backupccl: default to separated incremental_from for collections #75201
Labels
A-disaster-recovery
C-bug
Code not up to spec/doc, specs & docs deemed correct. Solution expected to change code/behavior.
T-disaster-recovery
In #72236, we added the ability to specify an
incremental_storage
option duringBACKUP
. When this option is present, incremental backups are stored in the specified directory rather than in the collection itself.We would like to be able to take advantage of this capability by default inside a backup collection.
Strawman requirements to be changed:
In a collection where the latest backup is a full backup, an incremental backup added via
BACKUP ... INTO LATEST IN
will be stored in an/incrementals
subdirectory/subprefix inside that collection, unless otherwise specified.In a collection where the latest backup is an incremental backup, an incremental backup added via
BACKUP INTO LATEST IN
will be stored directly in the collection.BACKUP ... INTO '{subdirectory}' IN '{destination}'
will continue to work as it does currently.RESTORE FROM {subdir} IN {destination}
without an explicitly incremental_storage option should search the appropriate directory outside of subdir (either the parent or some other directory specified by our default scheme) for full backups.Questions
Should this be behind a cluster setting in case people have automation built around the old layout? We've said in the past that we don't want people depending on a particular layout.
Should full backups also get their own subdirectory by default? I believe that without this, you could never have a lifecycle policy for full backups that didn't also apply to incrementals on S3. But, since most people likely want to retain incrementals for a shorter period than full backups, perhaps this isn't much of an issue. That said a layout like:
does have a nice look to it to me.
Jira issue: CRDB-12510
The text was updated successfully, but these errors were encountered: