何小伟-向Mongodb集群添加新的成员节点.doc

向Mongodb集群添加新的成员节点

向集群添加新节点有两种方法：
方 法

第一种方法
拷贝其他节点的数据文件到新的节点
第二种方法
向已有的replica set 添加新成员
一．第一种方法(拷贝其他节点的数据文件到新的节点)
我们在启动mongod服务之前一定要拷贝数据文件到本机。下面是空库添加到一个replica set中的日志记录：
第一步：#### 连接 Primary库
Tue Nov 1 14:22:57 [initandlisten] connection accepted from 10.250.7.220:54235 #2
Tue Nov 1 14:22:57 [initandlisten] connection accepted from 10.250.7.220:54236 #3
Tue Nov 1 14:22:57 [initandlisten] connection accepted from 10.250.7.220:54237 #4
Tue Nov 1 14:22:57 [rsStart] trying to contact 10.250.7.220:27018
Tue Nov 1 14:22:57 [rsStart] trying to contact 10.250.7.220:27019
Tue Nov 1 14:22:57 [rsStart] trying to contact 10.250.7.220:27020
第二步：####从primary库复制replica set的配置信息，并保存到本地 -dbpath 指定的文件目录####
Tue Nov 1 14:22:57 [rsStart] replSet got config version 2 from a remote, saving locally
Tue Nov 1 14:22:57 [rsStart] replSet info saving a newer config version to local.system.replset
Tue Nov 1 14:22:57 [FileAllocator] allocating new datafile /opt/mongodata/r1/local.ns, filling with zeroes...
Tue Nov 1 14:22:57 [FileAllocator] creating directory /opt/mongodata/r1/_tmp
Tue Nov 1 14:22:57 [FileAllocator] done allocating datafile /opt/mongodata/r1/local.ns, size: 16MB, took 0.118 secs
Tue Nov 1 14:22:57 [FileAllocator] allocating new datafile /opt/mongodata/r1/local.0, filling with zeroes...
Tue Nov 1 14:23:01 [FileAllocator] done allocating datafile /opt/mongodata/r1/local.0, size: 64MB, took 3.807 secs
Tue Nov 1 14:23:01 [FileAllocator] allocating new datafile /opt/mongodata/r1/local.1, filling with zeroes...
Tue Nov 1 14:23:02 [rsStart] replSet saveConfigLocally done
Tue Nov 1 14:23:02 [FileAllocator] done allocating datafile /opt/mongodata/r1/local.1, size: 128MB, took 1.068 secs
Tue Nov 1 14:23:02 [rsStart] replSet STARTUP2
Tue Nov 1 14:23:02 [rsMgr] replSet total number of votes is even - add arbiter or give one member an extra vote
Tue Nov 1 14:23:02 [rsSync] ******
第三步：####创建oplog 日志文件####
Tue Nov 1 14:23:02 [rsSync] creating replication oplog of size: 944MB...
Tue Nov 1 14:23:02 [FileAllocator] allocating new datafile /opt/mongodata/r1/local.2, filling with zeroes...
Tue Nov 1 14:23:04 [rsHealthPoll] replSet info member 10.250.7.220:27018 is up
Tue Nov 1 14:23:04 [rsHealthPoll] replSet member 10.250.7.220:27018 is now in state SECONDARY
Tue Nov 1 14:23:04 [rsHealthPoll] replSet info member 10.250.7.220:27019 is up
Tue Nov 1 14:23:04 [rsHealthPoll] replSet member 10.250.7.220:27019 is now in state SECONDARY
Tue Nov 1 14:23:04 [rsHealthPoll] replSet info member 10.250.7.220:27020 is up
Tue Nov 1 14:23:04 [rsHealthPoll] replSet member 10.250.7.220:27020 is now in state PRIMARY
Tue Nov 1 14:23:49 [FileAllocator] done allocating datafile /opt/mongodata/r1/local.2, size: 1024MB, took 46.28 secs
Tue Nov 1 14:23:50 [rsSync] ******
第四步：####应用主库的日志，复制数据文件####
Tue Nov 1 14:23:50 [rsSync] replSet initial sync pending
Tue Nov 1 14:23:50 [rsSync] replSet syncing to: 10.250.7.220:27020
Tue Nov 1 14:23:50 [rsSync] build index local.me { _id: 1 }
Tue Nov 1 14:23:50 [rsSync] build index done 0 records 0.003 secs
Tue Nov 1 14:23:50 [rsSync] replSet initial sync drop all databases
Tue Nov 1 14:23:50 [rsSync] dropAllDatabasesExceptLocal 1
Tue Nov 1 14:23:50 [rsSync] replSet initial sync clone all databases
Tue Nov 1 14:23:50 [rsSync] replSet initial sync cloning db: test
Tue Nov 1 14:23:50 [FileAllocator] allocating new datafile /opt/mongodata/r1/test.ns, filling with zeroes...
Tue Nov 1 14:23:51 [FileAllocator] done allocating datafile /opt/mongodata/r1/test.ns, size: 16MB, took 0.8 secs
Tue Nov 1 14:23:51 [FileAllocator] allocating new datafile /opt/mongodata/r1/test.0, filling with zeroes...
Tue Nov 1 14:23:55 [FileAllocator] done allocating datafile /opt/mongodata/r1/test.0, size: 64MB, took 3.643 secs
Tue Nov 1 14:23:55 [FileAllocator] allocating new datafile /opt/mongodata/r1/test.1, filling with zeroes...

Tue Nov 1 14:23:55 [rsSync] build index test.yql { _id: 1 }
Tue Nov 1 14:23:56 [rsSync] build index done 1 records 0.006 secs
Tue Nov 1 14:23:56 [rsSync] replSet initial sync query minValid
Tue Nov 1 14:23:56 [rsSync] replSet initial oplog application from 10.250.7.220:27020 starting at Nov 1 14:14:05:1 to Nov 1 14:14:05:1
Tue Nov 1 14:23:57 [rsSync] replSet initial sync finishing up
Tue Nov 1 14:23:57 [rsSync] replSet set minValid=4eaf8e2d:1
Tue Nov 1 14:23:57 [rsSync] build index local.replset.minvalid { _id: 1 }
Tue Nov 1 14:23:57 [rsSync] build index done 0 records 0.023 secs
Tue Nov 1 14:23:57 [rsSync] replSet initial sync done
Tue Nov 1 14:23:58 [rsSync] replSet syncing to: 10.250.7.220:27020
Tue Nov 1 14:23:58 [rsSync] replSet SECONDARY
Tue Nov 1 14:24:02 [FileAllocator] done allocating datafile /opt/mongodata/r1/test.1, size: 128MB, took 7.428 secs
Tue Nov 1 14:24:16 [clientcursormon] mem (MB) res:16 virt:2848 mapped:1312
Tue Nov 1 14:24:42 [initandlisten] connection accepted from 127.0.0.1:21141 #5
Tue Nov 1 14:29:16 [clientcursormon] mem (MB) res:16 virt:2849 mapped:1312
Tue Nov 1 14:34:16 [clientcursormon] mem (MB) res:16 virt:2913 mapped:1312
Tue Nov 1 14:39:16 [clientcursormon] mem (MB) res:16 virt:2913 mapped:1312
第五步：===下面是删除节点以后的日志==
Tue Nov 1 14:39:28 [conn2] end connection 10.250.7.220:54235
Tue Nov 1 14:39:29 [rsMgr] replset msgReceivedNewConfig version: version: 3
Tue Nov 1 14:39:29 [rsMgr] replSet info saving a newer config version to local.system.replset
Tue Nov 1 14:39:29 [rsMgr] replSet saveConfigLocally done
Tue Nov 1 14:39:29 [rsMgr] replSet error self not present in the repl set configuration:
Tue Nov 1 14:39:29 [rsMgr] { _id: "myset", version: 3, members: [ { _id: 0, host: "10.250.7.220:27018" }, { _id: 1, host: "10.250.7.220:27019" }, { _id: 2, host: "10.
250.7.220:27020" } ] }
-------------------------------------------------------------------------
这是拷贝replica set myset 中其他节点的数据文件!(记住是数据文件，一定不要mongod.lock ，每个mongod进程都需要自己的数据目录，如果你要运行3个mongod的实例，那么就需要3个独自的目录。mongod启动的时候会在数据目录创建一个mongod.lock文件，阻止其他进程使用此目录.)
第六步：####连接primary库
Tue Nov 1 15:53:24 [initandlisten] connection accepted from 10.250.7.220:54761 #2
Tue Nov 1 15:53:28 [initandlisten] connection accepted from 10.250.7.220:54763 #3
Tue Nov 1 15:53:28 [initandlisten] connection accepted from 10.250.7.220:54764 #4
Tue Nov 1 15:53:34 [clientcursormon] mem (MB) res:31 virt:2606 mapped:1232
Tue Nov 1 15:53:34 [rsStart] trying to contact 10.250.7.220:27018
Tue Nov 1 15:53:34 [rsStart] trying to contact 10.250.7.220:27019
Tue Nov 1 15:53:34 [rsStart] trying to contact 10.250.7.220:27020
第七步：####拷贝配置文件，并保存。
Tue Nov 1 15:53:34 [rsStart] replSet got config version 6 from a remote, saving locally
Tue Nov 1 15:53:34 [rsStart] replSet info saving a newer config version to local.system.replset
Tue Nov 1 15:53:34 [rsStart] replSet saveConfigLocally done

Tue Nov 1 15:53:34 [rsStart] replSet STARTUP2
Tue Nov 1 15:53:34 [rsMgr] replSet total number of votes is even - add arbiter or give one member an extra vote
Tue Nov 1 15:53:34 [rsSync] replSet SECONDARY
Tue Nov 1 15:53:34 [rsHealthPoll] replSet info member 10.250.7.220:27018 is up
Tue Nov 1 15:53:34 [rsHealthPoll] replSet member 10.250.7.220:27018 is now in state SECONDARY
Tue Nov 1 15:53:34 [rsHealthPoll] replSet info member 10.250.7.220:27019 is up
Tue Nov 1 15:53:34 [rsHealthPoll] replSet member 10.250.7.220:27019 is now in state SECONDARY
Tue Nov 1 15:53:34 [rsHealthPoll] replSet info member 10.250.7.220:27020 is up
Tue Nov 1 15:53:34 [rsHealthPoll] replSet member 10.250.7.220:27020 is now in state PRIMARY
Tue Nov 1 15:53:34 [rsMgr] replSet can't see a majority, will not try to elect self
Tue Nov 1 15:53:38 [rsSync] replSet syncing to: 10.250.7.220:27020

从上面的日志来看，这种方法初始化的速度比较快。当然考虑拷贝数据文件的时间，在大量数据的时候，我们可以选择里面一种方案！
二．第二种方法(向已有的replica set 添加新成员)
向已有的replica set 添加新成员是很简单的，新加的成员可以是空的。下面介绍一下如何添加新成员：
添加新成员时，必须在新的机器上先起来一个mongodb服务，必须指出要加入的replica set 集群的名字基本语法如下(其他是默认的)，然后再在replica set的master 库上执行添加操作：
./mongod -dbpath=/opt/mongodata/r1 -port 27017 -replSet myset
当然也可以使用下面的命令，将输出日志记录到/opt/mongodata/r1/rac3.27017.log 里面！
[mongodb@rac3 bin]$./mongod -dbpath=/opt/mongodata/r1 -port 27017 -replSet myset -logpath=/opt/mongodata/r1/rac3.27017.log -logappend &
第一步：在主节点上执行添加操作
[mongodb@rac4 bin]$ ./mongo 127.0.0.1:27020
MongoDB shell version: 2.0.1
connecting to: 127.0.0.1:27020/test
PRIMARY> rs.status();
{
"set" : "myset",
"date" : ISODate("2011-11-01T06:13:01Z"),
"myState" : 1,
"syncingTo" : "10.250.7.220:27018",
"members" : [
{
"_id" : 0,
"name" : "10.250.7.220:27018",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 62668,
"optime" : {
"t" : 1320066230000,
"i" : 1
},
"optimeDate" : ISODate("2011-10-31T13:03:50Z"),
"lastHeartbeat" : ISODate("2011-11-01T06:13:01Z"),
"pingMs" : 0
},
{
"_id" : 1,
"name" : "10.250.7.220:27019",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 63887,
"optime" : {
"t" : 1320066230000,
"i" : 1
},
"optimeDate" : ISODate("2011-10-31T13:03:50Z"),
"lastHeartbeat" : ISODate("2011-11-01T06:13:01Z"),
"pingMs" : 0
},
{
"_id" : 2,
"name" : "10.250.7.220:27020",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"optime" : {
"t" : 1320066230000,
"i" : 1
},
"optimeDate" : ISODate("2011-10-31T13:03:50Z"),
"self" : true
}
],
"ok" : 1
}
PRIMARY> rs.add("10.250.7.241:27017");
{ "ok" : 1 } --添加成功
添加之后，mongodb 会自动执行新加节点的恢复操作，所以第一次查询rs.status()时候，新加节点的信息为空，等待恢复完成之后再次查询就有了基本信息了。
PRIMARY> rs.status();
{
....省略....
{
"_id" : 2,
"name" : "10.250.7.220:27020",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"optime" : {
"t" : 1320128045000,
"i" : 1
},
"optimeDate" : ISODate("2011-11-01T06:14:05Z"),
"self" : true
},
{
"_id" : 3,
"name" : "10.250.7.241:27017",
"health" : 0,
"state" : 8,
"stateStr" : "(not reachable/healthy)",
"uptime" : 0,
"optime" : {
"t" : 0,
"i" : 0
},
"optimeDate" : ISODate("1970-01-01T00:00:00Z"),
"lastHeartbeat" : ISODate("2011-11-01T06:14:07Z"),
"pingMs" : 0,
"errmsg" : "still initializing"
}
],
"ok" : 1
}
PRIMARY> rs.status();
{
"set" : "myset",
"date" : ISODate("2011-11-01T06:16:08Z"),
"myState" : 1,
"syncingTo" : "10.250.7.220:27018",
"members" : [
.....省略....
{
"_id" : 3,
"name" : "10.250.7.241:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 117,
"optime" : {
"t" : 1320128045000,
"i" : 1
},
"optimeDate" : ISODate("2011-11-01T06:14:05Z"),
"lastHeartbeat" : ISODate("2011-11-01T06:16:07Z"),
"pingMs" : 0
}
],
"ok" : 1
}
PRIMARY>
NOTE:这里只是测试了使用空库来添加新节点的，当然也可以从其他节点拷贝数据文件过来，这样可以更快的和主库进行同步操作，从而节省恢复时间！