runner

master:

A personal project that runs commands and submits the execution result to a configurable instance of healthchecks.io (>=v3, requires auto-provisioning).

The URL for your HealthChecks instance, including your ping_key, is required at runtime*. Because the ping_key is considered a secret, users may wish to keep it out of their shell history and out of any cron scripts that are calling hc-runner; to this end, as an alternative to the --url flag, the URL can also be specified in a config file or by the HC_RUNNER_URL environment variable. All other options are taken only from command line flags.

Please consider restricting access (e.g. chmod 0600) to any files that contain your ping_key, possibly including the hc-runner config file.

* If you're using the hosted healthchecks server, your URL may look something like https://hc-ping.com/{ping_key}/.

Quickstart

$ cargo run -q -- --help
Command runner for healthchecks.io

Usage: hc-runner [OPTIONS] --slug <NAME> <COMMAND>...

Arguments:
  <COMMAND>...

Options:
  -q, --quiet              Silence logging / warnings. Does not affect called command's output
  -s, --slug <NAME>        Set healthchecks slug for this call
      --success-only       Disable calling `/start` and only ping healthchecks if the test was successful
  -t, --timeout <TIMEOUT>  Set timeout for requests to healthchecks server [default: 10]
  -u, --url <URL>          Specify the URL of the healthchecks server for this call
  -v, --verbose...         Increase logging verbosity. May be repeated. Defaults to `Level::WARN`
  -h, --help               Print help
  -V, --version            Print version

runner:

• by default sends a request to /start to mark the beginning of the scripts execution and uses ?create=1 here to create a new healthcheck for this slug if it doesn't already exist
• by default sends a request to /{status_code} to mark the end of execution and reflect the exit status (e.g. /0 for successful exit) and sends stderr as the body
• mirrors the exit status, stdout, and stderr of the called command
• can optionally only report successful runs with --success-only
  • this will prevent failure notifications for services that are expected to fail sometimes, but for which notifications are still desired if there isn't at least one successful run per (healthchecks-configured) time period
  • does not report execution time or collect stderr
• can disambiguate flags in the called command using -- trailing args syntax, e.g.:
  • hc-runner -v -- command makes hc-runner more verbose
  • hc-runner -- command -v passes the -v flag to command
  • hc-runner -v -- command -v does both
• prepends commands with /usr/bin/caffeinate on MacOS to keep long-running commands alive

Example:

$ git clone https://github.com/n8henrie/runner-rs.git
$ cd runner-rs
$ export HC_RUNNER_URL=http://your.server.url
$ cargo build --release
$ ./target/release/runner --slug say_foo -- echo foo
foo
$ echo $?
0
$ ./target/release/runner \
    --slug epic_fail \
    -- bash -c 'echo bar >/dev/stderr; exit 1'
bar
$ echo $?
1

Notes

debugging

-vvv is your friend. Note that the output will contain your ping_key.

Due to the default of create=1, you will pollute your HealthChecks instance when testing with fake slugs (--slug=foo), but your output will be cluttered with errors if you use a fake URL (--url=http://broken). During testing / experimentation, consider using --success-only with a command that is guaranteed to fail (e.g. false) which will prevent calls to the server entirely.

testing

The integration tests use the httpmock library to provide a mock server. Testing should be done with --test-threads=1 or errors will likely result. The make test target sets this for you.

macOS

On macOS, runner prefixes commands with caffeinate in order to keep long-running processes awake.

Newer versions of macOS have built-in privacy and security tools that may prevent runner from accessing sensitive directories like ~/Documents, particularly if run in an automated script from launchd. The install-macos target in the Makefile includes a workaround that should read RUNNER_URL from config.env (see config-sample.env), compile and install the project, then present some permissions dialogs to allow you to give access to these directories to runner. If the scripts you are running do not access sensitive directories such as ~/Desktop, ~/Documents, ~/Downloads, don't bother with this. If you do use this approach, you'll have to remove a check named runner-rs-setup-delete-me from your healthchecks.io instance.

Alternatives

There are several similar projects on crates.io that may be much more comprehensive and/or functional than this hobby project. I encourage you to check them out! Here are a couple:

• https://github.com/msfjarvis/healthchecks-rs
• https://github.com/dimo414/task-mon

Acknowledgements

• Pēteris Caune, creator of healthchecks.io!