config服务之间数据有差异

刚开始研发说mongo很慢，去看了一下mongo集群的状态，确实有两个mongo的主节点变成了recover状态，而SECONDARY 节点变成了主节点，查看了一下两个mongo节点日志的情况，发现SECONDARY直接变成了主，随后对主节点发出了stepdown的命令，主节点开始回滚事物，在此情况下，mongo数据库是坏了吗，我还在判断中，

一下子两个节点变成了recover状态确实问题很大，需要重新初始化数据吗？如果重新初始化，数据量很大。 

       还有从刚开始登陆操作系统，执行操作系统命令，发现很慢，看了top命令，发现内存不够，也许是mongo压力太大，直接就到副本集上去了！

     虽然两个mongo节点坏了，但是有secondary顶上去当了主节点，不至于查询操作很慢，最后决定，先重启一下mongo节点，重启完成后，发现mongo副本集的状态恢复正常，mongo主节点也恢复正常，在启动路由节点的时候失败了，去看日志，发现，报了以下错误:

2017-01-10T17:03:08.448+0800 I CONTROL  [mongosMain] options: { net: { bindIp: "1.1130.38", port: 62000 }, processManagement: { fork: true, pidFilePath: "/data/mongo2/pid/pid_ms.txt" }, sharding: { configDB: "1.1.130.36:42000,1.1.13.37:42000,1.1.13.38:42000" }, systemLog: { destination: "file", path: "/data/mongo2/log/log_ms.txt" } }

2017-01-10T17:03:10.218+0800 W SHARDING [mongosMain] config servers 1.1.13.36:42000 and 1.1.13.37:42000 differ
2017-01-10T17:03:10.219+0800 W SHARDING [mongosMain] config servers 1.1.13.36:42000 and 1.1.13.37:42000 differ
2017-01-10T17:03:10.220+0800 W SHARDING [mongosMain] config servers 1.1.13.36:42000 and 1.1.13.37:42000 differ
2017-01-10T17:03:10.221+0800 W SHARDING [mongosMain] config servers 1.1.13.36:42000 and 1.1.13.37:42000 differ
2017-01-10T17:03:10.221+0800 E SHARDING [mongosMain] could not verify that config servers are in sync :: caused by :: config servers 1.1.13.36:42000 and 1.1.13.37:42000 differ: { chunks: "72d01257b2a9a1f8f1045e1187e5a84c", collections: "6500f97a1a972fac8667f18129fa2fda", databases: "e42742394ff821ebb840395597398b44", shards: "84e83b83c34d5a8b0b00d2fd0dbcbd16", version: "bef8a99e420fc1bf6171016741a3bbe1" } vs { chunks: "e477af06af12d9dd35caba53942e33ff", collections: "6500f97a1a972fac8667f18129fa2fda", databases: "e42742394ff821ebb840395597398b44", shards: "84e83b83c34d5a8b0b00d2fd0dbcbd16", version: "bef8a99e420fc1bf6171016741a3bbe1" }
2017-01-10T17:03:10.221+0800 I -        [mongosMain] configServer connection startup check failed



解决办法：
原因：config服务之间数据有差异
解决：停掉所有的config服务，手动将期中一个config的数据拷到另一个中，再启动，问题应该可以解决了