Dashboard to view system resources and query logs
- view timeseries graphs for: cpu, memory, storage, & networking
- receive email alerts when a resource (cpu, memory, or storage) is in high use
- ingest docker (rootless) container logs
- ingest nginx access logs
- query the logs
You're not required to use docker, nginx, or setup email (to receive alerts).
NixOS and Nix Flakes on remote host for usage.
assumptions for simplification:
- docker containers run on same host
- nginx runs on same host
promtail should safely fail for its respective scrape job if you're not running docker or nginx
in your system flake.nix inputs, add
tele.url = "github:jtara1/tele";
in your modules or an imports, append
inputs.tele.nixosModules.default
where inputs is the 1st parameter in the function assigned to outputs.
in a nix module, enable it with config such as
services.tele = {
enable = true;
email = {
host = "mail.example.com:587";
senderAddress = "[email protected]";
receiverAddress = "[email protected]";
secretsFilePath = /etc/tele/secrets.json;
};
};
email is optional and used for sending emails for notifications based on alert rules.
secretsFilePath should be a file path to a file containing:
{
"emailPlaintextPassword": "your-email-password"
}On host OS and its network, you should expose or redirect to its http://localhost:3010,
grafana, the dashboard for querying and graphs
Add security. Change grafana dashboard password. Allowlist your IP from which you're accessing it. Encrypt the channel via TLS or other. etc.
Go to the url for grafana, defaults to http://localhost:3010
default login:
username: admin
password: admin
You can query logs, create visualizations, load a dashboard, or check alerts:
- Explore
- Dashboards
- Node Exporter Full http://localhost:3010/d/rYdddlPWk/node-exporter-full?orgId=1&refresh=1m
- Alerting
- migrate everything to nixos config declaration
- nixos config services.promtail
- relabel docker apps logs for readability (container name instead of container id, the long-hashes)
-
.sh instead of ansible? Makefile instead of ansible?can import nix flake - promtail to ingest logs for non-rootless (default) docker containers
- config to conditionally set docker deamon.json instead of docker rootless
- improve promtail job, docker
- re-add nginx access logs pipeline
- prometheus for system monitoring and metrics
- 1st alerts rule for memory
- alert notifications via SMTPS/email
- alert rules for core resources: cpu, storage, memory
- ingest logs from multiple virtual machines in dedicated tele server?
- fix and test the nginx config for local dev and for my server
- refactor into nix module(s) to be more portable for non-flake NixOS users
- core health dashboard - pre-configured visualization for core resources
- improve management of email password for ease of secure and declarative use
- publish to flakehub
- error: timestamp of a queried log is invalid
- alerts may need variable datasource uid