-
Notifications
You must be signed in to change notification settings - Fork 4k
sql: remove gossip-based DistSQL physical planning #157077
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
Conversation
In 25.1 release we switched to using SQL-instance-based DistSQL physical planning but left gossip-based planning just in case (controlled via a cluster setting). We haven't seen any fallout from that, and given that many different workloads have been upgraded to 25.2, it seems safe to completely remove gossip-based planning now. Release note: None
mgartner
left a comment
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.
@mgartner reviewed 4 of 4 files at r1, all commit messages.
Reviewable status:complete! 1 of 0 LGTMs obtained (waiting on @DrewKimball and @michae2)
michae2
left a comment
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.
@michae2 reviewed 4 of 4 files at r1, all commit messages.
Reviewable status:complete! 2 of 0 LGTMs obtained (waiting on @DrewKimball)
|
TFTRs! bors r+ |
|
An interesting side-effect of using sql-instance-based planning is that it protects placement of TableReaders against a range split within SQL row. This is the case because we stitch together spans that look like they belong to the same SQL row cockroach/pkg/sql/distsql_physical_planner.go Line 1695 in cddd383
|
157077: sql: remove gossip-based DistSQL physical planning r=yuzefovich a=yuzefovich In 25.1 release we switched to using SQL-instance-based DistSQL physical planning but left gossip-based planning just in case (controlled via a cluster setting). We haven't seen any fallout from that, and given that many different workloads have been upgraded to 25.2, it seems safe to completely remove gossip-based planning now. Epic: None Release note: None 157079: sql: speed up recently added TestTruncateLarge r=yuzefovich a=yuzefovich We just saw a timeout in tests where 15m limit was exceeded because more than 11m were consumed by TestTruncateLarge. In order to speed up the test, this commit reduces the number of tables affected as well as runs both subtests in parallel (which - hopefully - should be ok given that we skip this test under all heavy configs). Fixes: #156998. Release note: None 157086: go.mod: bump Pebble to f11535e4f8ec r=annrpom a=annrpom Changes: * [`f11535e4`](cockroachdb/pebble@f11535e4) db: use Peek in TestVersionSetSeqNums * [`1ba71aa2`](cockroachdb/pebble@1ba71aa2) wal: consolidate LogRecycler.[Set]MinRecycleLogNum * [`82d592ec`](cockroachdb/pebble@82d592ec) db: document that multiple flushableEntries may share a logNum * [`9b012186`](cockroachdb/pebble@9b012186) db: add metric encapsulating blob file rewrite bytes read, written * [`b0c3d6c1`](cockroachdb/pebble@b0c3d6c1) db: disable writing to value blocks if `DisableSeparationBySuffix` is true * [`562c8d4d`](cockroachdb/pebble@562c8d4d) internal/manifest: document relationship between LastSeqNum and WAL seqnums * [`6d692923`](cockroachdb/pebble@6d692923) github: make sure the AI review bot picks up a PR with label added after creation Release note: none. Epic: none. Co-authored-by: Yahor Yuzefovich <yahor@cockroachlabs.com> Co-authored-by: Annie Pompa <annie@cockroachlabs.com>
|
Build failed (retrying...): |
In 25.1 release we switched to using SQL-instance-based DistSQL physical planning but left gossip-based planning just in case (controlled via a cluster setting). We haven't seen any fallout from that, and given that many different workloads have been upgraded to 25.2, it seems safe to completely remove gossip-based planning now.
Epic: None
Release note: None