You can start a virtual environment in python in order to not pollute your environment, but otherwise the worker named "calculator.py" should have everything you need to have a running working. I used the dev instance and also played around with the docker containers to understand a little bit more about how open source users might be scaling this out.
The architecture of this follows what seemed to be best practices.
The worker is in a worker folder.
The workflows has two workflows, only one of which is used. That is the aws_workflow that runs the following activities.
- create vpc
- create subnet
- create internet gateway
- create route table
- associate route table with subnet
- create a security group in the subnet to associate with the instance id
- return the instance ID.
When finished running it should create a t2 micro, although that is configurable using a .env file. You'll need the following in the .env file in order to run this temporal workflow.
.env
AWS_ACCESS_KEY_ID=???
AWS_SECRET_ACCESS_KEY=????
AWS_REGION=us-west-1
SECURITY_GROUP_NAME=Temporal_Security_Group
KEY_PAIR_NAME=gmoney
AMI_ID=ami-0cbe318e714fc9a82 # just a standard aws ami but you could use whatever you want here too
Then you can run the below to start the task.
temporal workflow start \
--task-queue aws-setup-task-queue \
--type SETUP_AWS
Then you can use the below command to start the worker.
poetry run python worker/calculator.py Remember that poetry will be sad if you run it without a virtual environment.
Things that I would like to implement next time are a destory if there is a failure so that I don't need to clean things up by hand.
This felt generally straight forward, although I needed to set the retry's to 1 since in this application, if something fails, I need to cleanup the state before I want to continue. I could absolutely just take in the values I already have to destory right after I do it but want to keep them around to show I did them.
#TODO:
There are a few things I'll do in the coming days before the interview. 1.) Clean up the project 2.) Complete this readme
I spent a lot of project time implementing and getting my head around temporal, and if I had more time I would look into the below.
1.) I'll try to work in some more distributed system concepts like
- Timers for the the times I'm waiting for instances to come up
- Adding tests, I did play with those in other projects that I forked and they looked really helpful.
- Signals and Versioning.
- Automatic rollback if a workflow fails