Compute and derive metrics by watching a log file

Last update: Nov 4, 2022

Overview

Export Prometheus metrics from your web server's access log

Web servers only offer limited metrics. For example, nginx only offers stub_status unless you pay for NGINX Plus, which will only give a basic count of active connections and requests served across all virtual hosts.

However nearly all web servers are configured to log all requests to a file.

This tool attaches to the log file (like tail -f does) and exposes metrics for Prometheus.

Project status

This is ready for production use, and deployed to my personal server which hosts multiple services.

How to use

$ ./access-log-to-prometheus-metrics /var/log/nginx/access.log '$host $remote_addr - $remote_user [$time_local] "$request" $status $body_bytes_sent "$http_referer" "$http_user_agent"'
Starting server at 127.0.0.1:9898
$ curl http://127.0.0.1:9898/
# HELP requests The total number of requests per HTTP status code and virtual host name
# TYPE requests counter
requests{status="200",vhost="app.taguette.org"} 123
requests{status="304",vhost="app.taguette.org"} 27
requests{status="207",vhost="cloud.rampin.org"} 8

Supported log formats

You can pass your NGINX log_format configuration line. If you are using a different web server, you will have to convert your log format to the equivalent NGINX configuration.

The following variables are read:

The HTTP status code ($status), used as a label on the requests metric
The virtual host name ($host), used as a label on the requests metric
The presence of a user name, used as label on the requests metric (yes/no/unk)
The time to process the request ($request_time), used in the request_duration metric
The size of the response's body ($response_body_size), used in the response_body_size metric

Metrics

requests is a counter for requests (e.g. log lines), organized by status (the HTTP status code) and vhost (the virtual host name)
request_duration is a histogram for request durations, organized by status (the HTTP status code) and vhost (the virtual host name)
response_body_size if a histogram for responses' body sizes in bytes, organized by status (the HTTP status code) and vhost (the virtual host name)

Labels default to the string "unk" if they are unknown (e.g. those variables are not present in your log format).

Log agent to replay time-stamped log stream

replayman Log agent to replay time-stamped log stream. Getting started Installation cargo install replayman Prepare for your data TSV with a header li

3 Dec 23, 2022

⏱ Cross-platform Prometheus style process metrics collector of metrics crate

⏱ metrics-process This crate provides Prometheus style process metrics collector of metrics crate for Linux, macOS, and Windows. Collector code is man

12 Dec 16, 2022

Fiddi is a command line tool that does the boring and complex process of checking and processing/watching transactions on EVM compatible Blockchain.

7 Jan 9, 2023

zigfi is an open-source stocks, commodities and cryptocurrencies price monitoring CLI app, written fully in Rust, where you can organize assets you're watching easily into watchlists for easy access on your terminal.

zigfi zigfi is an open-source stocks, commodities and cryptocurrencies price monitoring CLI app, written fully in Rust, where you can organize assets

18 Oct 24, 2022

A dotfiles manager, with real time fle watching and 100% less sym-links!

Kubo A dotfile manager that watches files in real time. Usage Create a directory called .kubo in $HOME, then create a file called kubo.toml in .kubo.

5 Jul 24, 2023

Plex webhook service to automatically update your Anilist watching list.

anifunnel Plex webhook service to automatically update your Anilist watching list. Description anifunnel is a web server that will consume incoming Pl

3 Dec 12, 2022

Displays the Show/Movie you're currently watching on Discord!

Jellyfin-RPC Program used to display what you're currently watching on discord. Jellyfin-RPC uses the API to check what you're currently watching, thi

3 Jan 8, 2023

Distributed compute platform implemented in Rust, and powered by Apache Arrow.

Ballista: Distributed Compute Platform Overview Ballista is a distributed compute platform primarily implemented in Rust, powered by Apache Arrow. It

2.3k Jan 3, 2023

Distributed compute platform implemented in Rust, and powered by Apache Arrow.

Ballista: Distributed Compute Platform Overview Ballista is a distributed compute platform primarily implemented in Rust, powered by Apache Arrow. It

2.3k Jan 3, 2023

3d Cellular Automata using WGPU in Rust (for the web and using compute shaders)

3D-Cellular-Automata-WGPU 3d Cellular Automata using WGPU in Rust (for the web and using compute shaders) The branches are very messy... I recommend y

18 Dec 18, 2022

Sample code for compute shader 101 training

Sample code for Compute Shader 101 This repo contains sample code to help you get started writing applications using compute shaders.

332 Jan 1, 2023

Interface definitions for the Compute@Edge platform in witx.

🔗 compute-at-edge-abi This repository contains witx definitions for the Compute@Edge platform ABI. About witx The witx file format is an experimental

14 Apr 5, 2022

Viceroy provides local testing for developers working with Compute@Edge.

Viceroy provides local testing for developers working with Compute@Edge. It allows you to run services written against the Compute@Edge APIs on your local development machine, and allows you to configure testing backends for your service to communicate with.

99 Jan 7, 2023

Compute a pairwise SNP distance matrix from one or two alignment(s)

Comments

Add custom filters/labels
It would be cool if the user could customize the behavior from the command-line.

For example:

Custom filters, e.g. only process lines where "request starts with /api" or where "vhost is example.org" or "status is 5xx"

Custom labels, e.g. "set <version> from request using ^/api/(v[0-9]+)/"

This needs the log parser to keep the original variables & values in addition to the parsed values.
enhancement
opened by remram44 0