@carehart, we’ve managed to keep things under control. Adding the -XX:MaxMetaspaceSize helped because we’re able to allocate adequate memory to it (7GB). Looks like specifically allocating a specific (and adequate) metaspace memory helped.
@carehart The solution for us was to use servers with more memory and restart servers periodically.
The server restart is also the only thing working for me. I’m running 184.108.40.206.