[fix](load_stream) Fix flush token deadlock by ensuring wait_for_flush_tasks is called before destruction #60284
Conversation
This file contains hidden or 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
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
liaoxin01
added a commit
to liaoxin01/doris
that referenced
this pull request
Jan 27, 2026
…ait_for_flush_tasks is called before destruction apache#60284 Problem: When TabletStream is destroyed without pre_close() being called (e.g., on_idle_timeout scenario), the _flush_token destructor calls shutdown() which triggers deadlock detection if called from the pool thread. Root cause: - on_idle_timeout() directly calls brpc::StreamClose() without calling LoadStream::close() - This triggers the destruction chain without calling pre_close() on TabletStreams - If flush tasks are still running, TabletStream may be destroyed in pool thread Solution: - Add IndexStream::~IndexStream() to ensure wait_for_flush_tasks() is called on all TabletStreams - Add TabletStream::wait_for_flush_tasks() to wait for all flush tasks to complete - This ensures _flush_token is properly handled before TabletStream destruction - Revert commit c18ef17 (shared_from_this) as it is no longer needed
Contributor
Author
|
run buildall |
gavinchou
previously approved these changes
Jan 27, 2026
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
Contributor
|
PR approved by anyone and no changes requested. |
There was a problem hiding this comment.
Pull request overview
This PR fixes a deadlock issue that occurs when TabletStream is destroyed without pre_close() being called (e.g., in the on_idle_timeout scenario). The deadlock happened because the flush token's destructor calls shutdown(), which performs deadlock detection and throws if called from the pool thread itself.
Changes:
- Reverts PR #60148's use of
shared_from_this()in async lambdas, replacing with rawthispointer - Adds
TabletStream::wait_for_flush_tasks()to ensure flush tasks complete before destruction - Adds
IndexStream::~IndexStream()to callwait_for_flush_tasks()on all tablet streams - Refactors
TabletStream::pre_close()to call the newwait_for_flush_tasks()method
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| be/src/runtime/load_stream.h | Removes std::enable_shared_from_this from TabletStream, adds wait_for_flush_tasks() method and _flush_tasks_done flag, adds IndexStream destructor |
| be/src/runtime/load_stream.cpp | Implements wait_for_flush_tasks() and IndexStream destructor, reverts lambda captures from shared_from_this() to raw this, refactors pre_close() |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
liaoxin01
force-pushed
the
fix-load-stream-flush-token-deadlock-master
branch
from
86e21d6 to
c36ec0e
Compare
liaoxin01
added a commit
to liaoxin01/doris
that referenced
this pull request
Jan 27, 2026
…ait_for_flush_tasks is called before destruction apache#60284 Problem: When TabletStream is destroyed without pre_close() being called (e.g., on_idle_timeout scenario), the _flush_token destructor calls shutdown() which triggers deadlock detection if called from the pool thread. Root cause: - on_idle_timeout() directly calls brpc::StreamClose() without calling LoadStream::close() - This triggers the destruction chain without calling pre_close() on TabletStreams - If flush tasks are still running, TabletStream may be destroyed in pool thread Solution: - Add IndexStream::~IndexStream() to ensure wait_for_flush_tasks() is called on all TabletStreams - Add TabletStream::wait_for_flush_tasks() to wait for all flush tasks to complete - This ensures _flush_token is properly handled before TabletStream destruction - Revert commit c18ef17 (shared_from_this) as it is no longer needed
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 32864 ms |
github-actions
bot
removed
the
approved
TPC-H: Total hot run time: 32400 ms |
ClickBench: Total hot run time: 28.29 s |
Contributor
BE UT Coverage ReportIncrement line coverage Increment coverage report
|
Contributor
BE Regression && UT Coverage ReportIncrement line coverage Increment coverage report
|
…h_tasks is called before destruction Problem: When TabletStream is destroyed without pre_close() being called (e.g., on_idle_timeout scenario), the _flush_token destructor calls shutdown() which triggers deadlock detection if called from the pool thread. Root cause: - on_idle_timeout() directly calls brpc::StreamClose() without calling LoadStream::close() - This triggers the destruction chain without calling pre_close() on TabletStreams - If flush tasks are still running, TabletStream may be destroyed in pool thread Solution: - Add IndexStream::~IndexStream() to ensure wait_for_flush_tasks() is called on all TabletStreams - Add TabletStream::wait_for_flush_tasks() to wait for all flush tasks to complete - This ensures _flush_token is properly handled before TabletStream destruction - Revert commit c18ef17 (shared_from_this) as it is no longer needed
liaoxin01
force-pushed
the
fix-load-stream-flush-token-deadlock-master
branch
from
c36ec0e to
cc77be5
Compare
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 33039 ms |
ClickBench: Total hot run time: 28.37 s |
Contributor
BE UT Coverage ReportIncrement line coverage Increment coverage report
|
Contributor
BE Regression && UT Coverage ReportIncrement line coverage Increment coverage report
|
Contributor
BE Regression && UT Coverage ReportIncrement line coverage Increment coverage report
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
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.
What problem does this PR solve?
Issue Number: close #xxx
Related PR: #60148
Problem Summary:
Problem:
When TabletStream is destroyed without pre_close() being called (e.g., on_idle_timeout scenario), the _flush_token destructor calls shutdown() which triggers deadlock detection if called from the pool thread.
Root cause:
Solution:
Release note
None
Check List (For Author)
Test
Behavior changed:
Does this need documentation?
Check List (For Reviewer who merge this PR)