Specifying multiple backupPolicies to perform backup to multiple S3 bucket/endpoint

[y1r]

Currently, MySQLCluster v1beta2 / backupPolicyName is *string thus we can only specify one backupPolicy for each MySQLCluster. However, in our use case, we'd like to specify multiple backupPolicies to support multiple CronJob of backup operation.

By implementing this feature, we can perform more durable backup like:

• use multiple S3 endpoints like AWS's S3 and some in-house S3 compatible storage to backup
• use multiple S3 bucket with different storageClass in AWS's S3

Do you have any future plan to implement this? Or, do you have any idea to implement this feature by user-side?

[ymmt2005]

My thoughts:

For multiple CronJobs, we can extend BackupPolicy so that we can define multiple schedules.

For durable storage, we can also extend BackupPolicy to specify multiple storage buckets.
This is better than having multiple BackupPolicies/CronJobs because it can reuse the same
dump. That being said, I'd rather use rclone or something like that
to make backups more durable.

[y1r]

Thank you for your quick response!

we can also extend BackupPolicy to specify multiple storage buckets.
This is better than having multiple BackupPolicies/CronJobs because it can reuse the same
dump

This also works for our situation.

That being said, I'd rather use rclone or something like that
to make backups more durable.

Do you mean backup to bucket A, then perform sync from bucket A to bucket B by using rclone right? If the backup has corrupted on bucket A, the corruption will be propagated to bucket B. Also, backup failure on bucket A leads immediately fail on bucket B because we cannot read from bucket A. Therefore, I think uploading bucket A and bucket B simultaneously from moco-backup CronJob seems more durable for our situation.

[ymmt2005]

Therefore, I think uploading bucket A and bucket B simultaneously from moco-backup CronJob seems more durable for our situation.

That's fair.
Do you intend to contribute these enhancements?