We are seeing many errors like this:
Aug 05, 2024 9:20:46 PM INFO hudson.plugins.ec2.EC2Cloud getNewOrExistingAvailableSlave
SlaveTemplate{description='Jenkins-exec-old', labels='jenkins-exec-old'}. Cannot provision - no capacity for instances: 0

Which I took to mean AWS does not have capacity for the instance type and AZ we specified.  However, upon investigation this is not the case.

• While we are receiving errors such as the above, I can launch the same instance type in all AZs with no error from AWS.
• We have also swapped the instance type to several different types (m6i.4xl, m5i.4xl, m5.4xl, and r6i.4xl) with the same results
• Our AWS account's ec2 on-demand limit is under 50% utilization
• Jenkins internal limit for this executor has also not been reached.  We are at 0/20 and 18/20 executors for the groups exhibiting this behavior

The strangest thing is that when I check our Cloudtrail logs, Jenkins is not actually submitting RunInstances requests while it's throwing this error.  I've seen this error dozens of times in our Jenkins logs, with no corresponding RunInstances API calls to AWS.