The EC2 User Guide includes documentation on how to avoid these issues: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/nvme-ebs-volumes.html#timeout-nvme-

The EC2 User Guide includes documentation on how to avoid these issues: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/nvme-ebs-volumes.html#timeout-nvme-ebs-volumes https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/nvme-ebs... No need for a kernel with the default set - just put it in your kernel cmdline options in grub. Make an AMI with the change after that, and you're good to go. Edit: My apologies. I didn't read your comment carefully enough, and didn't realize you had specifically called out needing to make these changes, and that older kernel versions had the 255 maximum. Leaving the above for posterity, and so people can notice I'm a bad reader :) Did you see these issues even with it set to 255? That seems like it should have been enough timeout if everything is working normally. Perhaps small GP2 volumes that are out of credits with a large queue depth might see this? I can't say I've run into it yet.