zhang-arvin opened a new pull request, #67308:
URL: https://github.com/apache/doris/pull/67308

   Closes #67303.
   
   ## Problem
   The eraseTable and erasePartition methods in CatalogRecycleBin had an ABA 
race condition. They collected expired IDs under a read lock, then iterated 
over them under a write lock. However, between the read lock release and the 
write lock acquisition, entries could be re-added with the same ID (e.g., after 
a table/partition was restored and then dropped again), causing the write lock 
section to erase the wrong entry.
   
   ## Fix
   - Change the collection phase from List to Map that captures both the ID and 
the recycle time.
   - In the write lock section, re-validate that the recycle time hasn't 
changed and the entry is still expired before erasing.
   - Use get() then remove() instead of remove() directly, to allow the 
re-validation step.
   - Same fix applied to eraseDatabase, eraseTable, and erasePartition.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to