You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Worth to try reproducing it, by starting the upload of backup that has SSTables available for multipart upload in Scylla Manager (size must be greater that 250 Mb) and killing the agents during the upload.
Later on, there is an API of S3 available that can be used to list pending multipart uploads. https://docs.aws.amazon.com/cli/latest/reference/s3api/list-multipart-uploads.html
The way to go with this issue may be just an information added to the documentation about multipart upload life-cycle that can be attached to the backup bucket.
There is a possibility that multipart upload to S3 is broken/interrupted.
In such a situation, these mutlipart uploads should be aborted.
https://docs.aws.amazon.com/AmazonS3/latest/userguide/mpu-abort-incomplete-mpu-lifecycle-config.html
Multipart uploads abort can be delegated to RClone by calling to clean them up https://rclone.org/s3/#cleanup
We need to add the cleanup to purge stage of the backup task.
Update:
We can extend the S3 bucket setup guide to include LCM for backup buckets, so that it's delegated to S3.
https://docs.aws.amazon.com/AmazonS3/latest/userguide/mpu-abort-incomplete-mpu-lifecycle-config.html
The text was updated successfully, but these errors were encountered: