Hey MariaDB Team,
I've submitted the following bug to MySQL, as I have no MariaDB environment to test it myself, I'm only able to copy / paste here.
They have hidden their bug report.
Please mark as hidden as well if it makes sense for you!
This is a rather complicated problem with crashing the MySQL server, so that mysql_safe has to restart it.
I'm not sure how to classify the problem, so I try to describe it.
Choosing severity S2, because the whole mysqld crashes from a user proc.
Please note: the crash does happen with INNODB being used as engine for a temporary table.
A function is called
- the function calls a procedure to generate some cache
- the procedure creates a temporary table (if not exists)
- table is getting cleaned of outdated rows
- table is checked for cache data (for the arguments supplied)
- cache is generated
- the generated cache is queried and summarized to some extend
- function returns a value
Full stacktrace can be supplied when needed.
The proc and function is derived from a SLA calculation tool.
The crash does NOT happen when:
- ENGINE=MEMORY is used
- the table is not created as a temporary one