I have 12 node proceses across 2 servers that, at 2:39 this morning, simultaneously stopped calling epoll_wait and clock_gettime