[DocDB] Intents are populated back after restart and not cleaned up #16537
Labels
area/docdb
YugabyteDB core features
kind/bug
This issue is a bug
priority/medium
Medium priority issue
Jira Link: DB-5931
Description
On restart, we replay the transactions that were not flushed and also try to clean them up if we find that they were already applied or if we applied it again. While cleaning the intent, cleanup task might fail due to race conditions. Basically, we initialise the thread pool for clean up tasks after bootstrap.
So, the transaction metadata is cleaned up but the intents do remain in IntentDB and they only get cleaned up when compaction takes place as we have some optimisations where we delete the intent files as well.
I think this could have 'intermittent' performance issue
restart the cluster
This also causes perf issue.
If we flush all tablets and then restart the intent count goes back to 0.
./build/latest/bin/yb-ts-cli flush_all_tablets
Restart the cluster
Warning: Please confirm that this issue does not contain any sensitive information
The text was updated successfully, but these errors were encountered: