Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Incomplete
-
10.1.24, 10.2.8
-
CentOS 6.5 and 7.3
Description
I recently upgraded my servers from 5.5.34 to 10x. For my hosts running CentOS 6x I went with 10.1.24 and hosts running CentOS 7x I went with 10.2.8.
I have noticed a slow down in certain innodb operations on both 10.1.24 and 10.2.8.
For example, I have a replication master that is running 5.5.34 and a delete operation took 20 seconds. When it replicated to the 10.2.8 slave, it took 70 seconds. Some additional details:
• Both servers had been rebooted and all caches flushed.
• No other connection activity was happening, so not a potential concurrency issue.
• Both CentOS 7.3 configured the same (swapiness = 1, numa-interleaving enabled, etc). • Hardware is identical
• my.cnf(s) are the same except on the 10.2.8 slave commented out: innodb_additional_mem_pool_size – thread_cache_size – thread_concurrency
Are there any config settings in 10x I can tweak that address this kind of performance issue?