Видео

I appreciate you input. I agree that "just building something" is insufficient, however you will never gain "a deep understanding of HOW to build" without starting just building something.

Good suggestion! I will look into this!

That is very specific! I will look into this.

Hey! I promise I will make a revision video with OAC in the future. In the meantime, you can check out a revised GitHub repository I made for you that actually has this change: github.com/Cloudmancermedia/cdk-spa-deployment-oac.git In the README I have some notes on how it was achieved and why its necessary for the workaround. I hope this helps!

@@cloudmancer thanks a bunch 👍

Hey, I might be able to help with this, can you clarify what you mean by extensions? Did you mean VS Code extensions?

Hey, sure, this is a good idea. Was there a specific architecture you wanted to see using custom resources and/or lambda + SNS? What are you looking to use these custom resources to achieve?

Sure, check out the reply on your other comment.

Hey, to my knowledge, function URLs are not customizable.

Of course!

Awesome thanks for the feedback!

Hey, Strictly speaking, yes, this is a workaround and it does not do anything to change the default functionality of API Gateway. However most people who would encounter this problem are probably discovering it because they first tried to make a Lambda with API Gateway and it timed out. At least this is how the problem manifested itself for me.

It's a septum piercing. Thanks!

: (

Hey! Did you mean you wanted to migrate existing infrastructure where the templates are version-controlled through Azure Repos? Or were you simply wanting to migrate your version-controlled repos from Azure to AWS CodeCommit? Without knowing too much detail about your situation or the architecture you have provisioned in Azure, if your situation is more the former, it might be worthwhile to look into Terraform templates since its more platform-agnostic. Let me know if this answers your question!

@@cloudmancer Thanks! The scenario is my source code is on Azure Repos but I want to deploy on AWS (without using CodeCommit, just the other resources). Using an agnostic platform sounds nice, I am also checking serverless framework as my architechture is purely serverless

Yes, in that case serverless framework would be a good choice! You can keep your repos version controlled wherever you want and it should not impede your ability to use the CLI to deploy the infra to AWS or wherever you want

I am unsure what you mean by this!

Yes, any serverless infrastructure would be placed in the infrastructure folder. This will ensure that every time you push changes to source and the pipeline runs, any new lambdas or changes to existing lambdas or the api gateway will be deployed!

@@cloudmancer thank you very much !

No problem!

⁠@@cloudmancerhey Ryan,after I create the pipeline and the infrastructure code. I wonder for this CICD approach, would the development cycle be if I would like to create a new feature for my app, I first check out the dev branch and after finishing the new feature, I push it to the dev branch of my GitHub and test it with the dev environment. After testing, I would like to deploy to the production. I am not sure which below approach is correct to do. 1. I would need to merge it from the local dev branch to my local main branch and then push it to the main branch of my GitHub for the production environment 2. Send a pull request from the dev branch to the main branch and ask peer to review, if it is good, then merge it to the main branch. And I guess a merge commit will generate to the main branch then trigger codepipeline to execute

I would branch off dev and make a feature branch, work on whatever new feature, and then you could either use the console and open a PR or push directly to dev from your local. This would just depend on how protected your dev branch is and if you are working with other people on the same repo or if its just you and its a personal project. If you wanted to test your feature branch itself, you could always deploy a new CI/CD pipeline and have it point to your feature branch! Just follow the steps we did for making the dev/prod context objects and the allowed envs array and do it for your feature branch, with the --env=yourFeatureName. This would really ensure some extra safety in testing your feature. The nice part about this CI/CD pipeline is you can create/destroy as meany of them as you needed for however many envs you want. Generally speaking, I try to manage any merges into main from the console. I would push whatever you are working on to dev (whether directly or through the console with a PR) and let it deploy through the CI pipeline. If you like it and it works, I would head to GitHub and open a PR from dev into main. Once thats done, you (or a reviewer) can visually check and see all changes and make sure everything looks good before approving and merging into the main branch. This will trigger the pipeline and should deploy everything to prod. If for some reason there is a catastrophic failure from dev to prod you could always just revert main to the previous commit before it blew up and it would retrigger the pipeline to build the previous build!

Great suggestion! I will work on compiling some resources for learning. I agree there are not many good places to find stuff!

Hey! I would suggest using a Lambda layer! More than likely, you will want to use Axios for multiple lambdas, so a shared Lambda layer would be a good solution for this. Shawn Torsitano has a fantastic breakdown of how to do this on his blog at: shawntorsitano.com/blog/cdk-lambda-layers/ I also have a video showing how to do it that you can check out. I have a link to my GitHub repo in the description that contains the code. If you don't want to worry about layers, you can also package your function as a Docker image. I hope this helps!

@@cloudmancer Hey Ryan, thanks for the article . I am following your video series slowly but surely :)

Let me look into this today and I will get back to you!

Ok, figured it out. Two steps: 1. Install @types/uuid as a dev dependency in the root cdk project: npm install --save-dev @types/uuid 2. Ensure that your compiler options tsconfig.json object contains this key: "typeRoots": ["./node_modules/@types"] I believe in the repo I have on github, step two is already done so you should only need to do step one. I just pushed the change for both of these to the repo main branch, should just need to install the new changes. Let me know if this solves it for you!

Thanks for the kind words! I will do my best

Certs are great, they do definitely make you more marketable!

Great suggestion, thank you! I got you

Hey, I wanted to follow up with you on this. I have a video in the works about this, but for now I think this is probably some of the best advice I have found on how to structure CDK projects: rehanvdm.com/blog/aws-cdk-starter-configuration-multiple-environments-cicd Another option here: github.com/kevinslin/open-cdk?tab=readme-ov-file#structure I hope these help!

I am glad it helped you! To answer your question, kind of. You can have a single S3 bucket with subfolders like dev/ prod/ and have those serve as the source for a distribution. For your specific use case, I am pretty sure you would need a different distribution for each dev/ prod/ x/ folder you wanted to serve as your subdomains website. I know this is possible for sure, as far as how to configure it, let me do some digging. As far as it being a single distribution, I don't think so. While I know its possible to have multiple origins for a distribution, I believe this is intended for serving specific requests to specific paths. You can set it up to serve static content from the bucket and dynamic content from lets say a load balancer in front of an EC2, but this is meant to be the same "website" even though they are different origins. I am pretty sure the domain name wouldn't change for the client, each page would just serve content from different sources based on the path.

@@cloudmancer thanks for the explanation and tutorial. Subscribed

No problem! To follow up, I believe that in order to set up a subfolder as the origin you simply would just add the subfolder to the default root object of cloudfront. So just prod/index.html dev/index.html, etc. This way you can have just a single S3 bucket and have each distro with a subdomain set up to each subfolder in the bucket. TBH this is actually probably the preferred way to do it since there are bucket limits! Good suggestion!

@@cloudmancer thanks. I was thinking the same

Thanks for watching, I will dig into these for you!

Almost like you would have to use magic or something

Most definitely, I got you

Hey, just wanted to follow up with you, I just put out a video on how to do this

@@cloudmancer Thanks

I will come up with something for you!

Creating and deploying an AWS Lambda project for node.js, using an MVC or rather (Models, Routes and Controllers) design using the AWS CDK CLI. Just to understand how to incorporate multiple directory. Also touch on using the code below to use external files like googleapis credentials file. " SERVICE_ACCOUNT_KEY: serviceAccountKeySecret.secretValue.toString()," Thanks

This is a great suggestion and you aren't the first one to ask! I am current working on putting together a video for this.

I also wanted to follow up on the second part of your inquiry. It looks like you have your service account key for Google APIs stored in Secrets Manager, and are wanting to pass it into an env variable for a Lambda function. For this particular case, I would say an easy option would be to set your Lambda function up to fetch the secret value from Secrets Manager at execution time, rather than passing it into the Lambda as an env variable. This of course would require the correct permissions on your Lambda execution role, which I believe would just need "secretsmanager:GetSecretValue" added to the policy with the correct ARN for your secret as the resource. You can use the AWS SDK that is baked into the Lambda NodeJs runtime to fetch the secret in your code. This is a little more secure because 1, it doesn't hard code the secret value into the CloudFormation template when you deploy your CDK code and 2, it also keeps the secret from being stored in Lambda as an env variable. This way the secret stays secure in SM and isn't stored in two additional places. This also makes it so that if you ever needed to swap out the secret due to key expiration or whatever, it would not require you to re-deploy your CloudFormation stack for the changes to take effect, as Lambda would just pull the new value next time it ran.

I have never heard of that but I will check it out!

Thanks for the kind words!

I am glad it helped you!

I got you. Expect it soon.

Hey, just wanted you to know I just dropped an intro to Step Functions video today! Thanks for the request! Let me know of any others you wanna see!

Hey there, yes I do remember! Thank you for helping me!