs.influxdb_v2]]
urls = ["http://influxdb:8086"]
token = "${INFLUX_TOKEN:?INFLUX_TOKEN is required}" # set in /etc/default/telegraf or the unit environment
organization = "myorg"
bucket = "infrastructure"
[[inputs.cpu]]
percpu = true
totalcpu = true
collect_cpu_time = false
[[inputs.mem]]
[[inputs.disk]]
ignore_fs = ["tmpfs", "devtmpfs", "devfs", "iso9660", "overlay", "aufs", "squashfs"]
[[inputs.diskio]]
[[inputs.net]]
interfaces = ["eth0", "ens5"]
[[inputs.system]]
[[inputs.processes]]
[[inputs.kernel]]
```
### Task B: Application and Database Inputs
```toml
# /etc/telegraf/telegraf.d/databases.conf — Database monitoring inputs
[[inputs.postgresql]]
address = "postgres://telegraf:${PG_PASSWORD}@localhost:5432/myapp?sslmode=disable"
databases = ["myapp"]
[inputs.postgresql.tags]
service = "postgres"
[[inputs.mysql]]
servers = ["telegraf:${MYSQL_PASSWORD}@tcp(localhost:3306)/"]
metric_version = 2
gather_table_schema = true
gather_process_list = true
gather_replica_status = true
[inputs.mysql.tags]
service = "mysql"
[[inputs.redis]]
servers = ["tcp://localhost:6379"]
password = "$REDIS_PASSWORD"
[inputs.redis.tags]
service = "redis"
[[inputs.nginx]] # needs ngx_http_stub_status_module enabled
urls = ["http://localhost:8080/nginx_status"]
[inputs.nginx.tags]
service = "nginx"
```
```toml
# /etc/telegraf/telegraf.d/docker.conf — Container metrics
[[inputs.docker]]
endpoint = "unix:///var/run/docker.sock"
gather_services = false
container_name_include = []
perdevice_include = ["cpu"]
total_include = ["cpu", "blkio", "network"]
[inputs.docker.tags]
input = "docker"
```
### Task C: Prometheus Input/Output
```toml
# /etc/telegraf/telegraf.d/prometheus.conf — Scrape Prometheus endpoints
[[inputs.prometheus]]
urls = [
"http://app-server:8080/metrics",
"http://payment-service:8080/metrics",
]
metric_version = 2
url_tag = "scrape_url"
# Expose metrics as Prometheus endpoint for Prometheus to scrape
[[outputs.prometheus_client]]
listen = ":9273"
metric_version = 2
export_timestamp = true
```
### Task D: Metric Processing and Filtering
Filter at the plugin that produces the metric whenever you can; it is cheaper than dropping later. `namepass` on a *processor* only restricts which metrics that processor touches, the rest still flow on.
```toml
# /etc/telegraf/telegraf.d/processing.conf — Filter and transform metrics
# Drop noisy fields at the input (add this to the Task A [[inputs.disk]] instead of a second
# [[inputs.disk]] block, which would collect every disk twice)
[[inputs.disk]]
fieldexclude = ["inodes_*"]
# Drop whole metrics from the stream (processors.filter, Telegraf 1.29+)
[[processors.filter]]
order = 1
default = "pass"
[[processors.filter.rule]]
name = ["docker_container_blkio"]
action = "drop"
[[processors.converter]]
order = 2
[processors.converter.fields]
float = ["usage_idle", "usage_user", "usage_system"]
# Tag metrics based on field values
[[processors.starlark]]
order = 3
namepass = ["cpu"]
source = '''
def apply(metric):
idle = metric.fields.get("usage_idle")
if idle != None:
used = 100 - idle
if used > 90:
metric.tags["cpu_alert"] = "critical"
elif used > 70:
metric.tags["cpu_alert"] = "warning"
return metric
'''
# Rename last, so earlier plugins and filters still see the original name
[[processors.rename]]
order = 4
[[processors.rename.replace]]
measurement = "cpu"
dest = "system_cpu"
# Aggregate before sending: emits extra metrics with _mean/_max/_min/_count field suffixes
[[aggregators.basicstats]]
period = "60s"
drop_original = false
stats = ["mean", "max", "min", "count"]
namepass = ["http_response"]
```
### Task E: HTTP and Custom Inputs
```toml
# /etc/telegraf/telegraf.d/http.conf — HTTP endpoint checks and API polling
[[inputs.http_response]]
urls = [
"https://api.shopfront.dev/health",
"https://www.shopfront.dev",
]
response_timeout = "5s"
method = "GET"
follow_redirects = true
response_status_code = 200
[inputs.http_response.tags]
check = "uptime"
[[inputs.http]]
urls = ["http://app-server:8080/api/stats"]
method = "GET"
data_format = "json"
json_name_key = "metric_name"
json_time_key = "timestamp"
json_time_format = "unix"
[inputs.http.tags]
source = "app-api"
# Execute custom scripts for metrics
[[inputs.exec]]
commands = [["/opt/scripts/check_queue_depth.sh", "--queue", "orders"]]
timeout = "5s"
data_format = "influx"
interval = "30s"
```
### Task F: Docker Deployment
```yaml
# docker-compose.yml — Telegraf with InfluxDB
services:
telegraf:
image: telegraf:1.40
volumes:
- ./telegraf.conf:/etc/telegraf/telegraf.conf:ro
- ./telegraf.d:/etc/telegraf/telegraf.d:ro
- /var/run/docker.sock:/var/run/docker.sock:ro
- /proc:/host/proc:ro
- /sys:/host/sys:ro
environment:
- HOST_PROC=/host/proc
- HOST_SYS=/host/sys
- INFLUX_TOKEN=${INFLUX_TOKEN}
group_add:
- "${DOCKER_GID}" # numeric group of the socket: stat -c '%g' /var/run/docker.sock
depends_on:
- influxdb
influxdb:
image: influxdb:2.8
ports:
- "127.0.0.1:8086:8086"
volumes:
- influxdb_data:/var/lib/influxdb2
environment:
- DOCKER_INFLUXDB_INIT_MODE=setup
- DOCKER_INFLUXDB_INIT_USERNAME=admin
- DOCKER_INFLUXDB_INIT_PASSWORD=${INFLUX_ADMIN_PASSWORD}
- DOCKER_INFLUXDB_INIT_ADMIN_TOKEN=${INFLUX_TOKEN}
- DOCKER_INFLUXDB_INIT_ORG=myorg
- DOCKER_INFLUXDB_INIT_BUCKET=infrastructure
volumes:
influxdb_data:
```
## Examples
### Example 1: "Send host metrics from my VM to InfluxDB and check the config first"
Write the Task A file to `/etc/telegraf/telegraf.conf`, put `INFLUX_TOKEN=...` in `/etc/default/telegraf`, then:
```bash
telegraf --config /etc/telegraf/telegraf.conf --test | head -5
sudo systemctl restart telegraf
```
`--test` prints line protocol such as `cpu,cpu=cpu-total,host=web-01 usage_idle=96.2,usage_user=2.1 ...` and exits without writing. With `${INFLUX_TOKEN:?...}` an unset variable makes the load fail with your message; a plain `$INFLUX_TOKEN` or `${INFLUX_TOKEN}` would not fail, and `--test` does not start outputs, so it cannot reveal a bad token.
### Example 2: "Expose Postgres and Redis metrics for my existing Prometheus"
Put the Task B inputs for `postgresql` and `redis` in `telegraf.d/databases.conf`, replace the InfluxDB output with `[[outputs.prometheus_client]]` (`listen = ":9273"`), and add a scrape job pointing at `host:9273/metrics`. Telegraf serves the latest values of every input as Prometheus text; with `metric_version = 2` field names become labels-friendly series such as `postgresql_xact_commit`.
## Guidelines
- Use `telegraf.d/` for modular configs, one file per input category.
- Set `metric_buffer_limit` high enough to ride out an output outage; metrics beyond it are dropped.
- Prefer `namepass`/`namedrop` and `fieldinclude`/`fieldexclude` on inputs to cut cardinality and storage cost.
- Never put passwords in the config file; reference environment variables, and mark the mandatory ones with `${NAME:?message}` because strict mode does not fail on unset variables.
- Create a read-only database user for `postgresql` and `mysql`; never monitor as a superuser.
- Mounting `/var/run/docker.sock` gives the container control of the Docker daemon even when mounted read-only; use `inputs.docker` only on hosts where that is acceptable, or point `endpoint` at a socket proxy.
- Use Starlark for logic that would take several chained processors; set `order` when processors depend on one another.
- For InfluxDB 3 Core/Enterprise use `[[outputs.influxdb_v3]]` (`urls`, `token`, `database`) instead of `influxdb_v2`.
- Since 1.36 the Linux `mem` input reports `used_percent` from `MemAvailable`, so it reads higher than on older versions; recheck alert thresholds after an upgrade.
- Monitor Telegraf itself with `[[inputs.internal]]` to catch dropped metrics and slow gathers.
Pliki w paczce (2)
- SKILL.md 10 KB
- _scores.json 2 KB
Wersje
W bazie nie ma innych skilli o nazwie telegraf.
Migawka bazy z 13.11.2026 · wersja paczki b8435e72a863 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100prometheus-configuration
Konfiguruje Prometheus do zbierania i przechowywania metryk oraz monitorowania infrastruktury i aplikacji, wraz z alertingiem. - 100prometheus-configuration
Konfiguruje Prometheus do zbierania i przechowywania metryk oraz monitorowania infrastruktury i aplikacji, wraz z alertingiem. - 100prometheus-configuration
Konfiguruje Prometheus do zbierania i przechowywania metryk oraz monitorowania infrastruktury i aplikacji, wraz z alertingiem. - 100prometheus-configuration
Konfiguruje Prometheus do zbierania i przechowywania metryk oraz monitorowania infrastruktury i aplikacji, wraz z alertingiem. - 100prometheus-configuration
Konfiguruje Prometheusa do zbierania, przechowywania i monitorowania metryk infrastruktury i aplikacji przy wdrażaniu zbierania metryk oraz infrastruktury monitoringu. - 100prometheus-configuration
Konfiguruje Prometheusa do zbierania, przechowywania i monitorowania metryk infrastruktury i aplikacji przy wdrażaniu zbierania metryk oraz infrastruktury monitoringu.