Tomasz Grygo created IGNITE-13093: ------------------------------------- Summary: Unidentified Apache Ignite worker blocked when inserting large amount of records to the persistent storage Key: IGNITE-13093 URL: https://issues.apache.org/jira/browse/IGNITE-13093 Project: Ignite Issue Type: Bug Components: cache Affects Versions: 2.8.1 Environment: Java 1.8.0_231 Apache Ignite 2.8.1 Windows 10, 64G memory
<?xml version="1.0" encoding="UTF-8"?> <beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation=" http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd"> <bean class="org.apache.ignite.configuration.IgniteConfiguration"> <property name="gridLogger"> <bean class="org.apache.ignite.logger.log4j2.Log4J2Logger"> <constructor-arg type="java.lang.String" value="log4j2.xml"/> </bean> </property> <property name="communicationSpi"> <bean class="org.apache.ignite.spi.communication.tcp.TcpCommunicationSpi"> <!-- Override local port. --> <property name="localPort" value="47400"/> </bean> </property> <!-- Configure internal thread pool. 64--> <property name="publicThreadPoolSize" value="16"/> <!-- Configure system thread pool. 32--> <property name="systemThreadPoolSize" value="8"/> <property name="systemWorkerBlockedTimeout" value="#{5 * 60 * 1000}"/> <property name="failureHandler"> <bean class="org.apache.ignite.failure.StopNodeFailureHandler"> <!-- uncomment to enable this handler to process critical workers' hung-ups --> <property name="ignoredFailureTypes"> <list> </list> </property> </bean> </property> <!-- Set to true to enable distributed class loading for examples, default is false. --> <property name="peerClassLoadingEnabled" value="false"/> <property name="dataStorageConfiguration"> <bean class="org.apache.ignite.configuration.DataStorageConfiguration"> <!-- Sets a path to the root directory where data and indexes are to be persisted. It's assumed the directory is on a separated SSD. --> <property name="storagePath" value="persistence"/> <!-- Sets a path to the directory where WAL is stored. It's assumed the directory is on a separated HDD. --> <property name="walPath" value="wal"/> <!-- Sets a path to the directory where WAL archive is stored. The directory is on the same HDD as the WAL. --> <property name="walArchivePath" value="wal/archive"/> <!-- Changing WAL Mode. --> <property name="walMode" value="NONE"/> <!-- Set the page size to 4 KB, default --> <!-- limit 1 KB - 16 KB --> <property name="pageSize" value="#{4 * 1024}"/> <!-- Enable write throttling. --> <property name="writeThrottlingEnabled" value="false"/> <property name="checkpointFrequency" value="500"/> <property name="lockWaitTime" value="2000"/> <property name="checkpointThreads" value="1"/> <property name="checkpointWriteOrder" value="RANDOM"/> <!-- Default memory region that grows endlessly. A cache is bound to this memory region unless it sets another one in its CacheConfiguration. --> <property name="defaultDataRegionConfiguration"> <bean class="org.apache.ignite.configuration.DataRegionConfiguration"> <!--property name="name" value="Default_Region"/--> <property name="name" value="default"/> <!-- 100 MB memory region with disabled eviction --> <property name="initialSize" value="#{100L * 1024 * 1024}"/> <!-- maxSize 20 MB is too little --> <!-- sum of all maxSize values has to be less than total memory of the system --> <!-- limits size in memory, not on disk --> <!-- default value 1.2GB --> <property name="maxSize" value="#{2L * 1024 * 1024 * 1024}"/> <property name="persistenceEnabled" value="true"/> <!-- Increasing the buffer size to 1 GB. --> <property name="checkpointPageBufferSize" value="#{1L * 1024 * 1024 * 1024}"/> </bean> </property> <!-- Defining several data regions for different memory regions --> <property name="dataRegionConfigurations"> <list> </list> </property> </bean> </property> <property name="cacheConfiguration"> <list> <!--bean class="org.apache.ignite.configuration.CacheConfiguration"> <property name="dataRegionName" value="default"/> <property name="name" value=".ShardDetectorStorage"/> <property name="onheapCacheEnabled" value="true"/> </bean> <bean class="org.apache.ignite.configuration.CacheConfiguration"> <property name="dataRegionName" value="default"/> <property name="name" value=".ChildrenStore"/> <property name="onheapCacheEnabled" value="true"/> </bean> <bean class="org.apache.ignite.configuration.CacheConfiguration"> <property name="dataRegionName" value="default"/> <property name="name" value=".ChildrenStore.listsize"/> <property name="onheapCacheEnabled" value="true"/> </bean> <bean class="org.apache.ignite.configuration.CacheConfiguration"> <property name="dataRegionName" value="default"/> <property name="name" value=".RootStorage"/> <property name="onheapCacheEnabled" value="true"/> </bean> <bean class="org.apache.ignite.configuration.CacheConfiguration"> <property name="dataRegionName" value="default"/> <property name="name" value=".QualifierStorage"/> <property name="onheapCacheEnabled" value="true"/> </bean--> </list> </property> </bean> </beans> Java settings -Xms1024m -Xmx50g -Xss1024m -Xverify:none -server -DIGNITE_QUIET=true -XX:+UseG1GC -XX:+DisableExplicitGC -Djava.net.preferIPv4Stack=true -XX:+AlwaysPreTouch -XX:+ScavengeBeforeFullGC -XX:+AggressiveOpts partial thread dump during slowdown "db-checkpoint-thread-#54" #99 prio=5 os_prio=0 tid=0x0000000070344800 nid=0x2d54 runnable [0x0000001c5df3e000] java.lang.Thread.State: RUNNABLE at org.apache.ignite.internal.processors.cache.persistence.pagemem.PageMemoryImpl$Segment.removePageForReplacement(PageMemoryImpl.java:2398) at org.apache.ignite.internal.processors.cache.persistence.pagemem.PageMemoryImpl$Segment.access$900(PageMemoryImpl.java:2093) at org.apache.ignite.internal.processors.cache.persistence.pagemem.PageMemoryImpl.acquirePage(PageMemoryImpl.java:773) at org.apache.ignite.internal.processors.cache.persistence.pagemem.PageMemoryImpl.acquirePage(PageMemoryImpl.java:701) at org.apache.ignite.internal.processors.cache.persistence.GridCacheOffheapManager.saveStoreMetadata(GridCacheOffheapManager.java:342) at org.apache.ignite.internal.processors.cache.persistence.GridCacheOffheapManager.syncMetadata(GridCacheOffheapManager.java:268) at org.apache.ignite.internal.processors.cache.persistence.GridCacheOffheapManager.syncMetadata(GridCacheOffheapManager.java:254) at org.apache.ignite.internal.processors.cache.persistence.GridCacheOffheapManager.beforeCheckpointBegin(GridCacheOffheapManager.java:226) at org.apache.ignite.internal.processors.cache.persistence.GridCacheDatabaseSharedManager$Checkpointer.markCheckpointBegin(GridCacheDatabaseSharedManager.java:4125) at org.apache.ignite.internal.processors.cache.persistence.GridCacheDatabaseSharedManager$Checkpointer.doCheckpoint(GridCacheDatabaseSharedManager.java:3738) at org.apache.ignite.internal.processors.cache.persistence.GridCacheDatabaseSharedManager$Checkpointer.body(GridCacheDatabaseSharedManager.java:3623) at org.apache.ignite.internal.util.worker.GridWorker.run(GridWorker.java:120) at java.lang.Thread.run(Thread.java:748) Locked ownable synchronizers: - <0x000000009b71a250> (a java.util.concurrent.locks.ReentrantReadWriteLock$NonfairSync) Reporter: Tomasz Grygo I'm looking at Apache Ignite to use as a fast database. Performance is very important, I need to build it as fast as possible with resources available. First I copy all (450M) records from my original test database to Ignite caches through IgniteDataStreams using PK as a key. Database does not fit in memory so I have disk persistence enabled and eviction disabled. Data is inserted in parallel using 8 threads. I have only one but fairly powerful Windows PC doing all the work, no separate Ignite cluster. I'm not interested in cache recovery so WAL is disabled. Everything goes well until I hit around 310 million entries (2 hours of work). At this point Ignite starts to choke, inserts slow down and then stop with exceptions. Exception is triggered by systemWorkerBlockedTimeout setting set to 5 minutes. Extending this time does not help at all. Based on heap dump I tried adding -DIGNITE_PAGES_LIST_DISABLE_ONHEAP_CACHING=true and it failed slightly later but still could not finish the job. I read the performance guides and I tried tweaking other Ignite settings too but didn't see any impact. How can if find which worker is being blocked and why? 2020-05-27 21:54:26,176 [Storage2 ] [ERROR] - DTR_0030 worker Storage2 had error: FATAL ERROR java.lang.IllegalStateException: Data streamer has been closed. java.lang.IllegalStateException: Data streamer has been closed. at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.closedException(DataStreamerImpl.java:1095) at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.lock(DataStreamerImpl.java:446) at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.addDataInternal(DataStreamerImpl.java:646) at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.addDataInternal(DataStreamerImpl.java:631) at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.addData(DataStreamerImpl.java:753) at com.sc.extr.cache.PureIgniteDynamicRowStorage.putIfAbsent(PureIgniteDynamicRowStorage.java:83) at com.sc.extr.cache.PureIgniteDynamicRowStorage.addRowOnKey(PureIgniteDynamicRowStorage.java:160) at com.sc.extr.tree.MultiCacheTreeBuilder.addRootRowToCache(MultiCacheTreeBuilder.java:409) at com.sc.extr.tree.MultiCacheTreeBuilder.parentRev1to1(MultiCacheTreeBuilder.java:237) at com.sc.extr.tree.MultiCacheTreeBuilder.addRowToCache(MultiCacheTreeBuilder.java:333) at com.sc.extr.tree.MultiCacheTreeBuilder.parentRev(MultiCacheTreeBuilder.java:274) at com.sc.extr.tree.MultiCacheTreeBuilder.addRow(MultiCacheTreeBuilder.java:379) at com.sc.extr.tree.MultiCacheTreeBuilder.process(MultiCacheTreeBuilder.java:206) at com.sc.bi.workflow.WorkTransformer.processOne(WorkTransformer.java:84) at com.sc.bi.workflow.WorkTransformer.doWork(WorkTransformer.java:145) at com.sc.bi.workflow.WorkTransformer.processQueue(WorkTransformer.java:210) at com.sc.bi.workflow.WorkTransformer.run(WorkTransformer.java:169) Caused by: class org.apache.ignite.IgniteCheckedException: Data streamer has been cancelled: DataStreamerImpl [bufLdrSzPerThread=4096, rcvr=org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl$IsolatedUpdater@381b03ed, ioPlcRslvr=null, cacheName=PERSON.PTINTN, bufSize=512, parallelOps=0, timeout=-1, autoFlushFreq=0, bufMappings=ConcurrentHashMap {03e74462-12ec-4140-b9fb-a975572ac3bb=Buffer [node=TcpDiscoveryNode [id=03e74462-12ec-4140-b9fb-a975572ac3bb, consistentId=b01eb38b-7728-4e43-a697-0bc52f872e44, addrs=ArrayList [127.0.0.1, 172.27.179.112], sockAddrs=HashSet [SOFTBI-DEV.sc.com/172.27.179.112:47500, /127.0.0.1:47500], discPort=47500, order=1, intOrder=1, lastExchangeTime=1590614830815, loc=true, ver=2.8.1#20200521-sha1:86422096, isClient=false], isLocNode=true, idGen=0, sem=java.util.concurrent.Semaphore@2a869d9[Permits = 64], perNodeParallelOps=64, entriesCnt=2048, locFutsSize=0, reqsSize=0]}, cacheObjProc=GridProcessorAdapter [], cacheObjCtx=org.apache.ignite.internal.processors.cache.CacheObjectContext@2a5313b0, cancelled=true, cancellationReason=null, failCntr=0, activeFuts=GridConcurrentHashSet [GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=2102798044], GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=1195632760], GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=370791970], GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=420732031], GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=1453517070]], jobPda=null, depCls=null, fut=DataStreamerFuture [super=GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=1165180540]], publicFut=IgniteFuture [orig=DataStreamerFuture [super=GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, hash=1165180540]]], disconnectErr=null, closed=true, lastFlushTime=1590629894701, skipStore=false, keepBinary=false, maxRemapCnt=32, remapSem=java.util.concurrent.Semaphore@6e6f060b[Permits = 2147483647], remapOwning=false] at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.closeEx(DataStreamerImpl.java:1347) at org.apache.ignite.internal.processors.datastreamer.DataStreamerImpl.closeEx(DataStreamerImpl.java:1318) at org.apache.ignite.internal.processors.datastreamer.DataStreamProcessor.onKernalStop(DataStreamProcessor.java:155) at org.apache.ignite.internal.IgniteKernal.stop0(IgniteKernal.java:2551) at org.apache.ignite.internal.IgniteKernal.stop(IgniteKernal.java:2499) at org.apache.ignite.internal.IgnitionEx$IgniteNamedInstance.stop0(IgnitionEx.java:2650) at org.apache.ignite.internal.IgnitionEx$IgniteNamedInstance.stop(IgnitionEx.java:2613) at org.apache.ignite.internal.IgnitionEx.stop(IgnitionEx.java:339) at org.apache.ignite.failure.StopNodeFailureHandler$1.run(StopNodeFailureHandler.java:36) at java.lang.Thread.run(Thread.java:748) -- This message was sent by Atlassian Jira (v8.3.4#803005)