Hbase regionserver 逐个挂掉的问题分析

最近遇到一个比较诡异的问题，一个regionserver由于GC的原因，导致与zookeeper链接超时，最终被踢出集群。但是，接下来的现象，才是噩梦的开始！！！

1. 一个regionserver由于GC的原因，导致与zookeeper链接超时，最终被踢出集群。

~~~Hbase regionserver log~~~2018-05-31 11:42:17,739 INFO  [MemStoreFlusher.0] regionserver.HRegion: Started memstore flush for cn_kong_groups,\x00\x00\xBB\xE9\x03\x03\x00D\xDF,1527701650816.a177e358544ffe3157a4c0531feb8e5a., current region memstore size 123.40 MB, and 1/1 column families' memstores are being flushed.2018-05-31 11:42:17,740 WARN  [JvmPauseMonitor] util.JvmPauseMonitor: Detected pause in JVM or host machine (eg GC): pause of approximately 52612msGC pool 'ParNew' had collection(s): count=1 time=45897msGC pool 'ConcurrentMarkSweep' had collection(s): count=2 time=6814ms2018-05-31 11:42:17,741 WARN  [B.defaultRpcServer.handler=0,queue=0,port=16020] ipc.RpcServer: (responseTooSlow): {"processingtimems":52721,"call":"Scan(org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ScanRequest)","client":"10.21.23.232:55676","starttimems":1527738085020,"queuetimems":0,"class":"HRegionServer","responsesize":15,"method":"Scan"}2018-05-31 11:42:17,745 INFO  [regionserver/regionserver1.bigdata.com/172.16.11.66:16020-SendThread(ip-10-21-14-154.bigdata.com:2181)] zookeeper.ClientCnxn: Client session timed out, have not heard from server in 61597ms for sessionid 0x15f3454790276db, closing socket connection and attempting reconnect2018-05-31 11:42:17,745 INFO  [main-SendThread(ip-10-21-14-154.bigdata.com:2181)] zookeeper.ClientCnxn: Client session timed out, have not heard from server in 61595ms for sessionid 0x15f3454790276da, closing socket connection and attempting reconnect2018-05-31 11:42:17,893 INFO  [sync.3] wal.FSHLog: Slow sync cost: 148 ms, current pipeline: [DatanodeInfoWithStorage[172.16.11.66:50010,DS-ac448b6f-2964-4900-aeda-4547f2d956b8,DISK], DatanodeInfoWithStorage[172.16.11.67:50010,DS-e8b1727a-81d5-4b65-9854-6b6ad6749b64,DISK], DatanodeInfoWithStorage[10.21.23.41:50010,DS-56eb0e28-5b3c-4047-acc3-d79f3f6bacc2,DISK]]2018-05-31 11:42:17,893 INFO  [sync.2] wal.FSHLog: Slow sync cost: 150 ms, current pipeline: [DatanodeInfoWithStorage[172.16.11.66:50010,DS-ac448b6f-2964-4900-aeda-4547f2d956b8,DISK], DatanodeInfoWithStorage[172.16.11.67:50010,DS-e8b1727a-81d5-4b65-9854-6b6ad6749b64,DISK], DatanodeInfoWithStorage[10.21.23.41:50010,DS-56eb0e28-5b3c-4047-acc3-d79f3f6bacc2,DISK]]

2. Full GC的问题，抛开先不管，一个诡异的问题出现了，启动这个regionserver之后，不长时间这个regionserver又挂掉了。而且，陆续有其他regionserver开始挂掉。

2018-05-31 12:20:12,919 ERROR [RS_OPEN_REGION-regionserver1:16020-2] coprocessor.CoprocessorHost: The coprocessor org.apache.dfs.storage.hbase.cube.v1.coprocessor.observer.AggregateRegionObserver threw org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.ipc.StandbyException): Operation category READ is not supported in state standby        at org.apache.hadoop.hdfs.server.namenode.ha.StandbyState.checkOperation(StandbyState.java:87)        at org.apache.hadoop.hdfs.server.namenode.NameNode$NameNodeHAContext.checkOperation(NameNode.java:1774)        at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkOperation(FSNamesystem.java:1313)        at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getFileInfo(FSNamesystem.java:3856)        at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.getFileInfo(NameNodeRpcServer.java:1006)        at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.getFileInfo(ClientNamenodeProtocolServerSideTranslatorPB.java:843)        at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)        at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:616)        at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:969)        at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2049)        at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2045)        at java.security.AccessController.doPrivileged(Native Method)        at javax.security.auth.Subject.doAs(Subject.java:415)        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1657)        at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2043)2018-05-31 12:20:12,923 ERROR [RS_OPEN_REGION-regionserver1:16020-2] regionserver.HRegionServer: ABORTING region server regionserver1.bigdata.com,16020,1527740222472: The coprocessor org.apache.dfs.storage.hbase.cube.v1.coprocessor.observer.AggregateRegionObserver threw org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.ipc.StandbyException): Operation category READ is not supported in state standby        at org.apache.hadoop.hdfs.server.namenode.ha.StandbyState.checkOperation(StandbyState.java:87)        at org.apache.hadoop.hdfs.server.namenode.NameNode$NameNodeHAContext.checkOperation(NameNode.java:1774)        at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkOperation(FSNamesystem.java:1313)

3. 下面是另一个regionserver的日志，都可以看到"Operation category READ is not supported in state standby"，然后"ABORTING region server"

org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.ipc.StandbyException): Operation category READ is not supported in state standbyat org.apache.hadoop.hdfs.server.namenode.ha.StandbyState.checkOperation(StandbyState.java:87)at org.apache.hadoop.hdfs.server.namenode.NameNode$NameNodeHAContext.checkOperation(NameNode.java:1774)at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkOperation(FSNamesystem.java:1313)at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getFileInfo(FSNamesystem.java:3856)at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.getFileInfo(NameNodeRpcServer.java:1006)at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.getFileInfo(ClientNamenodeProtocolServerSideTranslatorPB.java:843)at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:616)at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:969)at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2049)at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2045)at java.security.AccessController.doPrivileged(Native Method)at javax.security.auth.Subject.doAs(Subject.java:415)at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1657)at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2043)2018-05-31 12:20:13,678 ERROR [RS_OPEN_REGION-regionserver2:16020-2] regionserver.HRegionServer: ABORTING region server regionserver2.bigdata.com,16020,1519906845598: The coprocessor

4. 为什么会出现这个情况呢？感觉十分诡异。

最后经过查找，并且有老司机的指引，发现了问题点。部分hbase表，使用了coprocessor功能，而这个也不是关键

hbase 表的信息如下，注意这部分：hdfs://dfs/dfs/dfs_metadata/coprocessor/*****

dfs_ZZQSWFZ4VW', {TABLE_ATTRIBUTES => {coprocessor$1 => 'hdfs://dfs/dfs/dfs_metadata/coprocessor/dfs-coprocessor-1.5.4.1-0.jar|org.apache.dfs.storage.hbase.cube.v2.coprocessor.endpoint.CubeVisitService|1001|', coprocessor$2 => 'hdfs://dfs/dfs/dfs_metadata/coprocessor/dfs-coprocessor-1.5.4.1-0.jar|org.apache.dfs.storage.hbase.cube.v1.coprocessor.observer.AggregateRegionObserver|1002|', METADATA => {'CREATION_TIME' => '1478338090339', 'GIT_COMMIT' => 'c4e31c1b3a664f598352061ae8703812e9d9cef7;', 'dfs_HOST' => 'dfs_metadata', 'OWNER' => 'xxxx.owner@bigdata.com', 'SEGMENT' => 'WindGreenwichOffline_dev[20161105070130_20161105092132]', 'SPLIT_POLICY' => 'org.apache.hadoop.hbase.regionserver.DisabledRegionSplitPolicy', 'USER' => 'ADMIN'}}, {NAME => 'F1', DATA_BLOCK_ENCODING => 'FAST_DIFF', BLOOMFILTER => 'NONE', COMPRESSION => 'SNAPPY'}

5. 关键点来了，当前的hdfs namenode是HA模式的，但是这个集群，最早是单点namenode的，而当时创建的一些，使用了coprocessor的hbase表，是问题的触发点。

这些coprocessor，指定了hdfs的访问名称，所以，只有原来的namenode是active的状态，才能正常访问。一旦active-standby发生切换，这部分表，就无法正常加载与访问。最终导致整个regionserver的异常。

hbase 表的信息如下，注意这部分：hdfs://old_namenode_host.bigdata.com:9000/dfs/dfs_metadata/***** 这里使用了具体的主机名

'dfs_1KT8V5FL1C', {TABLE_ATTRIBUTES => {coprocessor$1 => '|org.apache.hadoop.hbase.security.access.SecureBulkLoadEndpoint|107374663|', coprocessor$2 => 'hdfs://old_namenode_host.bigdata.com:9000/dfs/dfs_metadata/coprocessor/dfs-coprocessor-1.5.4.1-1.jar|org.apache.dfs.storage.hbase.cube.v2.coprocessor.endpoint.CubeVisitService|1001|', coprocessor$3 => 'hdfs://old_namenode_host.bigdata.com:9000/dfs/dfs_metadata/coprocessor/dfs-coprocessor-1.5.4.1-1.jar|org.apache.dfs.storage.hbase.cube.v1.coprocessor.observer.AggregateRegionObserver|1002|', METADATA => {'CREATION_TIME' => '1493632152602', 'GIT_COMMIT' => 'c4e31c1b3a664f598352061ae8703812e9d9cef7;', 'dfs_HOST' => 'dfs_metadata', 'OWNER' => 'xxxx.owner@bigdata.com', 'SEGMENT' => 'WindCGN2_clone[20160501000000_20170501000000]', 'SPLIT_POLICY' => 'org.apache.hadoop.hbase.regionserver.DisabledRegionSplitPolicy', 'USER' => 'ADMIN'}}, {NAME => 'F1', DATA_BLOCK_ENCODING => 'FAST_DIFF', BLOOMFILTER => 'NONE', COMPRESSION => 'SNAPPY'}

最后总结，为什么会其他的regionserver逐个挂掉呢？为什么namenode已经切换很久，在GC引发regionserver挂掉，才会导致regionserver逐个挂掉的连锁反应？

1. GC引起第一个regionserver挂掉，而其上的region，一定会被master分配到其他的regionserver上，而其他的regionserver也不能正确接管这个region，所以也是以相同的错误，导致第二个regionserver挂掉，然后第三个。理论上，全部的regionserver都会挂掉。

2. 由于在第一个regionserver挂掉前，namenode active-standby发生切换之前，hbase regionserver已经启动，并且能正常识别region，所以没有引发问题，但是regionserver发生重启，就肯定不能正常挂载这个region的。

关于coprocessor功能，请参考

http://www.zhyea.com/2017/04/13/using-hbase-coprocessor.html

https://www.3pillarglobal.com/insights/hbase-coprocessors