Thanks for the great S3 backup feature! It changed our offsite setup, and backups to S3 run reliably every night.
The one weak spot we found is restoring large guests. Now and then one chunk read from S3 hangs and then fails with error reading a body from connection, and the whole restore aborts. With a 400 GiB container that is ~265 GiB to download, and each attempt starts from scratch, so whether a restore finishes turns into a matter of luck (for us, 2 of 2 direct restores failed). Pulling the snapshot into a local datastore with a sync job first and restoring from there works well, because a failed sync can simply be rerun. Is a retry for failed chunk reads planned?
Or is there a recommended setting, such as limit-passive-requests, that helps in this case?