hadoop - Lily Hbase Indexers 无故退出

Question

我正在运行 Cloudera/Solr 集群，并尝试使用 hbase-solr (Lily) 索引器将 Hbase 的 NRT 索引编入 Solr。批处理模式索引工作正常。

但是，在我开始以恒定流加载数据后，Lily 索引器开始一个接一个地死掉。他们不会打印出向我跳出的特定错误消息，但都以相同的方式结束:

2014-09-10 16:04:56,770 INFO org.apache.hadoop.hbase.zookeeper.RecoverableZooKeeper: Process identifier=ip-172-31-1-204.ap-southeast-2.compute.internal,44013,1410329096767 connecting to ZooKeeper ensemble=ip-172-31-1-205.ap-southeast-2.compute.internal:2181,ip-172-31-1-206.ap-southeast-2.compute.internal:2181,ip-172-31-1-204.ap-southeast-2.compute.internal:2181
2014-09-10 16:04:56,771 INFO org.apache.zookeeper.ClientCnxn: Opening socket connection to server ip-172-31-1-206.ap-southeast-2.compute.internal/172.31.1.206:2181. Will not attempt to authenticate using SASL (unknown error)
2014-09-10 16:04:56,772 INFO org.apache.hadoop.ipc.RpcServer: RpcServer.listener,port=44013: starting
2014-09-10 16:04:56,771 INFO org.apache.hadoop.ipc.RpcServer: RpcServer.responder: starting
2014-09-10 16:04:56,773 INFO org.apache.zookeeper.ClientCnxn: Socket connection established to ip-172-31-1-206.ap-southeast-2.compute.internal/172.31.1.206:2181, initiating session
2014-09-10 16:04:56,775 INFO com.ngdata.hbaseindexer.supervisor.IndexerSupervisor: Started indexer for indexFeature
2014-09-10 16:04:56,776 INFO org.apache.zookeeper.ClientCnxn: Session establishment complete on server ip-172-31-1-206.ap-southeast-2.compute.internal/172.31.1.206:2181, sessionid = 0x1485c7ff13602fd, negotiated timeout = 60000
2014-09-10 16:04:56,813 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:04:57,287 INFO org.kitesdk.morphline.api.MorphlineContext: Done importing commands
2014-09-10 16:04:57,289 INFO org.apache.solr.client.solrj.impl.HttpClientUtil: Creating new http client, config:
2014-09-10 16:04:57,297 INFO org.apache.hadoop.ipc.RpcServer: regionserver/ip-172-31-1-204.ap-southeast-2.compute.internal/172.31.1.204:0: started 10 reader(s).
2014-09-10 16:04:57,299 INFO org.apache.zookeeper.ZooKeeper: Initiating client connection, connectString=ip-172-31-1-205.ap-southeast-2.compute.internal:2181,ip-172-31-1-206.ap-southeast-2.compute.internal:2181,ip-172-31-1-204.ap-southeast-2.compute.internal:2181 sessionTimeout=60000 watcher=ip-172-31-1-204.ap-southeast-2.compute.internal,44713,1410329097297, quorum=ip-172-31-1-205.ap-southeast-2.compute.internal:2181,ip-172-31-1-206.ap-southeast-2.compute.internal:2181,ip-172-31-1-204.ap-southeast-2.compute.internal:2181, baseZNode=/hbase
2014-09-10 16:04:57,301 INFO org.apache.hadoop.hbase.zookeeper.RecoverableZooKeeper: Process identifier=ip-172-31-1-204.ap-southeast-2.compute.internal,44713,1410329097297 connecting to ZooKeeper ensemble=ip-172-31-1-205.ap-southeast-2.compute.internal:2181,ip-172-31-1-206.ap-southeast-2.compute.internal:2181,ip-172-31-1-204.ap-southeast-2.compute.internal:2181
2014-09-10 16:04:57,302 INFO org.apache.zookeeper.ClientCnxn: Opening socket connection to server ip-172-31-1-204.ap-southeast-2.compute.internal/172.31.1.204:2181. Will not attempt to authenticate using SASL (unknown error)
2014-09-10 16:04:57,303 INFO org.apache.hadoop.ipc.RpcServer: RpcServer.responder: starting
2014-09-10 16:04:57,303 INFO org.apache.zookeeper.ClientCnxn: Socket connection established to ip-172-31-1-204.ap-southeast-2.compute.internal/172.31.1.204:2181, initiating session
2014-09-10 16:04:57,304 INFO org.apache.hadoop.ipc.RpcServer: RpcServer.listener,port=44713: starting
2014-09-10 16:04:57,306 INFO com.ngdata.hbaseindexer.supervisor.IndexerSupervisor: Started indexer for indexSeenBlock
2014-09-10 16:04:57,307 INFO org.apache.zookeeper.ClientCnxn: Session establishment complete on server ip-172-31-1-204.ap-southeast-2.compute.internal/172.31.1.204:2181, sessionid = 0x3485c7fee8f0374, negotiated timeout = 60000
2014-09-10 16:04:57,349 INFO org.mortbay.log: Logging to org.slf4j.impl.Log4jLoggerAdapter(org.mortbay.log) via org.mortbay.log.Slf4jLog
2014-09-10 16:04:57,536 INFO org.mortbay.log: jetty-6.1.26.cloudera.2
2014-09-10 16:04:58,663 INFO org.mortbay.log: Started SelectChannelConnector@0.0.0.0:11060
2014-09-10 16:05:01,591 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,597 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,641 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,650 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,688 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,726 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,732 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,740 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands
2014-09-10 16:05:01,752 INFO org.kitesdk.morphline.api.MorphlineContext: Importing commands

Cloudera 管理器除了说进程已经退出外，也没有提供任何有用的信息。有些记录确实在 Solr 索引中得到更新，表明索引器至少在一段时间内正常工作。

我在 RHEL6.5 和 JDK7 上运行最新的 CDH 5.1。

Answer 1

如果您正在使用 WAL，即在您的 hbase 客户端程序(例如 mapreduce)中写入头日志选项为 FALSE(出于性能原因)。它不会写入 HLOG。

NRT 适用于 WAL/HLOG。

Batch Index 工作正常，因为它不适用于 WAL