MySQL8 instance crashes on CloudSQL

I had a MySQL 5.7 instance on Google CloudSQL. Now i created a MySQL 8 instance.

The configuration is pretty much the same (except i use 2 CPUs instead of one, 3.75GB of RAM). Now the default config for MySQL Memory usage (like innodb_buffer_pool_size,..) seems to be the same.

I migrated about half of my applications to use this instance. What is happening now is, that the instance Memory usage goes above 3.XX GB and the service gets restarted. Which is super annoying because in that time my applications obviously crash. It seems like memory usage grows with every select statement, and everything is cached.

Here are some of the config values:

|                          key_buffer_size |           8.00  MB | |                  innodb_buffer_pool_size |        1408.00  MB | |                   innodb_log_buffer_size |          16.00  MB | |                         sort_buffer_size |           0.25  MB | |                         read_buffer_size |           0.125 MB | |                     read_rnd_buffer_size |           0.25  MB | |                         join_buffer_size |           0.25  MB | |                             thread_stack |           0.273 MB | |                        binlog_cache_size |           0.031 MB | |                           tmp_table_size |          16.00  MB | 

This makes CloudSQL pretty much unusable to me. I need MySQL 8 without crashing several times a day.