DieterDP-ng commented on code in PR #5868:
URL: https://github.com/apache/hbase/pull/5868#discussion_r1590689156
##########
hbase-backup/src/main/java/org/apache/hadoop/hbase/backup/impl/BackupManager.java:
##########
@@ -295,51 +298,31 @@ public ArrayList<BackupImage> getAncestors(BackupInfo
backupInfo) throws IOExcep
.withRootDir(backup.getBackupRootDir()).withTableList(backup.getTableNames())
.withStartTime(backup.getStartTs()).withCompleteTime(backup.getCompleteTs()).build();
- // Only direct ancestors for a backup are required and not entire
history of backup for this
- // table resulting in verifying all of the previous backups which is
unnecessary and backup
- // paths need not be valid beyond the lifetime of a backup.
- //
- // RootDir is way of grouping a single backup including one full and
many incremental backups
+ // If the image has a different rootDir, it cannot be an ancestor.
if (!image.getRootDir().equals(backupInfo.getBackupRootDir())) {
continue;
}
- // add the full backup image as an ancestor until the last incremental
backup
- if (backup.getType().equals(BackupType.FULL)) {
- // check the backup image coverage, if previous image could be covered
by the newer ones,
- // then no need to add
- if (!BackupManifest.canCoverImage(ancestors, image)) {
- ancestors.add(image);
- }
+ // The ancestors consist of the most recent FULL backups that cover the
list of tables
+ // required in the new backup and all INCREMENTAL backups that came
after one of those FULL
+ // backups.
+ if (backup.getType().equals(BackupType.INCREMENTAL)) {
Review Comment:
I thought this was a consequence of the incremental backups using the WAL to
track changes. Incremental backups are created using the WAL-records since the
previous full backup. So I figured that any incremental backup would fully
consume that historic WAL so disk space is reclaimed. Assuming the WAL would be
cleaned up after creating the incremental backup, that means you can't have
multiple incremental backups spanning the same time period, hence it would make
sense to include all tables in the incremental backup that create.
Some experiments last week (as result of your other feedback), learned me
that using different root folders, it is possible to have incremental backups
spanning the same time period. So perhaps I was wrong about my other idea as
well.
I'll also browse some source code to find a lead on this, but isn't there a
technical design written somewhere about how backup-restore is supposed to work?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@hbase.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
