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.

Sign up for GitHub

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

Jump to bottom

release-21.2: kvclient: ignore stale lease information from lagging replicas #88738

Merged
merged 2 commits into from
Oct 3, 2022
Merged

release-21.2: kvclient: ignore stale lease information from lagging replicas #88738

merged 2 commits into from
Oct 3, 2022

Conversation

aayushshah15
Copy link
Contributor

Backport:

Please see individual PRs for details.

Release justification: bug fix / stability improvement.

/cc @cockroachdb/release

@aayushshah15 aayushshah15 requested a review from a team as a code owner September 26, 2022 17:12
@blathers-crl
Copy link

blathers-crl bot commented Sep 26, 2022

Thanks for opening a backport.

Please check the backport criteria before merging:

  • Patches should only be created for serious issues or test-only changes.
  • Patches should not break backwards-compatibility.
  • Patches should change as little code as possible.
  • Patches should not change on-disk formats or node communication protocols.
  • Patches should not add new functionality.
  • Patches must not add, edit, or otherwise modify cluster versions; or add version gates.
If some of the basic criteria cannot be satisfied, ensure that the exceptional criteria are satisfied within.
  • There is a high priority need for the functionality that cannot wait until the next release and is difficult to address in another way.
  • The new functionality is additive-only and only runs for clusters which have specifically “opted in” to it (e.g. by a cluster setting).
  • New code is protected by a conditional check that is trivial to verify and ensures that it only runs for opt-in clusters.
  • The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.

Add a brief release justification to the body of your PR to justify this backport.

Some other things to consider:

  • What did we do to ensure that a user that doesn’t know & care about this backport, has no idea that it happened?
  • Will this work in a cluster of mixed patch versions? Did we test that?
  • If a user upgrades a patch version, uses this feature, and then downgrades, what happens?

@cockroach-teamcity
Copy link
Member

This change is Reviewable

@aayushshah15 aayushshah15 force-pushed the backport21.2-73697-76756 branch from 9716eda to ec70797 Compare September 26, 2022 19:14
@aayushshah15
kvclient: ignore stale lease information from lagging replicas
002021b 
This commit makes it such that the `DistSender`'s range descriptor cache
doesn't trigger a cache eviction based on incompatible lease information in a
`NotLeaseHolderError` when it is coming from a replica that has a stale view of
the range's descriptor (characterized by an older `DescriptorGeneration` on the
replica)

Not doing so before was hazardous because, if we received an NLHE that pointed
to a replica that did not belong in the cached descriptor, we'd trigger a cache
evicion. This assumed that the replica returning the error had a fresher view
of the range than what we had in the cache, which is not always true. This
meant that we'd keep doing range lookups and subsequent evictions until this
lagging replica caught up to the current state of the range.

Release note (bug fix): A bug that caused high SQL tail latencies during
background rebalancing in the cluster has been fixed.
@aayushshah15 aayushshah15 force-pushed the backport21.2-73697-76756 branch from ec70797 to f891489 Compare September 26, 2022 19:14
nvanbenschoten
nvanbenschoten approved these changes Sep 27, 2022
View reviewed changes
Copy link
Member

@nvanbenschoten nvanbenschoten left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

:lgtm: thanks!

Reviewed 6 of 6 files at r1, 5 of 5 files at r2, all commit messages.
Reviewable status: :shipit: complete! 1 of 0 LGTMs obtained (waiting on @arulajmani)

@arulajmani @aayushshah15
kvclient: ship range descriptor in NLHE
95c9392 
We recently changed a NLHE to carry the range descriptor generation to
ensure to avoid thrashing the range cache if the replica had a stale
view of the range. In cockroachdb#75742, we saw issues caused by the dist sender
having a stale range descriptor. This patch switches from sending just
the descriptor generation back in NLHE to shipping back the entire
range descriptor. In the future, we may want to solve the issue above
by updating the range cache with the fresher range descriptor thus
skipping a cache eviction and range descriptor lookup.

References cockroachdb#75742

Release note: None
@aayushshah15 aayushshah15 force-pushed the backport21.2-73697-76756 branch from f891489 to 95c9392 Compare September 28, 2022 16:43
@aayushshah15 aayushshah15 merged commit c14b173 into cockroachdb:release-21.2 Oct 3, 2022
@aayushshah15 aayushshah15 deleted the backport21.2-73697-76756 branch October 3, 2022 07:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@nvanbenschoten nvanbenschoten nvanbenschoten approved these changes

@arulajmani arulajmani Awaiting requested review from arulajmani

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@aayushshah15 @cockroach-teamcity @nvanbenschoten @arulajmani