rtot

README

rtot

Run This Over There (Arr-tot)

Another async remote execution thing because that's just what the world needs, right?

No really

Jobs live in memory and they aren't automatically garbage collected. Jobs run as the same user running the rtot. SSL isn't built in, so if you're feeling paranoid you should probably put this behind nginx or whatever.

Example usage

Run the server somewhere:

rtot -a=':8457' -s='supersecret'

Hit the server over HTTP:

curl -H 'Authorization: rtot supersecret' \
  -d 'echo wat is happening' \
  http://other-server.example.com:8457/jobs

which returns the newly-created job, including its id and relative URL:

{
  "jobs": [
    {
      "href": "/jobs/0",
      "id": 0,
      "state": "new",
      "start": "0001-01-01 00:00:00 +0000 UTC",
      "complete": "0001-01-01 00:00:00 +0000 UTC",
      "create": "2014-01-12 03:42:32.314152969 +0000 UTC"
    }
  ]
}

Check back on that job:

curl -H 'Authorization: rtot supersecret' \
  http://other-server.example.com:8457/jobs/0

which returns JSON including the job's stdout, stderr, and the exit error if any:

{
  "jobs": [
    {
      "href": "/jobs/0",
      "id": 0,
      "out": "wat is happening\n",
      "state": "complete",
      "start": "2014-01-12 03:42:32.315039718 +0000 UTC",
      "complete": "2014-01-12 03:42:32.328346325 +0000 UTC",
      "create": "2014-01-12 03:42:32.314152969 +0000 UTC"
    }
  ]
}

Jobs that haven't exited yet will return with a state of "running" and any stdout or stderr that have been collected so far:

{
  "jobs": [
    {
      "href": "/jobs/0",
      "id": 0,
      "out": "ready\nset\nwait for it\n",
      "state": "running",
      "start": "2014-01-12 03:46:38.297295634 +0000 UTC",
      "complete": "0001-01-01 00:00:00 +0000 UTC",
      "create": "2014-01-12 03:46:38.296604443 +0000 UTC"
    }
  ]
}

A note on shebangs

If the data POSTed to the server does not start with #!, a shebang of #!/bin/bash is prepended, otherwise it's assumed that the shebang provided will be understood by the kernel. The server does not try to do anything fancy based on the content type of the request, and is sure to offend purists.

Job cleanup

As mentioned above, jobs are not automatically garbage collected. Instead, it's up to you to clean up after yourself:

curl -H 'Authorization: rtot supersecret' \
  -X DELETE \
  http://other-server.example.com:8457/jobs/1

The response for a successful job delete will have a status of 204 and no body.

Death

Since rtot is all about arbitrary superpowers, it's also possible to make it exit via the web, which is one way to restart it and/or purge all jobs:

curl -H 'Authorization: rtot supersecret' \
  -X DELETE \
  http://other-server.example.com:8457/

Documentation

Overview

+build debug