Fix volatile result sent before it is fully committed #2598
+146
−21
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
⚪
|
|
⚪
|
azevaykin
approved these changes
Mar 11, 2024
snaury
added a commit
to snaury/ydb
that referenced
this pull request
Mar 11, 2024
This was referenced Mar 20, 2024
Merged
Merged
This was referenced Mar 29, 2024
Merged
This was referenced Apr 6, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Changelog entry
Fix volatile result sent before it is fully committed.
Changelog category
Additional information
A rare failure was detected with Jepsen when volatile transactions feature is enabled and used. Investigation showed that YDB would sometimes reply SUCCESS to a commit that has actually failed. The underlying issue is that since volatile transactions are prepared in-memory, a PREPARED reply is sent very early in the pipeline, before the (readonly) propose transaction is committed (which is expected, volatile transactions don't have storage in their critical path until they are executed). When storage is a bit slow the volatile transaction may be fast enough to plan and execute while a previous localdb transaction is still committing. When propose localdb transaction finally commits it would erroneously observe the new operation result (which optimistically has SUCCESS while it is waiting for localdb commit and other participants), and would send it thinking this is a propose phase result. The localdb commit with execution side effects may fail however, but the SUCCESS result has already been sent by that time. This is fixed by marking operations that send propose result early, and ignoring operation result on propose completion in such operations.
Fixes KIKIMR-21156.