slumberd

An HTTP server which sleeps for a configurable amount of time, largely useful as a way to test functionality of HTTP proxies when dealing with long-lived requests.

Available on Docker Hub as naftulikay/slumberd.

Usage

slumberd is extremely flexible and can provide static or random request durations configurable via the CLI, specific request paths, query string parameters, and/or request headers. slumberd also returns a JSON response describing what it did, and sets response headers providing similar information.

Sleep for 500ms

$ curl -i http://127.0.0.1:8080/sleep/500
HTTP/1.1 200 OK
content-length: 143
x-request-id: 4699f361-3f53-4de9-ae01-1f36e8316cf5
content-type: application/json
x-slumber-time-millis: 500
x-slumber-type: fixed
x-slumber-time: 500ms
date: Wed, 23 Oct 2019 21:17:37 GMT

{
  "slumber": {
    "type": "fixed",
    "time_millis": 500,
    "time": "500ms"
  },
  "request_id": "4699f361-3f53-4de9-ae01-1f36e8316cf5"
}

Sleep for a Random Duration Between 500-1000ms

$ curl -i http://127.0.0.1:8080/random/500/1000
HTTP/1.1 200 OK
content-length: 255
x-slumber-min-time-millis: 500
x-request-id: c4f2c297-f95c-4e14-9e1b-7f634ff2afdd
x-slumber-min-time: 500ms
content-type: application/json
x-slumber-max-time-millis: 1000
x-slumber-max-time: 1s
x-slumber-time-millis: 541
x-slumber-type: random
x-slumber-time: 541.962699ms
date: Wed, 23 Oct 2019 21:18:16 GMT

{
  "slumber": {
    "type": "random",
    "time_millis": 541,
    "time": "541.962699ms",
    "max_time": "1s",
    "max_time_millis": 1000,
    "min_time_millis": 500,
    "min_time": "500ms"
  },
  "request_id": "c4f2c297-f95c-4e14-9e1b-7f634ff2afdd"
}

For complete usage information, see USAGE.md.

Performance and Portability

slumberd uses actix-web as a platform and uses Rust's zero-cost futures to "sleep" on each request. Individual requests do not block the thread that they are executing on, and as such, slumberd should scale fairly linearly with the amount of requests it receives versus the available network bandwidth and number of logical CPU cores available on the host.

One operating system thread per logical CPU is spawned, with each running a Tokio event loop.
Memory usage on cold boot is around 25MiB. I have not profiled slumberd under load, but similar Actix applications
I have written in the past will usually remain in the realm of ~100-200MiB under load.
The --release stripped static binary for Linux at time of writing is 4.06MiB. The Docker image should also be of
a similar size as it contains nothing but the binary.
The static musl binary has zero system requirements other than an x86_64 architecture and any modern Linux kernel
(I'm assuming that any kernel version >=2.6 should work just fine).