💥 crashie — a little failure in a box

Crashie is a Command-Line Utility that exits with a random exit code after a configurable delay. Use it when you want to test restart behaviors or anything that requires an application to fail.

Sleeping for 12.72 seconds, then exiting with code 130
Exiting with code 130

Usage Example

If you would like to randomly fail with a SIGINT (code 130) or SIGKILL (137) after 10 ± 2 seconds, run:

crashie --sigint --sigkill --delay=10 --delay-stddev=2
crashie --signals=2,3 --delay=10 --delay-stddev=2
echo $?

Alternatively, provide options using environment variables:

CRASHIE_SIGNALS=2,3 CRASHIE_SLEEP_DELAY=10 CRASHIE_SLEEP_DELAY_STDDEV=2 crashie

Crashie provides TCP and UDP echo functionalities. This comes in handy if you wand to test resilient connection logic, port forwarding (notably Kubernetes' kubectl port-forward) or similar aspects.

To bind crashie to TCP sockets, use the CRASHIE_BIND_TCP_ECHO environment variable or run e.g.

crashie --bind-tcp-echo 127.0.0.1:8080

Likewise, UDP echo is supported. For that, use the CRASHIE_BIND_UDP_ECHO environment variable or run e.g.

crashie --bind-udp-echo 127.0.0.1:8080

On Linux, you can test the echo behavior e.g. using netcat (nc 127.0.0.1 8080 for TCP or nc -u 127.0.0.1 8080 for UDP).

To simplify work with HTTP connections, you can also bind an HTTP "echo". For that, use the CRASHIE_BIND_HTTP_ECHO environment variable or run e.g.

crashie --bind-http-echo 127.0.0.1:8080

You can test the connection e.g. with curl (curl -v localhost:8080). As of now, the server always ignores the request specifics and responds with 204 No Content.

To support cases where responses must be 200 OK exactly - e.g. for liveness probes in ingress checks - you can provide the CRASHIE_HTTP_LIVENESS_PROBE_PATH or --http-liveness-probe-path argument:

crashie --bind-http-echo 127.0.0.1:8080 --http-liveness-probe-path /.health/livez

In this situation, calls to curl -v localhost:8080 result in a 204 No Content:

* processing: localhost:8080
*   Trying [::1]:8080...
* Connected to localhost (::1) port 8080
> GET / HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/8.2.1
> Accept: */*
>
< HTTP/1.1 204 No Content
< Server: crashie/0.3.0
< Date: Sat, 06 Jan 2024 14:44:53 GMT
< Content-Length: 0
< Cache-Control: no-cache, no-store
<
* Connection #0 to host localhost left intact

... while curl -v localhost:8080/.health/livez results in a 200 OK:

* processing: localhost:8080/.health/livez
*   Trying [::1]:8080...
* Connected to localhost (127.0.0.1) port 8080
> GET /.health/livez HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/8.2.1
> Accept: */*
>
< HTTP/1.1 200 OK
< Server: crashie/0.3.0
< Date: Sat, 06 Jan 2024 14:44:59 GMT
< Content-Length: 0
< Cache-Control: no-cache, no-store
<
* Connection #0 to host localhost left intact

Running via Docker

The application is available as the sunside/crashie Docker image. To run crashie via Docker, use e.g.

docker run --rm sunside/crashie --help

Provide command-line arguments as if you were running it locally:

docker run --rm sunside/crashie --sigint --sigkill --delay=10 --delay-stddev=2
echo $?

Alternatively, provide configuration via environment variables:

docker run --rm \
  --env CRASHIE_SIGNALS=2,3 \
  --env CRASHIE_SLEEP_DELAY=10 \
  --env CRASHIE_SLEEP_DELAY_STDDEV=2 \
  sunside/crashie
echo $?

Local Installation from crates.io

To install crashie from crates.io, run

cargo install crashie

Run from source

To get a documentation, run

cargo run -- --help