accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <>
Subject [jira] [Commented] (ACCUMULO-1355) Accumulo ingestion hangs/fails for 300 MB+ file, local machine memory usage grows to 10GB+
Date Fri, 26 Apr 2013 22:16:15 GMT


Josh Elser commented on ACCUMULO-1355:

Is this a one-node instance?

Also, I assume you can reliably replicate this? Take a look at the GC debug statements printed
in $ACCUMULO_LOG_DIR/tserver_$hostname.debug.log. I'm curious if the TServer is struggling
to allocate the memory in some "contiguous" region of heap. Does the TServer eventually die
if you let it hang?

And, if you're adventurous, I'd be curious to see what happens when you use the native maps
instead of the java version (`src/server/src/main/c++/make`).

I'm not entirely sure how things work when you have 300MB contained all within one k/v pair.
Others may have a better understanding of this without me going diving into code to make a
> Accumulo ingestion hangs/fails for 300 MB+ file, local machine memory usage grows to
> ------------------------------------------------------------------------------------------
>                 Key: ACCUMULO-1355
>                 URL:
>             Project: Accumulo
>          Issue Type: Bug
>          Components: client, thrift, tserver
>    Affects Versions: 1.4.2, 1.4.3
>            Reporter: Nasheb Ismaily
>            Assignee: Billie Rinaldi
>            Priority: Blocker
> *Bug:*
> I am attempting to ingest a 300+ MB file into Accumulo, however the ingestion process
hangs and my local machines memory consumption grows to 10GB+.
> The mutation is never put into accumulo.
> *The file I am ingesting can be found here (needs to be unzipped):*
> *I’ve attached a snippet of code used to ingest the data here:*
> *Initial Investigation:*
> While attempting to upload the file, the thread which sends the mutation to the tablet
‘waits’ in the waitRTE() method which is called by synchronized void flush() in
and is never ‘woken up’ by a notify call. While ‘waiting’ my local machines memory
allocation grows to 10 GB+. 
> *Thread Stack Trace (from java profiler):*
> java.nio.Bits.copyFromArray(
> java.nio.DirectByteBuffer.put(
> java.nio.DirectByteBuffer.put(
> org.apache.thrift.transport.TIOStreamTransport.write(
> org.apache.thrift.transport.TFramedTransport.flush(
> org.apache.accumulo.core.client.impl.ThriftTransportPool$CachedTTransport.flush(
> org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Client.send_applyUpdates(
> org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Client.applyUpdates(
> sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> sun.reflect.NativeMethodAccessorImpl.invoke(
> sun.reflect.DelegatingMethodAccessorImpl.invoke(
> java.lang.reflect.Method.invoke(
> org.apache.accumulo.cloudtrace.instrument.thrift.TraceWrap$2.invoke(
> com.sun.proxy.$Proxy7.applyUpdates(Unknown Source)
> org.apache.accumulo.core.client.impl.TabletServerBatchWriter$MutationWriter.sendMutationsToTabletServer(
> org.apache.accumulo.core.client.impl.TabletServerBatchWriter$MutationWriter.access$1400(
> org.apache.accumulo.core.client.impl.TabletServerBatchWriter$MutationWriter$SendTask.send(
> org.apache.accumulo.core.client.impl.TabletServerBatchWriter$MutationWriter$
> java.util.concurrent.Executors$
> java.util.concurrent.FutureTask$Sync.innerRun(
> java.util.concurrent.ThreadPoolExecutor$Worker.runTask(
> java.util.concurrent.ThreadPoolExecutor$
> *Tserver log error:*
> 2013-04-26 15:36:25,253 [server.TNonblockingServer] ERROR: Unexpected exception while
> java.lang.RuntimeException: No Such SessionID
> 	at org.apache.accumulo.server.tabletserver.TabletServer$ThriftClientHandler.applyUpdates(
> 	at sun.reflect.GeneratedMethodAccessor12.invoke(Unknown Source)
> 	at sun.reflect.DelegatingMethodAccessorImpl.invoke(
> 	at java.lang.reflect.Method.invoke(
> 	at org.apache.accumulo.cloudtrace.instrument.thrift.TraceWrap$1.invoke(
> 	at com.sun.proxy.$Proxy2.applyUpdates(Unknown Source)
> 	at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor$applyUpdates.process(
> 	at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor.process(
> 	at org.apache.accumulo.server.util.TServerUtils$TimedProcessor.process(
> 	at org.apache.thrift.server.TNonblockingServer$FrameBuffer.invoke(
> 	at org.apache.accumulo.server.util.TServerUtils$THsHaServer$
> 	at java.util.concurrent.ThreadPoolExecutor.runWorker(
> 	at java.util.concurrent.ThreadPoolExecutor$
> 	at
> 	at

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message