1

当我的 Primary 的 mongo shell 突然变成 Secondary 时，我正在练习一些 MapReduce。我与其他辅助服务器通过 SSH 连接到另外两个虚拟机，发现 mongod 已无法运行。我杀了他们，我发出了mongod --config /etc/mongod.conf把他们踢开的命令，然后我进入了mongo外壳。几秒钟后，他们被打断了：

2014-09-14T22:29:54.142-0500 DBClientCursor::init call() failed
2014-09-14T22:29:54.143-0500 trying reconnect to 127.0.0.1:27017 (127.0.0.1) failed
2014-09-14T22:29:54.143-0500 warning: Failed to connect to 127.0.0.1:27017, reason: errno:111 Connection refused
2014-09-14T22:29:54.143-0500 reconnect 127.0.0.1:27017 (127.0.0.1) failed failed couldn't connect to server 127.0.0.1:27017 (127.0.0.1), connection attempt failed
>

这是来自他们的（副本集中的两个原始辅助节点）日志：

2014-09-14T22:09:21.879-0500 [rsBackgroundSync] replSet syncing to: vm-billing-001:27017
2014-09-14T22:09:21.880-0500 [rsSync] replSet still syncing, not yet to minValid optime 54165090:1
2014-09-14T22:09:21.882-0500 [rsBackgroundSync] replset setting syncSourceFeedback to vm-billing-001:27017
2014-09-14T22:09:21.886-0500 [rsSync] replSet SECONDARY
2014-09-14T22:09:21.886-0500 [repl writer worker 1] build index on: test.tmp.mr.CCS.nonconforming_1_inc properties: { v: 1, key: { 0: 1 }, name: "_temp_0", ns: "test.tmp.mr.CCS.nonconforming_1_inc" }
2014-09-14T22:09:21.887-0500 [repl writer worker 1]      added index to empty collection
2014-09-14T22:09:21.887-0500 [repl writer worker 1] build index on: test.tmp.mr.CCS.nonconforming_1 properties: { v: 1, key: { _id: 1 }, name: "_id_", ns: "test.tmp.mr.CCS.nonconforming_1" }
2014-09-14T22:09:21.887-0500 [repl writer worker 1]      added index to empty collection
2014-09-14T22:09:21.888-0500 [repl writer worker 1] build index on: test.tmp.mr.CCS.nonconforming_1 properties: { v: 1, unique: true, key: { id: 1.0 }, name: "id_1", ns: "test.tmp.mr.CCS.nonconforming_1" }
2014-09-14T22:09:21.888-0500 [repl writer worker 1]      added index to empty collection
2014-09-14T22:09:21.891-0500 [repl writer worker 2] ERROR: writer worker caught exception:  :: caused by :: 11000 insertDocument :: caused by :: 11000 E11000 duplicate key error index: cisco.tmp.mr.CCS.nonconforming_1.$id_1  dup key: { : null } on: { ts: Timestamp 1410748561000|46, h: 9014687153249982311, v: 2, op: "i", ns: "cisco.tmp.mr.CCS.nonconforming_1", o: { _id: 14, value: 1.0 } }
2014-09-14T22:09:21.891-0500 [repl writer worker 2] Fatal Assertion 16360
2014-09-14T22:09:21.891-0500 [repl writer worker 2]

我可以mongo --host ... --port ...从两个无法启动 mongo 的 VM 向原始主 mongo 发出问题，但我确实在错误日志中看到了上面的一些连接被拒绝注释。

我原来的主mongod仍然可以在mongoshell中连接，但它是一个主。我可以杀死它并重新启动它，它将在二级启动。

如何回滚到最后一个已知状态并重新启动我的副本集？

4

0 回答 0