我正在使用带有 RF=3 的副本集的 MongoDB。文档中有一个事件“ha”,我看到它对更新操作的影响是巨大的:
[Wed Apr 29 2015 09:00:13 GMT+0300 (MSK)] [pid 7251] update took 0ms for uuid 08ecd93346daa1b8b6fef1772cea0469
[Wed Apr 29 2015 09:00:13 GMT+0300 (MSK)] [pid 7251] update took 1ms for uuid ea9a71aa09f0a8c3a169dcbb895e023d
[Wed Apr 29 2015 09:00:13 GMT+0300 (MSK)] [pid 7251] update took 1ms for uuid c4b51c3c1e7c18ab9c4f1e410af26a1b
[Wed Apr 29 2015 09:00:14 GMT+0300 (MSK)] [pid 7251] [event ha] ["start",{"norepeat":false,"id":8,"state":{"primary":"...","secondaries":["...", "..."]}}]
[Wed Apr 29 2015 09:00:14 GMT+0300 (MSK)] [pid 7251] [event ha_connect] ["start",{"norepeat":false,"id":8,"state":{"primary":"...","secondaries":["...", "..."]}}]
[Wed Apr 29 2015 09:00:15 GMT+0300 (MSK)] [pid 7251] update took 2124ms for uuid 07de843901cf5bc107f11e9965204920
[Wed Apr 29 2015 09:00:15 GMT+0300 (MSK)] [pid 7251] update took 2122ms for uuid 021c149b4a5272ef7b49296667484e1b
[Wed Apr 29 2015 09:00:15 GMT+0300 (MSK)] [pid 7251] update took 2122ms for uuid d5a7cfcc47aa5c238d33f2257edab674
据我所知,这个事件来自 mongodb-core。但是这个事件的意义是什么?副本集连接是否每 N 秒丢失并恢复?