MongoDB moveChunk在oplog中的表现

db1库下的hoo集合使用hash分片，通过手工拆分shrs02的chunk触发moveChunk
可以观察到shrs02的oplog执行了delete操作{"op":"d"}
shrs02:PRIMARY> db.oplog.rs.find({"ns" : "db1.hoo"}).sort({$natural:-1}).limit(5);
{ "ts" : Timestamp(1550548015, 279), "t" : NumberLong(1), "h" : NumberLong("-304152233719500506"), "v" : 2, "op" : "d", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:46:55.082Z"), "o" : { "x" : 2, "_id" : ObjectId("5c6b76c0e5c6c8a235bca558") } }
{ "ts" : Timestamp(1550548015, 278), "t" : NumberLong(1), "h" : NumberLong("-5045599070557482756"), "v" : 2, "op" : "d", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:46:55.082Z"), "o" : { "x" : 2, "_id" : ObjectId("5c6b76c0e5c6c8a235bca556") } }
{ "ts" : Timestamp(1550548015, 277), "t" : NumberLong(1), "h" : NumberLong("5120191885956451298"), "v" : 2, "op" : "d", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:46:55.082Z"), "o" : { "x" : 2, "_id" : ObjectId("5c6b76c0e5c6c8a235bca543") } }
{ "ts" : Timestamp(1550548015, 276), "t" : NumberLong(1), "h" : NumberLong("4061724698943275872"), "v" : 2, "op" : "d", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:46:55.082Z"), "o" : { "x" : 2, "_id" : ObjectId("5c6b76c0e5c6c8a235bca53c") } }
{ "ts" : Timestamp(1550548015, 275), "t" : NumberLong(1), "h" : NumberLong("-4863719428805872629"), "v" : 2, "op" : "d", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:46:55.082Z"), "o" : { "x" : 2, "_id" : ObjectId("5c6b76c0e5c6c8a235bca538") } }
同样可以看到shrs01的oplog记录着insert操作{"op":"i"}
shrs01:PRIMARY> use local
switched to db local
shrs01:PRIMARY> db.oplog.rs.find({"ns" : "db1.hoo"}).sort({$natural:-1}).limit(5);
{ "ts" : Timestamp(1550547112, 11052), "t" : NumberLong(1), "h" : NumberLong("-305777054114754335"), "v" : 2, "op" : "i", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:31:52.515Z"), "o" : { "_id" : ObjectId("5c6b76c0e5c6c8a235bca558"), "x" : 2, "udt" : ISODate("2019-02-19T03:23:44.990Z"), "type" : 0, "type2" : 1 } }
{ "ts" : Timestamp(1550547112, 11051), "t" : NumberLong(1), "h" : NumberLong("669359981315100370"), "v" : 2, "op" : "i", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:31:52.515Z"), "o" : { "_id" : ObjectId("5c6b76c0e5c6c8a235bca556"), "x" : 2, "udt" : ISODate("2019-02-19T03:23:44.987Z"), "type" : 2, "type2" : 1 } }
{ "ts" : Timestamp(1550547112, 11050), "t" : NumberLong(1), "h" : NumberLong("-4175246498915868787"), "v" : 2, "op" : "i", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:31:52.515Z"), "o" : { "_id" : ObjectId("5c6b76c0e5c6c8a235bca543"), "x" : 2, "udt" : ISODate("2019-02-19T03:23:44.958Z"), "type" : 2, "type2" : 0 } }
{ "ts" : Timestamp(1550547112, 11049), "t" : NumberLong(1), "h" : NumberLong("8144974434003347592"), "v" : 2, "op" : "i", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:31:52.515Z"), "o" : { "_id" : ObjectId("5c6b76c0e5c6c8a235bca53c"), "x" : 2, "udt" : ISODate("2019-02-19T03:23:44.946Z"), "type" : 6, "type2" : 4 } }
{ "ts" : Timestamp(1550547112, 11048), "t" : NumberLong(1), "h" : NumberLong("-1340638049932755147"), "v" : 2, "op" : "i", "ns" : "db1.hoo", "ui" : UUID("8d7b3877-cc2e-4e86-9a45-e66061e58697"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T03:31:52.515Z"), "o" : { "_id" : ObjectId("5c6b76c0e5c6c8a235bca538"), "x" : 2, "udt" : ISODate("2019-02-19T03:23:44.938Z"), "type" : 4, "type2" : 6 } }

db1库下的t_user集合使用range分片，通过手工拆分shrs02的chunk触发moveChunk
可以观察到shrs01的oplog也是记录insert操作{"op":"i"}
mongos> sh.splitFind("db1.t_user", {userid:50000});
shrs01:PRIMARY> use local
switched to db local
shrs01:PRIMARY> db.oplog.rs.find({"ns" : "db1.t_user"}).sort({$natural:-1}).limit(5);
{ "ts" : Timestamp(1550544266, 23696), "t" : NumberLong(1), "h" : NumberLong("6640796598824740378"), "v" : 2, "op" : "i", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:44:26.617Z"), "o" : { "_id" : ObjectId("5c6b6c54806fa64b3bd55089"), "userid" : 50000, "name" : "user50000", "age" : 25 } }
{ "ts" : Timestamp(1550544266, 23695), "t" : NumberLong(1), "h" : NumberLong("5097978940295012072"), "v" : 2, "op" : "i", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:44:26.617Z"), "o" : { "_id" : ObjectId("5c6b6c54806fa64b3bd55088"), "userid" : 49999, "name" : "user49999", "age" : 25 } }
{ "ts" : Timestamp(1550544266, 23694), "t" : NumberLong(1), "h" : NumberLong("-9206589727349063515"), "v" : 2, "op" : "i", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:44:26.617Z"), "o" : { "_id" : ObjectId("5c6b6c54806fa64b3bd55087"), "userid" : 49998, "name" : "user49998", "age" : 25 } }
{ "ts" : Timestamp(1550544266, 23693), "t" : NumberLong(1), "h" : NumberLong("-5836469842737896486"), "v" : 2, "op" : "i", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:44:26.617Z"), "o" : { "_id" : ObjectId("5c6b6c54806fa64b3bd55086"), "userid" : 49997, "name" : "user49997", "age" : 25 } }
{ "ts" : Timestamp(1550544266, 23692), "t" : NumberLong(1), "h" : NumberLong("6973611600053813704"), "v" : 2, "op" : "i", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:44:26.617Z"), "o" : { "_id" : ObjectId("5c6b6c54806fa64b3bd55085"), "userid" : 49996, "name" : "user49996", "age" : 25 } }
可以观察到shrs02的oplog也是记录insert操作{"op":"d"}
shrs02:PRIMARY> use local
switched to db local
shrs02:PRIMARY> db.oplog.rs.find({"ns" : "db1.t_user"}).sort({$natural:-1}).limit(5);
{ "ts" : Timestamp(1550545177, 464), "t" : NumberLong(1), "h" : NumberLong("6274443387502896231"), "v" : 2, "op" : "d", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:59:37.085Z"), "o" : { "userid" : 50000, "_id" : ObjectId("5c6b6c54806fa64b3bd55089") } }
{ "ts" : Timestamp(1550545177, 463), "t" : NumberLong(1), "h" : NumberLong("4608755019998843904"), "v" : 2, "op" : "d", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:59:37.085Z"), "o" : { "userid" : 49999, "_id" : ObjectId("5c6b6c54806fa64b3bd55088") } }
{ "ts" : Timestamp(1550545177, 462), "t" : NumberLong(1), "h" : NumberLong("-1809926020796113867"), "v" : 2, "op" : "d", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:59:37.085Z"), "o" : { "userid" : 49998, "_id" : ObjectId("5c6b6c54806fa64b3bd55087") } }
{ "ts" : Timestamp(1550545177, 461), "t" : NumberLong(1), "h" : NumberLong("-4040712242599584545"), "v" : 2, "op" : "d", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:59:37.085Z"), "o" : { "userid" : 49997, "_id" : ObjectId("5c6b6c54806fa64b3bd55086") } }
{ "ts" : Timestamp(1550545177, 460), "t" : NumberLong(1), "h" : NumberLong("3894943139412829378"), "v" : 2, "op" : "d", "ns" : "db1.t_user", "ui" : UUID("25190278-5b47-4c80-8138-a832f40840cb"), "fromMigrate" : true, "wall" : ISODate("2019-02-19T02:59:37.085Z"), "o" : { "userid" : 49996, "_id" : ObjectId("5c6b6c54806fa64b3bd55085") } }

从日志中可以看出chunk迁移在源分片上表现为删除操作，在目标分片上表现为插入操作

mongos> sh.status() #最终的分片集群信息
--- Sharding Status --- 
  sharding version: {
  	"_id" : 1,
  	"minCompatibleVersion" : 5,
  	"currentVersion" : 6,
  	"clusterId" : ObjectId("5c6b6a17e5837b6787ff6a09")
  }
  shards:
        {  "_id" : "shrs01",  "host" : "shrs01/10.11.12.140:7101,10.11.12.140:7102,10.11.12.140:7103",  "state" : 1 }
        {  "_id" : "shrs02",  "host" : "shrs02/10.11.12.140:7201,10.11.12.140:7202,10.11.12.140:7203",  "state" : 1 }
  active mongoses:
        "3.6.10" : 1
  autosplit:
        Currently enabled: yes
  balancer:
        Currently enabled:  yes
        Currently running:  no
        Failed balancer rounds in last 5 attempts:  0
        Migration Results for the last 24 hours: 
                3 : Success
  databases:
        {  "_id" : "config",  "primary" : "config",  "partitioned" : true }
                config.system.sessions
                        shard key: { "_id" : 1 }
                        unique: false
                        balancing: true
                        chunks:
                                shrs01	1
                        { "_id" : { "$minKey" : 1 } } -->> { "_id" : { "$maxKey" : 1 } } on : shrs01 Timestamp(1, 0) 
        {  "_id" : "db1",  "primary" : "shrs02",  "partitioned" : true }
                db1.hoo
                        shard key: { "x" : "hashed" }
                        unique: false
                        balancing: true
                        chunks:
                                shrs01	3
                                shrs02	3
                        { "x" : { "$minKey" : 1 } } -->> { "x" : NumberLong("-4611686018427387902") } on : shrs01 Timestamp(2, 2) 
                        { "x" : NumberLong("-4611686018427387902") } -->> { "x" : NumberLong(0) } on : shrs01 Timestamp(2, 3) 
                        { "x" : NumberLong(0) } -->> { "x" : NumberLong("2433984061343616404") } on : shrs01 Timestamp(3, 0) 
                        { "x" : NumberLong("2433984061343616404") } -->> { "x" : NumberLong("4470791281878691347") } on : shrs02 Timestamp(3, 1) 
                        { "x" : NumberLong("4470791281878691347") } -->> { "x" : NumberLong("4611686018427387902") } on : shrs02 Timestamp(2, 9) 
                        { "x" : NumberLong("4611686018427387902") } -->> { "x" : { "$maxKey" : 1 } } on : shrs02 Timestamp(2, 5) 
                db1.t_user
                        shard key: { "userid" : 1 }
                        unique: true
                        balancing: true
                        chunks:
                                shrs01	1
                                shrs02	1
                        { "userid" : { "$minKey" : 1 } } -->> { "userid" : 50001 } on : shrs01 Timestamp(2, 0) 
                        { "userid" : 50001 } -->> { "userid" : { "$maxKey" : 1 } } on : shrs02 Timestamp(2, 1) 

总结：
(1) mongos分片集群最小的逻辑存储单元（默认64M），表示自身所在分片下一个分片key值的连续范围
(2) chunk迁移在shard之间的oplog记录的是普通的插入删除操作
(3) chunk迁移完成以后才更新config server的元数据，最后才异步删除源shard的数据
发送评论 编辑评论

发送评论编辑评论
