Skip to content

Conversation

@ankitsultana
Copy link
Contributor

@ankitsultana ankitsultana commented Feb 10, 2024

See issue for explanation of the issue: #12398

This PR fixes the issue by de-duplicating the records before the map is updated. This adds a small cost in terms of memory, but it should be quite low (e.g. for 1M records, with 32 byte primary-keys, we will have 32MB + memory address and other Record info parameter costs).

There's a small latency cost also added due to the extra dedup step. But we think it's quite low compared to the rest of operations so shouldn't be too big an issue (e.g. after dedup the ConcurrentHashMap is updated for all records, which can contend with Realtime ingestion and other addOrReplaceSegment calls).

We only do this for Partial Upsert tables right now since Full Upsert tables don't suffer from the issue mentioned in #12398.

Test Plan: Added a unit-test.

@codecov-commenter
Copy link

codecov-commenter commented Feb 11, 2024

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (5c81c0a) 61.56% compared to head (563fba0) 61.63%.
Report is 24 commits behind head on master.

Files Patch % Lines
...cal/upsert/BasePartitionUpsertMetadataManager.java 84.61% 1 Missing and 1 partial ⚠️
...t/ConcurrentMapPartitionUpsertMetadataManager.java 0.00% 1 Missing and 1 partial ⚠️
Additional details and impacted files
@@             Coverage Diff              @@
##             master   #12395      +/-   ##
============================================
+ Coverage     61.56%   61.63%   +0.06%     
- Complexity      201      207       +6     
============================================
  Files          2426     2436      +10     
  Lines        132709   133117     +408     
  Branches      20519    20620     +101     
============================================
+ Hits          81706    82050     +344     
- Misses        44994    45027      +33     
- Partials       6009     6040      +31     
Flag Coverage Δ
custom-integration1 <0.01% <0.00%> (?)
integration <0.01% <0.00%> (+<0.01%) ⬆️
integration1 <0.01% <0.00%> (?)
integration2 0.00% <0.00%> (ø)
java-11 27.73% <73.33%> (+27.73%) ⬆️
java-21 34.89% <0.00%> (-26.68%) ⬇️
skip-bytebuffers-false 61.62% <73.33%> (+0.07%) ⬆️
skip-bytebuffers-true 34.85% <0.00%> (+0.02%) ⬆️
temurin 61.63% <73.33%> (+0.06%) ⬆️
unittests 61.63% <73.33%> (+0.06%) ⬆️
unittests1 46.73% <0.00%> (+<0.01%) ⬆️
unittests2 27.73% <73.33%> (+0.03%) ⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@ankitsultana ankitsultana changed the title [Do Not Merge] Fix Bug in Handling Equal Comparison Column Values in Upsert Fix Bug in Handling Equal Comparison Column Values in Upsert Feb 16, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants