[
https://issues.apache.org/jira/browse/PHOENIX-4685?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16434244#comment-16434244
]
Josh Elser commented on PHOENIX-4685:
-------------------------------------
{quote} * the whole purpose of using CoprocessorHConnectionTableFactory is to
build the connection with custom configuration. if we use the shared connection
then our index configuration will not take effect like our ServerRpcController
, threads etc . see HBASE-18359 and PHOENIX-3994.{quote}
This seems really important to me. That's a good catch, Ankit.
{quote} * default value of 256 for "hbase.hconnection.threads.max" may be less
for all index writers on the regionserver but in general scenario 8 threads per
core(with common 32 core server) should be sufficient but not sure such limit
result in any regression from previous versions or require documentation
change.{quote}
Makes me wonder about how we can evaluate whether this is a problem in the
future. It would be neat if we could track hconnection thread saturation via
Phoenix (along with the other 2ndary index metrics we added recently). I'm not
sure how difficult that would be...
{quote}So , how about creating our own ShortCircuitConnection from the
environment with custom config and pool , and shared the same across all region
while writing?
{quote}
Make sure I'm understanding this correctly: we'd create+cache our own
HConnection from the CPEnv, with custom Phoenix properties to control the
resources on that HConn? I like that – avoids us causing other HBase problems
trying to tweak Phoenix things.
> Parallel writes continuously to indexed table failing with OOME very quickly
> in 5.x branch
> ------------------------------------------------------------------------------------------
>
> Key: PHOENIX-4685
> URL: https://issues.apache.org/jira/browse/PHOENIX-4685
> Project: Phoenix
> Issue Type: Bug
> Reporter: Rajeshbabu Chintaguntla
> Assignee: Rajeshbabu Chintaguntla
> Priority: Major
> Fix For: 5.0.0
>
> Attachments: PHOENIX-4685.patch, PHOENIX-4685_jstack
>
>
> Currently trying to write data to indexed table failing with OOME where
> unable to create native threads. But it's working fine with 4.7.x branches.
> Found many threads created for meta lookup and shared threads and no space to
> create threads. This is happening even with short circuit writes enabled.
> {noformat}
> 2018-04-08 13:06:04,747 WARN
> [RpcServer.default.FPBQ.Fifo.handler=9,queue=0,port=16020]
> index.PhoenixIndexFailurePolicy: handleFailure failed
> java.io.IOException: java.lang.reflect.UndeclaredThrowableException
> at org.apache.hadoop.hbase.security.User.runAsLoginUser(User.java:185)
> at
> org.apache.phoenix.index.PhoenixIndexFailurePolicy.handleFailureWithExceptions(PhoenixIndexFailurePolicy.java:217)
> at
> org.apache.phoenix.index.PhoenixIndexFailurePolicy.handleFailure(PhoenixIndexFailurePolicy.java:143)
> at
> org.apache.phoenix.hbase.index.write.IndexWriter.writeAndKillYourselfOnFailure(IndexWriter.java:160)
> at
> org.apache.phoenix.hbase.index.write.IndexWriter.writeAndKillYourselfOnFailure(IndexWriter.java:144)
> at
> org.apache.phoenix.hbase.index.Indexer.doPostWithExceptions(Indexer.java:632)
> at org.apache.phoenix.hbase.index.Indexer.doPost(Indexer.java:607)
> at
> org.apache.phoenix.hbase.index.Indexer.postBatchMutateIndispensably(Indexer.java:590)
> at
> org.apache.hadoop.hbase.regionserver.RegionCoprocessorHost$30.call(RegionCoprocessorHost.java:1037)
> at
> org.apache.hadoop.hbase.regionserver.RegionCoprocessorHost$30.call(RegionCoprocessorHost.java:1034)
> at
> org.apache.hadoop.hbase.coprocessor.CoprocessorHost$ObserverOperationWithoutResult.callObserver(CoprocessorHost.java:540)
> at
> org.apache.hadoop.hbase.coprocessor.CoprocessorHost.execOperation(CoprocessorHost.java:614)
> at
> org.apache.hadoop.hbase.regionserver.RegionCoprocessorHost.postBatchMutateIndispensably(RegionCoprocessorHost.java:1034)
> at
> org.apache.hadoop.hbase.regionserver.HRegion$MutationBatchOperation.doPostOpCleanupForMiniBatch(HRegion.java:3533)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.doMiniBatchMutate(HRegion.java:3914)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.batchMutate(HRegion.java:3822)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.batchMutate(HRegion.java:3753)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.doBatchOp(RSRpcServices.java:1027)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.doNonAtomicBatchOp(RSRpcServices.java:959)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.doNonAtomicRegionMutation(RSRpcServices.java:922)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.multi(RSRpcServices.java:2666)
> at
> org.apache.hadoop.hbase.shaded.protobuf.generated.ClientProtos$ClientService$2.callBlockingMethod(ClientProtos.java:42014)
> at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:409)
> at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:130)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:324)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:304)
> Caused by: java.lang.reflect.UndeclaredThrowableException
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1761)
> at
> org.apache.hadoop.security.SecurityUtil.doAsUser(SecurityUtil.java:448)
> at
> org.apache.hadoop.security.SecurityUtil.doAsLoginUser(SecurityUtil.java:429)
> at sun.reflect.GeneratedMethodAccessor28.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:497)
> at org.apache.hadoop.hbase.util.Methods.call(Methods.java:40)
> at org.apache.hadoop.hbase.security.User.runAsLoginUser(User.java:183)
> ... 25 more
> Caused by: java.lang.Exception: java.lang.OutOfMemoryError: unable to create
> new native thread
> at
> org.apache.phoenix.index.PhoenixIndexFailurePolicy$1.run(PhoenixIndexFailurePolicy.java:266)
> at
> org.apache.phoenix.index.PhoenixIndexFailurePolicy$1.run(PhoenixIndexFailurePolicy.java:217)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1746)
> ... 32 more
> Caused by: java.lang.OutOfMemoryError: unable to create new native thread
> at java.lang.Thread.start0(Native Method)
> at java.lang.Thread.start(Thread.java:714)
> at
> java.util.concurrent.ThreadPoolExecutor.addWorker(ThreadPoolExecutor.java:950)
> at
> java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1357)
> at
> java.util.concurrent.AbstractExecutorService.submit(AbstractExecutorService.java:134)
> at
> org.apache.hadoop.hbase.client.HTable.coprocessorService(HTable.java:1007)
> at
> org.apache.hadoop.hbase.client.HTable.coprocessorService(HTable.java:986)
> at
> org.apache.phoenix.util.IndexUtil.updateIndexState(IndexUtil.java:724)
> at
> org.apache.phoenix.util.IndexUtil.updateIndexState(IndexUtil.java:709)
> at
> org.apache.phoenix.index.PhoenixIndexFailurePolicy$1.run(PhoenixIndexFailurePolicy.java:236)
> ... 36 more
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
