Error on ASG - UnfulfillableCapacity

[frantisekrokusek]

Hi Michael,

First thanks for this awesome template!

I am getting those errors, no matter the configuration...

Launching a new EC2 instance. Status Reason: Could not launch Spot Instances. UnfulfillableCapacity - Unable to fulfill capacity due to your request configuration. Please adjust your request and try again. Launching EC2 instance failed.

Status Reason: Could not launch Spot Instances. InvalidFleetConfiguration - Your requested instance type (p3.8xlarge) is not supported in your requested Availability Zone (us-east-2c). Launching EC2 instance failed.

• us-east-2a, us-east-2b, us-east-2c

[mludvig]

Hi Frantisek, it could be a number of reasons:

• Wait a bit longer. The ASG tries different combinations of instance types and availability zones and it may take ~30 mins or more to find one that’s available.

• Have you got high enough Service limits? You can increase them using tools/incease-quotas.sh for each different instance type.

• Try a different region, instance availability varies a lot between the regions. Try us-east-1, or us-west-2 that have the new g5.xlarge.

Hope that helps :)

[frantisekrokusek]

Thanks for your answer!

I indeed tried:

• changing region: oregon and ohio
• waiting hours...

I will try to increase service limits.

[mludvig]

In Oregon you can also enable the Los Angeles local zone - it usually has g4dn capacity available.

See here for the steps: Instnace g4dn.xlarge not starting

Also check out my article on Medium with some more info: Mining Ethereum on AWS — is it worth it?

[frantisekrokusek]

Thank you saddly I still can't figure out why it won't start...

[frantisekrokusek]

I am trying now to create a spot request manually and launch it myself.

[frantisekrokusek]

Ok I think I know what's the issue.. I have all spot requests limits to zero for G and P instance types! 🤯 Why is that?

[mludvig]

That's because that's the default for new accounts. See my comment in the first response ;)

Have you got high enough Service limits? You can increase them using tools/increase-quotas.sh for each different instance type.

Simply run tools/increase-quotas.sh g4dn from AWS CloudShell and it will raise the support requests to increase the quotas in all the regions where g4dn is available (this limit increase is valid for g5 as well). Then do the same for p3.

Start with the default - increase to 64 vCPU and once they approve it take it a step higher to e.g. 192 vCPU. If you try to raise straight from 0 to 192 they may take a long time to approve it and question why you need it.

Let me know how you go.

[frantisekrokusek]

Waiting for approval right now!

[mludvig]

Closing, hope that the limit increase sorted the issues.

[frantisekrokusek]

Actually, AWS refused to increase my limits...

[mludvig]

You can try smaller limits, e.g. 16 instead of the default 64 and only in one or two regions (e.g. us-east-1, us-west-2). You can do it manually through the support centre. There is a high demand for GPU instances so they're sometimes reluctant unless you've got a good reason to request them, esp with new accounts. Good luck with that :)

[frantisekrokusek]

I tried to increase to 4 to try the g5.xlarge