听了mjj的建议把apache handler换成了fpm+nginx, 内存是下去了，cpu又干上来了。这正常吗？ 不过oom崩数据库倒是没有了。 pm.max_children = 10 pm.start_servers = 4 pm.min_spare_servers = 2 pm.max_spare_servers = 4 pm.max_spawn_rate = 32 pm.process_idle_timeout = 10s; pm.max_requests = 100 15:39:41 up 22:33, 6 users, load average: 9.66, 7.07, 6.53 内存, 听了, 换成了
