How have I fixed «InnoDB: Semaphore wait has lasted > 600 seconds. We intentionally crash the server because it appears to be hung.»

MySQL
1

These 3 settings fixed the issue. I do not know which one exactly.

innodb_flush_neighbors = 0
thread_cache_size = 100
table_open_cache = 50000

bugs.launchpad.net/percona-server/+bug/1653764/comments/15

2

Today this issue occured again, and I have fixed it by commenting the thread_cache_size and table_open_cache options out.