Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: Communications link failure 偶发断链
[TiDB Usage Environment] Production Environment
[TiDB Version] 6.5.1 and 6.5.0
[Reproduction Path] Very low reproduction rate, occurs once every 1-3 days, adjusted timeout parameters, adjusted jdbcurl parameters
[Encountered Problem: Phenomenon and Impact]
The executed statement is “replace into … select” usage, at the same time point, the executed insert has no issues.
Using dolphinscheduler to submit SQL statements, through underlying F5 load balancing.
Disconnection occurs within 15 seconds, sometimes 1 second, sometimes around 10 seconds. Not caused by OOM.
[Resource Configuration]
[Attachments: Screenshots/Logs/Monitoring]
JDBC URL parameters: keepAlive=true&rewriteBatchedStatements=true&useConfigs=maxPerformance&failOverReadOnly=false&allowMultiQueries=true&wait_timeout=31536000&autoReconnect=true&maxReconnects=10&useServerPrepStmts=true&useSSL=false&useUnicode=true&prepStmtCacheSqlLimit=2048000&cachePrepStmts=true&characterEncoding=utf-8&useCursorFetch=true&prepStmtCacheSize=1000
Did the SQL execute? Did TiDB report any errors? What is the status of the DolphinScheduler task?
I encountered the same situation. How did you solve it?
When executing multiple truncate table operations, there are occasional Communications link failure errors. Checking the log, there are no OOM-related logs. How do the experts solve this situation?
It seems to be a protocol issue when the application interacts with TiDB. I suggest trying a different application.
Please provide the connection pool configuration.