m1.small capacity issues in use1-az6
There seems to be some issue with capacity at least for m1.small instances in use1-az6 availability zone. I had several scheduled retirements for instances, and when I went to try to start the instances again after stopping them to clear the retirement, I got InsufficientInstanceCapacity errors. It hadn't occurred to me to create capacity reservations before, we're fairly small and have never had capacity issues before. I have created a capacity reservation now with as many m1.small as it would let me (which is actually still fewer than are currently running).
Mainly I'm posting here to make sure folks at AWS are aware of this issue and hopefully are working to reallocate capacity to alleviate this. I had to switch some instance types around to get my required instances running again after stopping them for the retirements, and I'd like to get them back to their proper types ASAP. Also never know when more retirements will come in.
I can't comment on capacity issue in general but I'd note that the m1 instance types are quite old - we refer to them as "previous generation" instances. They are based on older hardware (think the underlying CPU and motherboards). And as Werner Vogels (CTO of Amazon) says "Everything fails all the time". So if hardware fails (as it appears to have done in your case) it may not be able to be replaced. As you say, more retirements may happen in the future - and this is definitely more likely as the hardware ages.
I'd strongly recommend that you move to more modern instance type. Note that you may need to update the drivers that are being used. The m1 instances have been around for more than a decade so it's definitely time for an upgrade.
Yes, I have been wanting to rebuild all of our instances starting from more modern AMIs that can be run on newer instance types. I just can't find the hours to do it. One aspect that limits what instance types that can be selected is that many of these older instances are 32-bit linux, which I guess was decided at the time on the assumption of lower memory usage overhead.
You might also try another AZ within us-east-1; there may be additional capacity there. But for longevity I definitely recommend "finding the hours"; those instance types have been superseded many times now.
Check instance capacityasked 21 days ago
Does RI purchase for RDS guarantee capacity reservation?Accepted Answerasked 2 years ago
"Failed to start the instance. Insufficient Capacity" errorasked 4 months ago
unused reserved capacity with regional reserved InstanceAccepted Answerasked 6 months ago
Unable to ssh to Lightsail server when in burst zone with ample remaining burst capacityasked 4 months ago
RDS Reserved InstancesAccepted Answerasked 6 years ago
Are there capacity issues for g4dn instances in eu-west-1?asked 8 months ago
Instance availabilityasked 8 months ago
ECS Capacity providers best practicesAccepted Answerasked 2 months ago
m1.small capacity issues in use1-az6asked 13 days ago