Re: InnoDB: Warning: a long semaphore wait:. X-lock on RW-latch

[Rick James]

SET timestamp is a way to get a predictable NOW(). The slow query follows it.

Is the index on orders.id UNIQUE (or PRIMARY KEY)? Ditto for sessions.id? Or, at least, is there only one row with the id in question?

Did you kill any threads? I'm fishing for a JOIN between one of these tables and some other table, such that there is an index missing somewhere and it tied up all the rows, thereby blocking the UPDATE. But then it is puzzling that the SELECT would be blocked for 7 minutes.

Sorry, I don't have a quick answer.