Complete SSH documentation

ssh.md

@@ -1,8 +1,40 @@
 # Debugging via SSH
 
 You can debug a submission by clicking the "Debug via SSH" button in
-the action bar at the bottom of a submission.
+the action bar at the bottom of a submission. This will allow you to
+log into a container that has your autograder code and the student's
+code downloaded, set up in the same way as an actual autograder run.
 
 [![Debug via SSH](debug_via_ssh.png)](debug_via_ssh.png)
 
-If you haven't done this before, you'll need to provide an SSH public key.
+## Adding an SSH key
+
+If you haven't done this before, you'll need to provide an SSH public
+key. Click on the link in the popup to do so:
+
+[![Debug via SSH: No public key](debug_via_ssh_no_public_key.png)](debug_via_ssh_no_public_key.png)
+
+You'll be asked to paste your public key into a text field. It should
+start with something like `ssh-rsa`, `ssh-dsa`, or something
+similar. If you don't have one, see [GitHub's instructions](https://help.github.com/articles/generating-a-new-ssh-key-and-adding-it-to-the-ssh-agent/#generating-a-new-ssh-key)
+for how to generate one. If copying from a terminal, make sure that
+you don't include any line breaks, because there should not be any
+line breaks in a public key.
+
+## Logging in via SSH
+
+Once you've set up your public key, clicking on the button will start
+a new container for you. As soon as it's ready, instructions will
+appear that give you a command to run to log into the container.
+
+[![SSH Instructions](ssh_instructions.png)](ssh_instructions.png)
+
+After logging in, you can inspect the filesytem and run the
+autograder. If the autograder fails to execute, you can make tweaks to
+see how to fix the autograder or the submission.
+
+## Persisting changes
+
+Once you get the autograder running correctly, you can persist changes
+by either resubmitting for the student, if the error is in the
+student's code, or by uploading a new version of the autograder.