Back when I was a dev manager, I used to monotonously do the same sequence of terminal commands 5-10 times a day: ssh into a particular host, sudo su to a user with correct permissions, cd to a project, and then open up rails console or hbase shell. I did this so many times that it became second nature. My product guy would have a problem and I'd say "Alright, let me look at HBase real quick, we'll figure that out" or "Oh I can fix that user's account, hold on".
Being an engineer, this same repetitive motion drove me nuts and I quickly got tired of it. I set out to automate as much of the process to get to a production shell prompt as soon as possible and that's when I came across expect. expect is a handy tool that allows you to write scripts against interactive text interfaces, which means you can automate a simple workflow like ssh'ing into a host and running commands against it. It fit perfectly and below I'll share with you an abstract version of my production rails console script and another I just developed to get into rails console in a Docker environment.
Sharing the code for these scripts will help explain a lot so I'm going to start off with that:
https://gist.github.com/564fbd2b3bf78bede84c4ba519d4b96c
The above is doing what we've discussed: ssh'ing into a host, changing to a different user, changing directories, and then opening up a rails console session. The way this works is that expect creates a new process (via spawn) for the ssh session and then issues (send) commands against that ssh process, 'expecting' (expect) certain output from those responses. If at any point what it expects doesn't match up after a certain time frame then it'll exit. Otherwise, it proceeds through these steps and we finally get to our interact command which brings the spawned process to the foreground allowing us to take over the session. Ta-da, we've got a remote rails console and it takes only a few seconds and considerably less typing!
The reason I got back into thinking about these expect scripts is due to Docker. I have a Dockerized Rails application that I was putting into production with AWS Elastic Beanstalk and I was updating the README with instructions on how to open rails console in that environment. While writing them I was thinking to myself "Well, this is going to be a PITA everytime", so I figured I could adapt my old prod_railsc.sh script to the task. This is what I ended up with:
https://gist.github.com/0bd53e34b8f437e6872b879bb3c483e4
This does everything very similar to our first script, except we have to fetch the image_id of the Rails Docker image before we can run the rails console command against it. expect allows us to use regular expressions to pull out specific output from our previous command, which we use above in the expect -re "(.*)\r\n(.*) " line. We then pull out the image_id variable and use that in our docker run command. This only works in Single Container nodes, but it could be adapted pretty easily if you're running other services on the same host.
We've got ourselves a fast, automated way to access rails console in both traditionally hosted and Docker hosted rail's environments. If you're interacting with your Rails instance in production to pull important database information, administer data, or even kick off one-time jobs then this script can save you sometime in the long run!
If you have an improvement or a better way of accessing your production shell then I would love to hear about it in the comments!