Cron jobs used up 1 million CPU seconds in a few hours

[hewgim]

Describe the bug
On two separate development servers, with pretty much a vanilla build of Magento 2.3.3, both servers used up over a million CPU seconds in a few hours hours after running for a few days. One was running in development mode and one in production mode. There were only 30 products and there was no activity at the time apart from cron.

To Reproduce
I can't give any information to exactly reproduce this. I was more wondering if anyone else had reported this problem. I don't know if it has anything to do with this known bug in Magento's cron: magento/magento2#11002

When the CPU seconds increased there were 2 spikes

The first spike seemed to coincide with the error:
[2019-12-21 01:05:05] main.WARNING: Memory size allocated for the temporary table is more than 20% of innodb_buffer_pool_size. Please update innodb_buffer_pool_size or decrease batch size value (which decreases memory usages for the temporary table). Current batch size: 2500; Allocated memory size: 320000000 bytes; InnoDB buffer pool size: 268435456 bytes. [] []

The second spike seemed to coincide with the error:
In Abstract.php line 144:

SQLSTATE[HY000] [2002] No such file or directory

Expected behavior
Given that the server was not under any load, I would not expect it to use a month's worth of CPU in a few hours

Screenshots

Environment (please complete the following information):

• Cron Job Manager Version 1.8.0
• Magento Version 2.3.3
• PHP Version 7.2.25

Additional context

[Ethan3600]

Hi @hewgim thank you for using the CJM!

I'm very familiar with these symptoms, and I've seen them on production and development environments since Magento 2.1.x. I think this has come up a few times already in the issues board, but it's most definitely a Magento bug.

Hopefully they fix this in 2.4.x. Going to close this issue as it is a Magento bug.

Back when I was working at an agency, Magento provided patches (took them a few patches before they got it right) to fix this bug. I suggest logging a ticket with them.