Details
-
Bug
-
Status: Closed (View Workflow)
-
Minor
-
Resolution: Won't Fix
-
10.0.24, 5.5(EOL), 10.0(EOL), 10.1(EOL)
-
CentOS 6.6 x86_64
Description
Our development team had run a query that uses the complex view.
When the query was executed by a single thread, it was no problem. (it took about 90 seconds.)
When the query was executed by multi threads, it had cause DB hang.
It did never-ending and the other threads could not even run a simple query like "desc table_name".
In the processlist, thread status was "Copyint to tmp table"
We performed test by changing parameter value in my.ncf.
We found that the value of aria_pagecahce_buffer_size is problem.
In my.cnf , the value of "aria_pagecahce_buffer_size" was 1M.
We changed the value from 1M to 128M and the query was executed without DB hang.
I think the DB hang occurs when the following conditions combined.
- A small size of "aria_pagecahce_buffer_size" .
- At the same time to run a query that uses a large internal temporary table from multiple threads.
What is this cause to have problem?
I attach a script to create a test environment.
Thanks.