I was testing a restore of my production data backup stored on a s3 bucket where data size is 170GB with RF=3 onto a separate cluster. So, net data would be 510GB. To my surprise, the restore is taking really a lot of time, around 9-10 hrs. S3 VPC endpoint is enabled and data download speeds are very quick when done manually. Not sure if this is the expected time to restore, please let me know if I might be doing something wrong here or is it an expected time period ? If yes, how can we maintain a good RTO ?
I suggest opening a new issue for the scylla-manager, including the manager version you’re using and the logs.
It’s indeed sound too long for the restore.
Worth mention the size of original cluster and the new cluster you’re trying to restore to.