Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

cherrypick-1.1: sqlccl: pass needed spans to prev backup checks #19286

Merged
merged 1 commit into from Oct 16, 2017

Conversation

dt
Copy link
Member

@dt dt commented Oct 16, 2017

Previously we validated previous backups were well-formed and ordered.
We did not, however, pass the spans which needed to be covered, the way we do
when running the matching check in RESTORE, meaning that we'd accept a set of
previous backups that didn't actually cover the spans being backed up.

Re-ordering the steps in the backup plan slightly computes the matching tables
and spans before validating the prior backups (and setting the start time), thus
catching that case where the set of tables (and thus the spans for which we need
complete history in order to restore) has changed.

Another potential approach would be to automatically change the startTime for
the spans for which we are missing history, effectiely de-incrementalizing those
tables. This opens up significant additional complexity though. A simple error
at BACKUP time should at least indicate there is an issue right away, rather
than letting an operator believe they are making usable BACKUPs that cannot
actually be RESTOREd.

Previously we validated previous backups were well-formed and ordered.
We did not, however, pass the spans which needed to be covered, the way we do
when running the matching check in RESTORE, meaning that we'd accept a set of
previous backups that didn't actually cover the spans being backed up.

Re-ordering the steps in the backup plan slightly computes the matching tables
and spans before validating the prior backups (and setting the start time), thus
catching that case where the set of tables (and thus the spans for which we need
complete history in order to restore) has changed.

Another potential approach would be to automatically change the startTime for
the spans for which we are missing history, effectiely de-incrementalizing those
tables. This opens up significant additional complexity though. A simple error
at BACKUP time should at least indicate there is an issue right away, rather
than letting an operator believe they are making usable BACKUPs that cannot
actually be RESTOREd.
@dt dt requested review from maddyblue and a team October 16, 2017 16:21
@cockroach-teamcity
Copy link
Member

This change is Reviewable

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

4 participants