There is a story behind this post. To make it short, the detailed
study presented here started when I noted old Bug
#72748 in "No feedback" status few days ago. I considered
this status totally wrong and started a discussion on Facebook about it (yes, this is what I usually do,
I discuss handling of MySQL bugs in public).
The technical topic of this discussion (blaming aside) was the way AUTO-INC locks are set by InnoDB with default innodb_autoinc_lock_mode=1. …