[core] Fix race condition for earliest snapshot #4930
+205
−46
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
2 tasks
wwj6591812
reviewed
Jan 17, 2025
| Long earliest; | ||
| if (startFromChangelog) { | ||
| Long earliestChangelog = earliestLongLivedChangelogId(); | ||
| earliest = earliestChangelog == null ? earliestSnapshot : earliestChangelog; | ||
| earliest = earliestChangelog == null ? earliestSnapshotId : earliestChangelog; |
| return null; | ||
| } | ||
|
|
||
| Long latest = null; |
| try { | ||
| earliestSnapshot = tryGetChangelogOrSnapshot(earliest); | ||
| break; | ||
| } catch (FileNotFoundException e) { |
| do { | ||
| try { | ||
| return tryGetSnapshot(snapshotId); | ||
| } catch (FileNotFoundException e) { |
| @@ -163,11 +163,28 @@ public Snapshot tryGetSnapshot(long snapshotId) throws FileNotFoundException { | |||
| return snapshot; | |||
| } | |||
|
|
|||
| private @Nullable Snapshot tryGetEarliestSnapshotLaterThanOrEqualTo( | |||
| long snapshotId, long stopId) { | |||
JingsongLi
reviewed
Feb 11, 2025
| Long earliestWatermark = null; | ||
| // find the first snapshot with watermark | ||
| if ((earliestWatermark = snapshot(earliest).watermark()) == null) { | ||
| if ((earliestWatermark = earliestSnapShot.watermark()) == null) { | ||
| while (earliest < latest) { |
There was a problem hiding this comment.
I think there are some risks. Earliest should be re-assigned.
Maybe you should just optimize earliestSnapshot. And use this method in many places.
There was a problem hiding this comment.
Given that some earliestSnapshot methods use binary search while others use linear, I extracted and optimized two kinds of paths. Please take a look.
yunfengzhou-hub
force-pushed
the
snapshot-earliest-race
branch
2 times, most recently
from
b4ad65b to
86fb5b5
Compare
yunfengzhou-hub
force-pushed
the
snapshot-earliest-race
branch
from
86fb5b5 to
d8e82fe
Compare
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.
Purpose
Linked issue: close #3479
This PR fixes a racing bug about earliest snapshot. When thread A is trying to get the earliest snapshot of a table, while the snapshot is being expired by thread B, thread A might throw exception that it cannot get the snapshot information from the earliest snapshot id.
This PR fixes the bug by adding retrying logic to seek the second earliest snapshot if the first one cannot be found.
Note that, theoretically speaking, any snapshot (including the latest one) could encounter this issue, but the probability is quite small, so this PR only fixes the bug for invocations on the earliest snapshot.
Tests
Tests in SnapshotManagerTest are used to verify against the changes.
API and Format
This change does not affect API or storage format.
Documentation
This change does not introduce a new feature.