The following property does exactly what I wanted our environment to do. I had a 4GiB Heap and ran the job and no jobs failed. Then I dropped our cluster heap to 1GiB and reran the same resource intensive task.
This property must be added to the "HBase Service Advanced Configuration Snippet (Safety Valve) for hbase-site.xml" <property> <name>hbase.client.scanner.max.result.size</name> <value>67108864</value> </property> We noted that 64MiB would be enough, but we also experimented 128MiB. I may do a write-up and elaborate some more on this. On Mon, May 12, 2014 at 1:38 PM, Vladimir Rodionov <vrodio...@carrieriq.com>wrote: > All your OOME are on the client side (map task). Your map tasks need more > heap. > Reduce # of map tasks and increase max heap size per map task. > > Best regards, > Vladimir Rodionov > Principal Platform Engineer > Carrier IQ, www.carrieriq.com > e-mail: vrodio...@carrieriq.com > > ________________________________________ > From: Geovanie Marquez [geovanie.marq...@gmail.com] > Sent: Thursday, May 08, 2014 2:35 PM > To: user@hbase.apache.org > Subject: Re: RPC Client OutOfMemoryError Java Heap Space > > sorry didn't include version > > CDH5 version - CDH-5.0.0-1.cdh5.0.0.p0.47 > > > On Thu, May 8, 2014 at 5:32 PM, Geovanie Marquez < > geovanie.marq...@gmail.com > > wrote: > > > Hey group, > > > > There is one job that scans HBase contents and is really resource > > intensive using all resources available to yarn (under Resource Manager). > > In my case, that is 8GB. My expectation here is that a properly > configured > > cluster would kill the application or degrade the application performance > > but never ever take a region server down. This is intended to be a > > multi-tenant environment where developers may submit jobs at will and I > > would want a configuration where the cluster services are not exited in > > this way because of memory. > > > > The simple solution here, is to change the way the job consumes resources > > so that when run it is not so resource greedy. I want to understand how I > > can mitigate this situation in general. > > > > **It FAILS with the following config:** > > The RPC client has 30 handlers > > write buffer of 2MiB > > The RegionServer heap is 4GiB > > Max Size of all memstores is 0.40 of total heap > > HFile Block Cache Size is 0.40 > > Low watermark for memstore flush is 0.38 > > HBase Memstore size is 128MiB > > > > **Job still FAILS with the following config:** > > Everything else the same except > > The RPC client has 10 handlers > > > > **Job still FAILS with the following config:** > > Everything else the same except > > HFile Block Cache Size is 0.10 > > > > > > When this runs I get the following error stacktrace: > > # > > #How do I avoid this via configuration. > > # > > > > java.lang.OutOfMemoryError: Java heap space > > at > org.apache.hadoop.hbase.ipc.RpcClient$Connection.readResponse(RpcClient.java:1100) > > at > org.apache.hadoop.hbase.ipc.RpcClient$Connection.run(RpcClient.java:721) > > 2014-05-08 16:23:54,705 WARN [IPC Client (1242056950) connection to > c1d001.in.wellcentive.com/10.2.4.21:60020 from hbase] > org.apache.hadoop.ipc.RpcClient: IPC Client (1242056950) connection to > c1d001.in.wellcentive.com/10.2.4.21:60020 from hbase: unexpected > exception receiving call responses > > # > > > > ###Yes, there was an RPC timeout this is what is killing the server > because the timeout is eventually (1minute later) reached. > > > > # > > > > java.lang.OutOfMemoryError: Java heap space > > at > org.apache.hadoop.hbase.ipc.RpcClient$Connection.readResponse(RpcClient.java:1100) > > at > org.apache.hadoop.hbase.ipc.RpcClient$Connection.run(RpcClient.java:721) > > 2014-05-08 16:23:55,319 INFO [main] > org.apache.hadoop.hbase.mapreduce.TableRecordReaderImpl: recovered from > org.apache.hadoop.hbase.DoNotRetryIOException: Failed after retry of > OutOfOrderScannerNextException: was there a rpc timeout? > > at > org.apache.hadoop.hbase.client.ClientScanner.next(ClientScanner.java:384) > > at > org.apache.hadoop.hbase.mapreduce.TableRecordReaderImpl.nextKeyValue(TableRecordReaderImpl.java:194) > > at > org.apache.hadoop.hbase.mapreduce.TableRecordReader.nextKeyValue(TableRecordReader.java:138) > > at > org.apache.hadoop.mapred.MapTask$NewTrackingRecordReader.nextKeyValue(MapTask.java:533) > > at > org.apache.hadoop.mapreduce.task.MapContextImpl.nextKeyValue(MapContextImpl.java:80) > > at > org.apache.hadoop.mapreduce.lib.map.WrappedMapper$Context.nextKeyValue(WrappedMapper.java:91) > > at org.apache.hadoop.mapreduce.Mapper.run(Mapper.java:144) > > at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:764) > > at org.apache.hadoop.mapred.MapTask.run(MapTask.java:340) > > at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:168) > > at java.security.AccessController.doPrivileged(Native Method) > > at javax.security.auth.Subject.doAs(Subject.java:415) > > at > org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1548) > > at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:163) > > > > # > > > > ## Probably caused by the OOME above > > > > # > > > > Caused by: > org.apache.hadoop.hbase.exceptions.OutOfOrderScannerNextException: > org.apache.hadoop.hbase.exceptions.OutOfOrderScannerNextException: Expected > nextCallSeq: 1 But the nextCallSeq got from client: 0; request=scanner_id: > 5612205039322936440 number_of_rows: 10000 close_scanner: false > next_call_seq: 0 > > at > org.apache.hadoop.hbase.regionserver.HRegionServer.scan(HRegionServer.java:3018) > > at > org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientService$2.callBlockingMethod(ClientProtos.java:26929) > > at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2175) > > at > org.apache.hadoop.hbase.ipc.RpcServer$Handler.run(RpcServer.java:1879) > > > > > > Confidentiality Notice: The information contained in this message, > including any attachments hereto, may be confidential and is intended to be > read only by the individual or entity to whom this message is addressed. If > the reader of this message is not the intended recipient or an agent or > designee of the intended recipient, please note that any review, use, > disclosure or distribution of this message or its attachments, in any form, > is strictly prohibited. If you have received this message in error, please > immediately notify the sender and/or notificati...@carrieriq.com and > delete or destroy any copy of this message and its attachments. >