Is Copilot still under active development?
The last non-CVE release was almost 1 year ago. Unanswered issues are piling up. Contributors haven't responded to multiple requests for comment:
- On a PR from last year where the "AWS Copilot CLI is in maintenance mode" banner was removed https://github.com/aws/copilot-cli/pull/5975#issuecomment-2512722068
- On an issue open since last year https://github.com/aws/copilot-cli/issues/5987
- "We are not actively developing any new features for Copilot but we still do maintenance work" - does this still apply? Or does the reverted PR #5975 supersede this? https://github.com/aws/copilot-cli/discussions/5925
I understand that you've technically removed the maintenance mode warning, but we need to know if we should start thinking about another system.
To be clear, this is because we get a lot of value from Copilot, and we appreciate the work you've put into it.
@bkaws @Madrigal Is AWS Copilot in maintenance mode? Or is there a roadmap? Thank you.
Was just about to post something similar and there was no clear messaging. Would be a shame not to have a tool like this, which is a good abstraction on top of ECS!
I have been advising clients for a while now not to adopt Copilot for new projects.
@lucix-aws @wty-Bryant @dannyrandall @iamhopaul123 @Lou1415926
Hello, as you've pointed out in bullet 3, that statement from 8/30/2024 summarizes the state of the tool: we are not doing active development, but will fix any high severity security vulnerabilities.
@bkaws how long do you plan on doing it?
The fact that this is in maintenance mode and no longer actively supported should be visible at the top of the README. Can we have an update to that effect?
The fact that this is in maintenance mode and no longer actively supported should be visible at the top of the README. Can we have an update to that effect?
They sort of did this, for less than a day back in November then immediately reverted the change. I have no idea why.
https://github.com/aws/copilot-cli/commit/949daed1e1f38c9b09a0b15fc3b895364c1cb153
The blog post referenced in the commit was also removed.
Would be useful to get more clarity on this 😉
@rubiconjosh what deployment framework/platform are you pointing them towards?
Sigh, this is so frustrating!
@rubiconjosh what deployment framework/platform are you pointing them towards?
Varies, I would say SST has been working out well for me.
Copilot is a great tool, and I can't understand why AWS would cease development for it. It's even linked all over the console.
Super disappointed to learn about this having just spent the past several weeks migrating away from CDK toward Copilot, which has been providing a much more pleasant experience. Although it works fine now, the chances of it being viable two years from now in the fast-moving AWS environment are pretty low.
Please consider adding a warning message to the CLI and its documentation pages. If the tool had said "Copilot is no longer being updated" when I was getting started, I'd have shrugged and found an alternative. Instead I've got hours of sunk cost.
We've been switching all our stuff to aws cdk at this point. There are some example repos under my profile. Don't know if that helps people, but it has been a valid solution for us.
I have asked Amazon Q, it says Copilot CLI has been deprecated and provided some replacements, here are the details below, hope it helps:
> Amazon Copilot CLI has been deprecated. AWS recommends migrating to AWS App Runner or Amazon ECS with AWS CDK or AWS CloudFormation for container deployment and management.
The specific replacement depends on your use case:
For simple web applications:
• **AWS App Runner** - Fully managed service for containerized web applications
• Provides automatic scaling, load balancing, and HTTPS
For more complex container orchestration:
• **Amazon ECS** with AWS Fargate - Serverless container platform
• **Amazon EKS** - Managed Kubernetes service
For infrastructure as code:
• **AWS CDK** - Define cloud resources using familiar programming languages
• **AWS CloudFormation** - JSON/YAML templates for infrastructure
Migration tools:
• AWS provides migration guides from Copilot to these services
• The AWS CLI (use_aws tool) can help manage these services
I will have a try and update you guys if it works
I have asked Amazon Q, it says Copilot CLI has been deprecated and provided some replacements, here are the details below, hope it helps:
> Amazon Copilot CLI has been deprecated. AWS recommends migrating to AWS App Runner or Amazon ECS with AWS CDK or AWS CloudFormation for container deployment and management. The specific replacement depends on your use case: For simple web applications: • **AWS App Runner** - Fully managed service for containerized web applications • Provides automatic scaling, load balancing, and HTTPS For more complex container orchestration: • **Amazon ECS** with AWS Fargate - Serverless container platform • **Amazon EKS** - Managed Kubernetes service For infrastructure as code: • **AWS CDK** - Define cloud resources using familiar programming languages • **AWS CloudFormation** - JSON/YAML templates for infrastructure Migration tools: • AWS provides migration guides from Copilot to these services • The AWS CLI (use_aws tool) can help manage these servicesI will have a try and update you guys if it works
Thanks but that's not very useful comment here and question was directed more towards AWS maintainers 😉
@mbevc1 I don't think you're going to get a response from the "AWS maintainers" but copilot is deprecated. Everyone should have or should be moving off of it. It wouldn't surprise me if the former team was no longer at Amazon at this point. That being said, here are some repos for AWS CDK, which has long term support, for creating an environment and deploying an ECS service:
Environment VPC/Subnets/etc. ECS Service/Task Def/etc.
The first sets your VPC, subnets, NAT, ECS Cluster, and other baseline infrastructure up and the second deploys an ECS service to the ECS cluster. Quick note, you should disable the ECS Stack on initial deployment and run the CodePipeLine. The reason is that if you deploy with the ECS stack initially, it will try to pull from ECR when an image hasn't been built yet. So deploy with ECS Stack commented out, let pipeline build, then uncomment and deploy again (or just check code in). There are alternative approaches such as writing a script that does a dockerbuild and push to ECR upon ECR's creation, but that's left for the reader.
Well, this looks like a good replacement now 🚀 https://aws.amazon.com/blogs/aws/build-production-ready-applications-without-infrastructure-complexity-using-amazon-ecs-express-mode/