docs update

docs/benchmark/README.md

@@ -0,0 +1,43 @@
+# HTTP Hosts Benchmarking
+
+This repository contains tools and scripts for benchmarking HTTP hosts performance.
+
+## Prerequisites
+
+Before running the benchmarks, you need to create a path list (`path_list_500k.txt`). You have two options:
+
+1. Generate from real-world server logs using `nginx_to_path_list.py`
+2. Generate randomly (Note: real-world usage patterns are typically non-random, e.g., ocean tiles are rarely accessed)
+
+## Important Notes
+
+- Run the benchmarks on `localhost`, and not over the internet! Otherwise you'd be just testing your internet speed.
+- The benchmark uses [wrk](https://github.com/wg/wrk) HTTP benchmarking tool
+
+## Usage
+
+Basic command:
+```bash
+wrk -c10 -t4 -d10s -s /data/ofm/benchmark/wrk_custom_list.lua http://localhost
+```
+
+### Parameters Explained
+
+- `-c10`: Number of connections to keep open
+- `-t4`: Number of threads to use
+- `-d10s`: Duration of the test (10 seconds)
+- `-s`: Script file to use
+
+### Thread Count Considerations
+
+- `-t1`: More accurate results as the URL list is loaded exactly in sequence
+- `-t4`: Better reflects real-world usage patterns
+
+## Results
+
+Benchmark results can be found in [results.md](results.md)
+
+## Contributing
+
+Feel free to submit your results including which hosts were used.
+

docs/benchmark/nginx_to_path_list.py

@@ -1,6 +1,10 @@
 import json
 
 
+# This script parses a nginx server log and creates a text file
+# which can be used in the Lua script.
+# The path file is not suppied in this repo.
+
 with open('access.jsonl') as fp:
     json_lines = fp.readlines()
 

docs/benchmark/wrk_custom_list.lua

@@ -1,6 +1,6 @@
 local counter = 1
 local lines = {}
-local url_base = "/planet/20231221_134737_pt/"  -- trailing slash
+local url_base = "/planet/fake_version/"  -- trailing slash
 local path_list_txt = "/data/ofm/benchmark/path_list_500k.txt"
 
 for line in io.lines(path_list_txt) do

init-server.py

@@ -129,6 +129,7 @@ def loadbalancer(hostname, user, port, noninteractive):
 @common_options
 def debug(hostname, user, port, noninteractive):
     c = get_connection(hostname, user, port)
+    run_http_host_sync(c)
 
 
 if __name__ == '__main__':

modules/http_host/benchmark/wrk_usage.txt

@@ -1,8 +0,0 @@
-wrk -c10 -t4 -d10s -s /data/ofm/benchmark/wrk_custom_list.lua http://localhost
-
-# -t1 => more correct, since the url list is loaded exactly in sequence
-# -t4 => reflecting real world usage
-
-
-
-