Return-Path: X-Original-To: archive-asf-public-internal@cust-asf2.ponee.io Delivered-To: archive-asf-public-internal@cust-asf2.ponee.io Received: from cust-asf.ponee.io (cust-asf.ponee.io [163.172.22.183]) by cust-asf2.ponee.io (Postfix) with ESMTP id 95B5D200B8C for ; Mon, 12 Sep 2016 23:56:23 +0200 (CEST) Received: by cust-asf.ponee.io (Postfix) id 940FD160AC8; Mon, 12 Sep 2016 21:56:23 +0000 (UTC) Delivered-To: archive-asf-public@cust-asf.ponee.io Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by cust-asf.ponee.io (Postfix) with SMTP id 68CB6160AB2 for ; Mon, 12 Sep 2016 23:56:22 +0200 (CEST) Received: (qmail 63034 invoked by uid 500); 12 Sep 2016 21:55:24 -0000 Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Delivered-To: mailing list issues@hbase.apache.org Received: (qmail 62592 invoked by uid 99); 12 Sep 2016 21:55:24 -0000 Received: from arcas.apache.org (HELO arcas) (140.211.11.28) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 12 Sep 2016 21:55:24 +0000 Received: from arcas.apache.org (localhost [127.0.0.1]) by arcas (Postfix) with ESMTP id 471F52C1B91 for ; Mon, 12 Sep 2016 21:55:21 +0000 (UTC) Date: Mon, 12 Sep 2016 21:55:21 +0000 (UTC) From: "Enis Soztutar (JIRA)" To: issues@hbase.apache.org Message-ID: In-Reply-To: References: Subject: [jira] [Comment Edited] (HBASE-16604) Scanner retries on IOException can cause the scans to miss data MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-JIRA-FingerPrint: 30527f35849b9dde25b450d4833f0394 archived-at: Mon, 12 Sep 2016 21:56:23 -0000 [ https://issues.apache.org/jira/browse/HBASE-16604?page=3Dcom.atlassia= n.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D154= 85409#comment-15485409 ]=20 Enis Soztutar edited comment on HBASE-16604 at 9/12/16 9:55 PM: ---------------------------------------------------------------- Some more info. KeyValueHeap.generalizedSeek() will leave the heap in "dirty" state by sett= ing the {{current =3D null}} if it gets an IOException:=20 {code} private boolean generalizedSeek(boolean isLazy, Cell seekKey, boolean forward, boolean useBloom) throws IOException { if (!isLazy && useBloom) { throw new IllegalArgumentException("Multi-column Bloom filter " + "optimization requires a lazy seek"); } if (current =3D=3D null) { return false; } heap.add(current); current =3D null; ... =20 {code} On the next call, this will return false, indicating that there are no more= values to return.=20 We can deal with this in a couple of different ways:=20 (1) Handle IOExceptions in individual KVHeap methods and make sure that sta= te is left consistent even in case of IOExceptions.=20 (2) Handle IOExceptions in HRegionScannerImpl and reset the whole RegionSca= nner state before returning=20 (3) Bubble the exception to the client, but make sure that the scanner is t= hrown away. The client will restart another RegionScanner, and possibly sta= rt from scanning from the start of the row throwing away partial results.= =20 I think, doing (1) will be very fragile. (2) also will not work, since ther= e should be a way to reset the scanner state reliable in case of an IOExcep= tion coming deep down from FS layer. If we are doing partial results, we ma= ybe left in the middle of a row, but with partially seek'ed. Thus I think (= 2) also won't cut.=20 (3) is the simplest and most reliable, which would reset the scanner back t= o the start of the row, and makes sure that the ScannerCallable returns. Th= e challenge with (3) is that, we want the ScannerCallable to not retry, but= we want the ClientScanner to retry. ClientScanner only handles a couple of= known exceptions which are derivatives of DNRIOE (UnknownScannerException,= NotServingRegionException, OutOfOrderScannerNextException, etc). We can in= troduce another exception type (ResetScannerException), but we have to be c= areful for BC for existing clients.=20 was (Author: enis): Some more info. KeyValueHeap.generalizedSeek() will leave the heap in "dirty" state by sett= ing the {{current =3D null}} if it gets an IOException:=20 {code } private boolean generalizedSeek(boolean isLazy, Cell seekKey, boolean forward, boolean useBloom) throws IOException { if (!isLazy && useBloom) { throw new IllegalArgumentException("Multi-column Bloom filter " + "optimization requires a lazy seek"); } if (current =3D=3D null) { return false; } heap.add(current); current =3D null; ... =20 {code} On the next call, this will return false, indicating that there are no more= values to return.=20 We can deal with this in a couple of different ways:=20 (1) Handle IOExceptions in individual KVHeap methods and make sure that sta= te is left consistent even in case of IOExceptions.=20 (2) Handle IOExceptions in HRegionScannerImpl and reset the whole RegionSca= nner state before returning=20 (3) Bubble the exception to the client, but make sure that the scanner is t= hrown away. The client will restart another RegionScanner, and possibly sta= rt from scanning from the start of the row throwing away partial results.= =20 I think, doing (1) will be very fragile. (2) also will not work, since ther= e should be a way to reset the scanner state reliable in case of an IOExcep= tion coming deep down from FS layer. If we are doing partial results, we ma= ybe left in the middle of a row, but with partially seek'ed. Thus I think (= 2) also won't cut.=20 (3) is the simplest, which would reset the scanner back to the start of the= row, and makes sure that the ScannerCallable returns. The challenge with (= 3) is that, we want the ScannerCallable to not retry, but we want the Clien= tScanner to retry. ClientScanner only handles a couple of known exceptions = which are derivatives of DNRIOE (UnknownScannerException, NotServingRegionE= xception, OutOfOrderScannerNextException, etc). We can introduce another ex= ception type (ResetScannerException), but we have to be careful for BC for = existing clients.=20 > Scanner retries on IOException can cause the scans to miss data=20 > ---------------------------------------------------------------- > > Key: HBASE-16604 > URL: https://issues.apache.org/jira/browse/HBASE-16604 > Project: HBase > Issue Type: Bug > Components: regionserver, Scanners > Reporter: Enis Soztutar > Assignee: Enis Soztutar > Fix For: 2.0.0, 1.3.0, 1.4.0, 1.1.7, 1.2.4 > > > Debugging an ITBLL failure, where the Verify did not "see" all the data i= n the cluster, I've noticed that if we end up getting a generic IOException= from the HFileReader level, we may end up missing the rest of the data in = the region. I was able to manually test this, and this stack trace helps to= understand what is going on:=20 > {code} > 2016-09-09 16:27:15,633 INFO [hconnection-0x71ad3d8a-shared--pool21-t9] = client.ScannerCallable(376): Open scanner=3D1 for scan=3D{"loadColumnFamili= esOnDemand":null,"startRow":"","stopRow":"","batch":-1,"cacheBlocks":true,"= totalColumns":1,"maxResultSize":2097152,"families":{"testFamily":["testFami= ly"]},"caching":100,"maxVersions":1,"timeRange":[0,9223372036854775807]} on= region region=3DtestScanThrowsException,,1473463632707.b2adfb618e5d0fe225c= 1dc40c0eabfee., hostname=3Dhw10676,51833,1473463626529, seqNum=3D2 > 2016-09-09 16:27:15,634 INFO [B.fifo.QRpcServer.handler=3D5,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2196): scan request:scanner_id: 1 nu= mber_of_rows: 100 close_scanner: false next_call_seq: 0 client_handles_part= ials: true client_handles_heartbeats: true renew: false > 2016-09-09 16:27:15,635 INFO [B.fifo.QRpcServer.handler=3D5,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2510): Rolling back next call seqId > 2016-09-09 16:27:15,635 INFO [B.fifo.QRpcServer.handler=3D5,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2565): Throwing new ServiceException= java.io.IOException: Could not reseek StoreFileScanner[HFileScanner for rea= der reader=3Dhdfs://localhost:51795/user/enis/test-data/d6fb1c70-93c1-4099-= acb7-5723fc05a737/data/default/testScanThrowsException/b2adfb618e5d0fe225c1= dc40c0eabfee/testFamily/5a213cc23b714e5e8e1a140ebbe72f2c, compression=3Dnon= e, cacheConf=3DblockCache=3DLruBlockCache{blockCount=3D0, currentSize=3D156= 7264, freeSize=3D1525578848, maxSize=3D1527146112, heapSize=3D1567264, minS= ize=3D1450788736, minFactor=3D0.95, multiSize=3D725394368, multiFactor=3D0.= 5, singleSize=3D362697184, singleFactor=3D0.25}, cacheDataOnRead=3Dtrue, ca= cheDataOnWrite=3Dfalse, cacheIndexesOnWrite=3Dfalse, cacheBloomsOnWrite=3Df= alse, cacheEvictOnClose=3Dfalse, cacheDataCompressed=3Dfalse, prefetchOnOpe= n=3Dfalse, firstKey=3Daaa/testFamily:testFamily/1473463633859/Put, lastKey= =3Dzzz/testFamily:testFamily/1473463634271/Put, avgKeyLen=3D35, avgValueLen= =3D3, entries=3D17576, length=3D866998, cur=3D/testFamily:/OLDEST_TIMESTAMP= /Minimum/vlen=3D0/seqid=3D0] to key /testFamily:testFamily/LATEST_TIMESTAMP= /Maximum/vlen=3D0/seqid=3D0 > 2016-09-09 16:27:15,635 DEBUG [B.fifo.QRpcServer.handler=3D5,queue=3D0,po= rt=3D51833] ipc.CallRunner(110): B.fifo.QRpcServer.handler=3D5,queue=3D0,po= rt=3D51833: callId: 26 service: ClientService methodName: Scan size: 26 con= nection: 192.168.42.75:51903 > java.io.IOException: Could not reseek StoreFileScanner[HFileScanner for r= eader reader=3Dhdfs://localhost:51795/user/enis/test-data/d6fb1c70-93c1-409= 9-acb7-5723fc05a737/data/default/testScanThrowsException/b2adfb618e5d0fe225= c1dc40c0eabfee/testFamily/5a213cc23b714e5e8e1a140ebbe72f2c, compression=3Dn= one, cacheConf=3DblockCache=3DLruBlockCache{blockCount=3D0, currentSize=3D1= 567264, freeSize=3D1525578848, maxSize=3D1527146112, heapSize=3D1567264, mi= nSize=3D1450788736, minFactor=3D0.95, multiSize=3D725394368, multiFactor=3D= 0.5, singleSize=3D362697184, singleFactor=3D0.25}, cacheDataOnRead=3Dtrue, = cacheDataOnWrite=3Dfalse, cacheIndexesOnWrite=3Dfalse, cacheBloomsOnWrite= =3Dfalse, cacheEvictOnClose=3Dfalse, cacheDataCompressed=3Dfalse, prefetchO= nOpen=3Dfalse, firstKey=3Daaa/testFamily:testFamily/1473463633859/Put, last= Key=3Dzzz/testFamily:testFamily/1473463634271/Put, avgKeyLen=3D35, avgValue= Len=3D3, entries=3D17576, length=3D866998, cur=3D/testFamily:/OLDEST_TIMEST= AMP/Minimum/vlen=3D0/seqid=3D0] to key /testFamily:testFamily/LATEST_TIMEST= AMP/Maximum/vlen=3D0/seqid=3D0 > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:224) > =09at org.apache.hadoop.hbase.regionserver.NonLazyKeyValueScanner.doRealS= eek(NonLazyKeyValueScanner.java:55) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.generalizedSeek(K= eyValueHeap.java:312) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.requestSeek(KeyVa= lueHeap.java:268) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.reseek(StoreScann= er.java:815) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.seekAsDirection(S= toreScanner.java:803) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.next(StoreScanner= .java:625) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.next(KeyValueHeap= .java:147) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.popu= lateResult(HRegion.java:5736) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Internal(HRegion.java:5887) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Raw(HRegion.java:5674) > =09at org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServic= es.java:2434) > =09at org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientServi= ce$2.callBlockingMethod(ClientProtos.java:32295) > =09at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2127) > =09at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:107) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor.consumerLoop(RpcExecutor.ja= va:133) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor$2.run(RpcExecutor.java:108) > =09at java.lang.Thread.run(Thread.java:745) > Caused by: java.io.IOException: foo bar baz > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:207) > =09... 17 more > 2016-09-09 16:27:15,653 INFO [hconnection-0x71ad3d8a-shared--pool21-t10]= client.ScannerCallable(261): Got exception making request scanner_id: 1 nu= mber_of_rows: 100 close_scanner: false next_call_seq: 0 client_handles_part= ials: true client_handles_heartbeats: true renew: false to region=3DtestSca= nThrowsException,,1473463632707.b2adfb618e5d0fe225c1dc40c0eabfee., hostname= =3Dhw10676,51833,1473463626529, seqNum=3D2 > java.io.IOException: java.io.IOException: Could not reseek StoreFileScann= er[HFileScanner for reader reader=3Dhdfs://localhost:51795/user/enis/test-d= ata/d6fb1c70-93c1-4099-acb7-5723fc05a737/data/default/testScanThrowsExcepti= on/b2adfb618e5d0fe225c1dc40c0eabfee/testFamily/5a213cc23b714e5e8e1a140ebbe7= 2f2c, compression=3Dnone, cacheConf=3DblockCache=3DLruBlockCache{blockCount= =3D0, currentSize=3D1567264, freeSize=3D1525578848, maxSize=3D1527146112, h= eapSize=3D1567264, minSize=3D1450788736, minFactor=3D0.95, multiSize=3D7253= 94368, multiFactor=3D0.5, singleSize=3D362697184, singleFactor=3D0.25}, cac= heDataOnRead=3Dtrue, cacheDataOnWrite=3Dfalse, cacheIndexesOnWrite=3Dfalse,= cacheBloomsOnWrite=3Dfalse, cacheEvictOnClose=3Dfalse, cacheDataCompressed= =3Dfalse, prefetchOnOpen=3Dfalse, firstKey=3Daaa/testFamily:testFamily/1473= 463633859/Put, lastKey=3Dzzz/testFamily:testFamily/1473463634271/Put, avgKe= yLen=3D35, avgValueLen=3D3, entries=3D17576, length=3D866998, cur=3D/testFa= mily:/OLDEST_TIMESTAMP/Minimum/vlen=3D0/seqid=3D0] to key /testFamily:testF= amily/LATEST_TIMESTAMP/Maximum/vlen=3D0/seqid=3D0 > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:224) > =09at org.apache.hadoop.hbase.regionserver.NonLazyKeyValueScanner.doRealS= eek(NonLazyKeyValueScanner.java:55) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.generalizedSeek(K= eyValueHeap.java:312) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.requestSeek(KeyVa= lueHeap.java:268) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.reseek(StoreScann= er.java:815) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.seekAsDirection(S= toreScanner.java:803) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.next(StoreScanner= .java:625) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.next(KeyValueHeap= .java:147) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.popu= lateResult(HRegion.java:5736) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Internal(HRegion.java:5887) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Raw(HRegion.java:5674) > =09at org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServic= es.java:2434) > =09at org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientServi= ce$2.callBlockingMethod(ClientProtos.java:32295) > =09at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2127) > =09at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:107) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor.consumerLoop(RpcExecutor.ja= va:133) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor$2.run(RpcExecutor.java:108) > =09at java.lang.Thread.run(Thread.java:745) > Caused by: java.io.IOException: foo bar baz > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:207) > =09... 17 more > =09at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Metho= d) > =09at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstru= ctorAccessorImpl.java:62) > =09at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(Delegatin= gConstructorAccessorImpl.java:45) > =09at java.lang.reflect.Constructor.newInstance(Constructor.java:422) > =09at org.apache.hadoop.ipc.RemoteException.instantiateException(RemoteEx= ception.java:106) > =09at org.apache.hadoop.ipc.RemoteException.unwrapRemoteException(RemoteE= xception.java:95) > =09at org.apache.hadoop.hbase.protobuf.ProtobufUtil.getRemoteException(Pr= otobufUtil.java:332) > =09at org.apache.hadoop.hbase.client.ScannerCallable.call(ScannerCallable= .java:256) > =09at org.apache.hadoop.hbase.client.ScannerCallable.call(ScannerCallable= .java:1) > =09at org.apache.hadoop.hbase.client.RpcRetryingCaller.callWithoutRetries= (RpcRetryingCaller.java:200) > =09at org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$Retrying= RPC.call(ScannerCallableWithReplicas.java:364) > =09at org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$Retrying= RPC.call(ScannerCallableWithReplicas.java:1) > =09at org.apache.hadoop.hbase.client.RpcRetryingCaller.callWithRetries(Rp= cRetryingCaller.java:126) > =09at org.apache.hadoop.hbase.client.ResultBoundedCompletionService$Queue= ingFuture.run(ResultBoundedCompletionService.java:65) > =09at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecuto= r.java:1142) > =09at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecut= or.java:617) > =09at java.lang.Thread.run(Thread.java:745) > Caused by: org.apache.hadoop.hbase.ipc.RemoteWithExtrasException(java.io.= IOException): java.io.IOException: Could not reseek StoreFileScanner[HFileS= canner for reader reader=3Dhdfs://localhost:51795/user/enis/test-data/d6fb1= c70-93c1-4099-acb7-5723fc05a737/data/default/testScanThrowsException/b2adfb= 618e5d0fe225c1dc40c0eabfee/testFamily/5a213cc23b714e5e8e1a140ebbe72f2c, com= pression=3Dnone, cacheConf=3DblockCache=3DLruBlockCache{blockCount=3D0, cur= rentSize=3D1567264, freeSize=3D1525578848, maxSize=3D1527146112, heapSize= =3D1567264, minSize=3D1450788736, minFactor=3D0.95, multiSize=3D725394368, = multiFactor=3D0.5, singleSize=3D362697184, singleFactor=3D0.25}, cacheDataO= nRead=3Dtrue, cacheDataOnWrite=3Dfalse, cacheIndexesOnWrite=3Dfalse, cacheB= loomsOnWrite=3Dfalse, cacheEvictOnClose=3Dfalse, cacheDataCompressed=3Dfals= e, prefetchOnOpen=3Dfalse, firstKey=3Daaa/testFamily:testFamily/14734636338= 59/Put, lastKey=3Dzzz/testFamily:testFamily/1473463634271/Put, avgKeyLen=3D= 35, avgValueLen=3D3, entries=3D17576, length=3D866998, cur=3D/testFamily:/O= LDEST_TIMESTAMP/Minimum/vlen=3D0/seqid=3D0] to key /testFamily:testFamily/L= ATEST_TIMESTAMP/Maximum/vlen=3D0/seqid=3D0 > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:224) > =09at org.apache.hadoop.hbase.regionserver.NonLazyKeyValueScanner.doRealS= eek(NonLazyKeyValueScanner.java:55) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.generalizedSeek(K= eyValueHeap.java:312) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.requestSeek(KeyVa= lueHeap.java:268) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.reseek(StoreScann= er.java:815) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.seekAsDirection(S= toreScanner.java:803) > =09at org.apache.hadoop.hbase.regionserver.StoreScanner.next(StoreScanner= .java:625) > =09at org.apache.hadoop.hbase.regionserver.KeyValueHeap.next(KeyValueHeap= .java:147) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.popu= lateResult(HRegion.java:5736) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Internal(HRegion.java:5887) > =09at org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.next= Raw(HRegion.java:5674) > =09at org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServic= es.java:2434) > =09at org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientServi= ce$2.callBlockingMethod(ClientProtos.java:32295) > =09at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2127) > =09at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:107) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor.consumerLoop(RpcExecutor.ja= va:133) > =09at org.apache.hadoop.hbase.ipc.RpcExecutor$2.run(RpcExecutor.java:108) > =09at java.lang.Thread.run(Thread.java:745) > Caused by: java.io.IOException: foo bar baz > =09at org.apache.hadoop.hbase.regionserver.StoreFileScanner.reseek(StoreF= ileScanner.java:207) > =09... 17 more > =09at org.apache.hadoop.hbase.ipc.RpcClientImpl.call(RpcClientImpl.java:1= 225) > =09at org.apache.hadoop.hbase.ipc.AbstractRpcClient.callBlockingMethod(Ab= stractRpcClient.java:213) > =09at org.apache.hadoop.hbase.ipc.AbstractRpcClient$BlockingRpcChannelImp= lementation.callBlockingMethod(AbstractRpcClient.java:287) > =09at org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientServi= ce$BlockingStub.scan(ClientProtos.java:32741) > =09at org.apache.hadoop.hbase.client.ScannerCallable.call(ScannerCallable= .java:214) > =09... 9 more > 2016-09-09 16:27:15,963 INFO [B.priority.fifo.QRpcServer.handler=3D8,que= ue=3D0,port=3D51824] regionserver.RSRpcServices(2196): scan request:region = { type: REGION_NAME value: "hbase:meta,,1" } scan { start_row: "testScanThr= owsException,,99999999999999" max_versions: 1 cache_blocks: true small: tru= e reversed: true caching: 1 } number_of_rows: 1 close_scanner: true client_= handles_partials: true client_handles_heartbeats: true > 2016-09-09 16:27:15,965 INFO [B.priority.fifo.QRpcServer.handler=3D8,que= ue=3D0,port=3D51824] regionserver.RSRpcServices(2549): class org.apache.had= oop.hbase.protobuf.generated.ClientProtos$ScanResponse:cells_per_result: 4 = scanner_id: 15 more_results: false stale: false partial_flag_per_result: fa= lse more_results_in_region: true heartbeat_message: false > 2016-09-09 16:27:15,967 INFO [B.fifo.QRpcServer.handler=3D0,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2196): scan request:scanner_id: 1 nu= mber_of_rows: 100 close_scanner: false next_call_seq: 0 client_handles_part= ials: true client_handles_heartbeats: true renew: false > 2016-09-09 16:27:15,967 INFO [B.fifo.QRpcServer.handler=3D0,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2549): class org.apache.hadoop.hbase= .protobuf.generated.ClientProtos$ScanResponse:scanner_id: 1 more_results: t= rue ttl: 60000 stale: false more_results_in_region: false > 2016-09-09 16:27:15,968 INFO [B.fifo.QRpcServer.handler=3D1,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2196): scan request:scanner_id: 1 nu= mber_of_rows: 0 close_scanner: true client_handles_partials: true client_ha= ndles_heartbeats: true > 2016-09-09 16:27:15,968 INFO [B.fifo.QRpcServer.handler=3D1,queue=3D0,po= rt=3D51833] regionserver.RSRpcServices(2549): class org.apache.hadoop.hbase= .protobuf.generated.ClientProtos$ScanResponse:scanner_id: 1 more_results: f= alse > 2016-09-09 16:27:15,971 INFO [main] hbase.HBaseTestingUtility(1111): Shu= tting down minicluster > {code} > The ScannerCallable is retryable if it gets a IOException. Before the HBA= SE-13662 fix, the retry was still failing due to OutOfOrderScannerException= , which then causes the Scanner stack to be reset, thus masking the issue. = After HBASE-13662, we can retry the same scan, but in case of we are doing = a {{reseek()}} which fails in middle with an IOException, the KVHeaps and t= he Scanner state on the server side is left dirty.=20 > It seems to be much safer to reset the scanner state in case of IOExcepti= ons.=20 -- This message was sent by Atlassian JIRA (v6.3.4#6332)