The default memory settings for flume are extremely low. Try giving it more Java memory.
On Jun 18, 2014, at 12:57 PM, David Gates <david.ga...@gmail.com> wrote: > When running a test to see if I can get Morphline Interceptor working to > convert some timestamps in logfiles, I am getting a java GC overhead limit > exceeded error. > > Command: > > flume-ng agent --conf ./conf -f testflume.conf > -Dflume.root.logger=DBUG,console -n agent > > testflume.conf: > > agent.channels.memory-channel.type = memory > agent.sources.spool-source.type = spooldir > agent.sources.spool-source.spoolDir = /home/impala/spool/ > agent.sources.spool-source.channels = memory-channel > agent.sources.spool-source.interceptors = morphlineinterceptor > agent.sources.spool-source.interceptors.morphlineinterceptor.type = > org.apache.flume.sink.solr.morphline.MorphlineInterceptor$Builder > agent.sources.spool-source.interceptors.morphlineinterceptor.morphlineFile > = /root/morphline.conf > agent.sources.spool-source.interceptors.morphlineinterceptor.morphlineId = > morphline1 > agent.sinks.hdfs-sink.channel = memory-channel > agent.sinks.hdfs-sink.type = hdfs > agent.sinks.hdfs-sink.hdfs.path = /user/impala/ > agent.sinks.hdfs-sink.hdfs.fileType = DataStream > agent.channels = memory-channel > agent.sources = spool-source > agent.sinks = hdfs-sink > > morphline.conf: > > morphlines : [ > { > id : morphline1 > importCommands : ["com.cloudera.**"] > > commands : [ > { > readCSV { > separator: ";" > trim: true > columns: > [Header1,Header2,Header3,ConnectionType,SessionID,ReleaseCause,StartTime,AnswerTime,ReleaseTime,MinutesWest,ReleaseCauseProto,ReleaseCauseNum,FirstReleaseDialogue,TrunkIDOrig,VOIPProtoOrig,SourceNumOrig,SourceHostOrig,DestNumOrig,DestHostOrig,OrigCallID,OrigRemotePayloadIP,OrigRemotePayloadUDP,OrigLocalPayloadIP,OrigLocalPayloadUDP,OrigCodecList,OrigIngressPackets,OrigEgressPackets,OrigIngressOctets,OrigEgressOctets,OrigIngressPacketLoss,OrigIngressDelay,OrigIngressJitter,TrunkIDTerm,VOIPProtoTerm,SourceNumTerm,SourceHostTerm,DestNumTerm,DestHostTerm,TermCallID,TermRemotePayloadIP,TermRemotePayloadUDP,TermLocalPayloadIP,TermLocalPayloadUDP,TermCodecList,TermIngressPackets,TermEgressPackets,TermIngressOctets,TermEgressOctets,TermIngpressPacketLoss,TermIngressDelay,TermIngressJitter,FinalRouteIndication,RoutingDigits,CallDurSec,PostDialDelaySec,RingTimeSec,DurMiniSec,ConfID,RPIDANI,RouteEntryIndex,RouteTable,LNPDip,IngressLRN,EgressLRN,CNAMDip,DNCDip,OrigTrunkAlias,TermTrunkAlias,ERSDip,OLIDigits] > } > } > ] > } > ] > > > I had originally also had a convertTimeStamp command in there but removed > it to troubleshoot. > > The error I get when running is as follows: > > 14/05/21 08:37:23 INFO api.MorphlineContext: Importing commands > 14/05/21 08:37:31 ERROR node.PollingPropertiesFileConfigurationProvider: > Unhandled error > java.lang.OutOfMemoryError: GC overhead limit exceeded > > > Ive tried googling but I can't find anything specific to flume/morphline > with GC limit exceeded, any help/ideas would be appreciated.