I am seeing quite a lot of the following errors in the alert log.
2017-06-27 22:31:14 42732 [ERROR] Failed to write to mysql.slow_log: Thread stack overrun: 116582671296 bytes used of a 262144 byte stack, and 81920 bytes needed. Use 'mysqld --thread_stack=#' to specify a bigger stack.
2017-06-27 22:31:14 42732 [ERROR] Failed to write to mysql.slow_log: Thread stack overrun: -8879168 bytes used of a 262144 byte stack, and 81920 bytes needed. Use 'mysqld --thread_stack=#' to specify a bigger stack.
* This error only appears in on production DB instance.
* This is an Amazon Aurora (5.6.10) Mysql instance configured as an 8xlarge (32 vCPU, 244G RAM).
1. The documentation on this parameter (thread_stack) is sparse and only mentions default values for 32 vs 64 bit hosts. This is set to the default of 256G [1]
2. I can find only a few references in blogs and forums that discuss this parameter. Most just recommend setting to the default. There is no discussion on effects (negative or positive) about adjusting this value.
3. There is absolutely no mention about a negative thread stack overrun. What does that mean?
4. Other then giving each thread/connection use a bit more memory, what other things should I consider when increasing this parameter?
5. should I increase just the 80K recommended in the error? or increase to something a bit larger?
6. the default is 256KB but the overrun above is 108.5GB but is asking for an 80K increase. I'm confused.
Any ideas?
Thanks
Eric
[1]
https://dev.mysql.com/doc/refman/5.6/en/server-system-variables.html#sysvar_thread_stack