Hey folks, I'm a new AWS user trying to launch an `f1.2xlarge` instance for some testing. My account has a quota allowing up to 8 F1 instances, so I thought I was good to go. However, no matter how many times I try, it just won't launch! I've checked everything: VPC, subnets, my AMI (FPGA Developer AMI), security groups, and even tried different Availability Zones. I created new launch templates and went through my configurations as thoroughly as possible, but nothing works. I'm also looking to upgrade to `f1.16xlarge` later, so fixing this is pretty important for my FPGA testing and development. The API hints that instances are available in certain zones, but then they fail to launch. Is there something I might be missing or a workaround that's known for new accounts? Thanks for any help!
6 Answers
Just a heads up, the F1 instances are reaching end-of-life in December 2025, and it’s worth checking if new customers might not have access anymore. It could be better to consider moving to F2 instances instead. Also, showing the exact error messages you're encountering would help others assist you. Check this notice: [GitHub End of Life Note](https://github.com/aws/aws-fpga)
Have you noticed if the instance tries to launch but immediately terminates? That could indicate a setup issue or a misconfiguration somewhere.
Have you tried launching it through the CLI? Sometimes, the web interface doesn't give detailed error messages, but the CLI could provide more context on what's going wrong. It could be worth a shot!
What error messages are you getting? Also, make sure to verify your quota for 'Running On-Demand F Instances' rather than Spot or dedicated instances. Remember, the `f1.2xlarge` has 8 vCPUs, so your quota might limit you to only one instance at a time since it's per vCPU.
Just a heads-up, the quotas for EC2 instances are actually about the number of vCPUs, not just the count of instances. If you have stopped instances, those count towards your quota too. So, make sure you're accounting for that! You might want to double-check your vCPU usage.
You should check CloudTrail for any clues regarding the RunInstances action. It might give you a better idea of what's happening behind the scenes. Just in case something's being blocked or there's a hidden issue.
I haven't tried that yet. I'll definitely give the CLI a go and see if I can spot any errors!