/dev/xvda seems to have lost files for a while - and of course no AWS support...

Are system and application logs checked?
I would need to confirm that the information that the file was lost is correct.

It would also be a good idea to check the load status of EC2 and EBS.
For example, if the volumeReadBytes value is high, it could be that there were many reads and a temporary error occurred.
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using_cloudwatch_ebs.html

If EBS read and write load is high, I think it can be improved by changing the volume type to IO2 and setting the throughput higher.
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/provisioned-iops.html

I cannot seem to find anything overt except a spike in calls on the dashboards or logs, nothing in my tomcat logs either except the 404's on those files. I will look at the volume types, that's good advice, thank you.