-
Notifications
You must be signed in to change notification settings - Fork 3.8k
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
sql: fix SCRUB index key order checking #32908
Conversation
Previously `SCRUB` would erroneously report that index keys were out of order for columns other than the first column in an index. This fixes the bug. Fixes cockroachdb#32874. Release note (bug fix): Fixed bug where SCRUB would erroneously report that index keys were out of order.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewable status: complete! 1 of 0 LGTMs obtained
} | ||
// After the first column with a differing value is found, the remaining | ||
// columns are skipped (see #32874). | ||
break |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
hmm, this doesn't seem quite right to me - won't this miss legitimately out of order columns? I feel like what we need to do is keep a "foundNewValue" bool for each column, setting it to true each time the column has a new value, and only enforce the ordering check for a column that follows a column where foundNewValue
is false... does that make sense?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nevermind - I got confused. The logic LGTM!
bors r+ |
32908: sql: fix SCRUB index key order checking r=lucy-zhang a=lucy-zhang Previously `SCRUB` would erroneously report that index keys were out of order for columns other than the first column in an index. This fixes the bug. Fixes #32874. Release note (bug fix): Fixed bug where SCRUB would erroneously report that index keys were out of order. Co-authored-by: Lucy Zhang <[email protected]>
Build succeeded |
The `scrub` tests have been passing on master, but failing on 2.1 because there was a bug (fixed in cockroachdb#32908). Release note: None
33573: roachtest: add MinVersion to scrub tests r=lucy-zhang a=lucy-zhang The `scrub` tests have been passing on master, but failing on 2.1 because there was a bug (fixed in #32908). Release note: None 33582: storage: bump RaftDelaySplitToSuppressSnapshotTicks r=knz a=knz The admin split path must accommodate a scenario where a range with not-yet-replicated followers is being manually split multiple times (eg. IMPORT during TPCC test fixtures). This scenario results in a bunch of replicas that all need to be populated with snapshots. To avoid backing up the raft snapshot queue, a heuristic was put in place (#32594) to delay the admin split if there is another snapshot being applied already. As shown by investigation in a failing test, there is a mismatch between the configured max delay for this heuristic (20s) and the actual duration of the snapshot application - the latter is limited by the max bandwidth for snapshots, 2 MB/s resulting in 32s applications in the worst case. We (Tobias and I) suspect that the heuristic thus fails to wait long enough to have the protective effect it was designed to provide. The current patch increases this delay to exceed this snapshot application duration estimate to about 50s. Note that this scenario is not likely to occur now that #32782 has been merged (this reduces the need for raft snapshots during splits); however in release-2.1 where that patch was not applied, the scenario is more likely. Release note (bug fix): resolve a cluster degradation scenario that could occur during IMPORT/RESTORE operations, manifested through a high number of pending Raft snapshots. Co-authored-by: Lucy Zhang <[email protected]> Co-authored-by: Raphael 'kena' Poss <[email protected]>
Previously
SCRUB
would erroneously report that index keys were out of orderfor columns other than the first column in an index. This fixes the bug.
Fixes #32874.
Release note (bug fix): Fixed bug where SCRUB would erroneously report that
index keys were out of order.