为什么 Apache Ignite 使用的内存比配置的多
Why does Apache Ignite use more memory than configured
在具有 16GB 内存的 Linux 主机上使用 Ignite 2.8.1-1 版本和默认配置（1GB 堆，20% 默认区域用于堆外存储和启用持久性）时，我注意到 ignite进程最多可以使用 11GB 的内存（通过检查顶部进程使用的内存常驻大小来验证，见附件）。当我检查日志中的指标时，消耗的内存（堆+堆外）加起来不接近 7GB。一种可能性是检查点缓冲区可以使用额外的内存，但默认情况下应该是默认区域的 1/4，即只有大约 0.25 * 0.2 * 16GB 的四分之一。
关于剩余内存的用途的任何提示？
谢谢！
Metrics for local node (to disable set 'metricsLogFrequency' to 0)
^-- Node [id=a52857f2, name=dummy-ignite-node, uptime=6 days, 10:41:45.541]
^-- H/N/C [hosts=98, nodes=221, CPUs=1481]
^-- CPU [cur=16.17%, avg=16.02%, GC=0%]
^-- PageMemory [pages=788278]
^-- Heap [used=1474MB, free=58.24%, comm=2020MB]
^-- Off-heap [used=3115MB, free=10.39%, comm=3376MB]
^-- sysMemPlc region [used=0MB, free=99.99%, comm=100MB]
^-- default region [used=3115MB, free=1.93%, comm=3176MB]
^-- metastoreMemPlc region [used=0MB, free=99.73%, comm=0MB]
^-- TxLog region [used=0MB, free=100%, comm=100MB]
^-- Ignite persistence [used=57084MB]
^-- sysMemPlc region [used=0MB]
^-- default region [used=57084MB]
^-- metastoreMemPlc region [used=0MB]
^-- TxLog region [used=0MB]
^-- Outbound messages queue [size=0]
^-- Public thread pool [active=0, idle=0, qSize=0]
^-- System thread pool [active=0, idle=8, qSize=0]
===一些关于设置的日志===
dsCfg=DataStorageConfiguration [sysRegionInitSize=41943040, sysRegionMaxSize=104857600, pageSize=0, concLvl=0, dfltDataRegConf=DataRegionConfiguration [name=default, *maxSize=3330655027, initSize=268435456,* swapPath=null, pageEvictionMode=DISABLED, evictionThreshold=0.9, emptyPagesPoolSize=100, metricsEnabled=false, metricsSubIntervalCount=5, metricsRateTimeInterval=60000, persistenceEnabled=true, checkpointPageBufSize=0, lazyMemoryAllocation=true], dataRegions=null, storagePath=null, checkpointFreq=180000, lockWaitTime=10000, checkpointThreads=4, checkpointWriteOrder=SEQUENTIAL, walHistSize=20, maxWalArchiveSize=1073741824, walSegments=10, walSegmentSize=67108864, walPath=db/wal, walArchivePath=db/wal/archive, metricsEnabled=false, walMode=FSYNC, walTlbSize=131072, walBuffSize=0, walFlushFreq=2000, walFsyncDelay=1000, walRecordIterBuffSize=67108864, alwaysWriteFullPages=false, fileIOFactory=org.apache.ignite.internal.processors.cache.persistence.file.AsyncFileIOFactory@285c08c8, metricsSubIntervalCnt=5, metricsRateTimeInterval=60000, walAutoArchiveAfterInactivity=-1, writeThrottlingEnabled=false, walCompactionEnabled=false, walCompactionLevel=1, checkpointReadLockTimeout=null, walPageCompression=DISABLED, walPageCompressionLevel=null], activeOnStart=true, autoActivation=true, longQryWarnTimeout=3000, sqlConnCfg=null, cliConnCfg=ClientConnectorConfiguration [host=null, port=10800, portRange=100, sockSndBufSize=0, sockRcvBufSize=0, tcpNoDelay=true, maxOpenCursorsPerConn=128, threadPoolSize=8, idleTimeout=0, handshakeTimeout=10000, jdbcEnabled=true, odbcEnabled=true, thinCliEnabled=true, sslEnabled=false, useIgniteSslCtxFactory=true, sslClientAuth=false, sslCtxFactory=null, thinCliCfg=ThinClientConfiguration [maxActiveTxPerConn=100]], mvccVacuumThreadCnt=2, mvccVacuumFreq=5000, authEnabled=false, failureHnd=RestartProcessFailureHandler [super=AbstractFailureHandler [ignoredFailureTypes=UnmodifiableSet []]], commFailureRslvr=DefaultCommunicationFailureResolver []]
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Daemon mode: on
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] OS: Linux 3.10.0-1160.49.1.el7.x86_64 amd64
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] OS user: root
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] PID: 7181
[2022-01-27T16:41:44,228][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Language runtime: Scala ver. 2.11.12
[2022-01-27T16:41:44,228][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM information: OpenJDK Runtime Environment 1.8.0_312-b07 Red Hat, Inc. OpenJDK 64-Bit Server VM 25.312-b07
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM total memory: 0.96GB
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Remote Management [restart: off, REST: off, JMX (remote: off)]
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Logger: Log4J2Logger [quiet=true, config=config/ignite-log4j2.xml]
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] IGNITE_HOME=/usr/share/apache-ignite
[2022-01-27T16:41:44,230][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM arguments: [-XX:+AggressiveOpts, *-Xms1g, -Xmx1g, -XX:MaxPermSize=128M*, -Dfile.encoding=UTF-8, -DIGNITE_QUIET=true, -DIGNITE_HOME=/usr/share/apache-ignite, -DIGNITE_PROG_NAME=/usr/share/apache-ignite/bin/ignitevisorcmd.sh, -DIGNITE_DEPLOYMENT_MODE_OVERRIDE=ISOLATED]
Screen shot of top command output
是的，这里也考虑了检查点缓冲区大小，如果您没有覆盖默认值，它应该是 3GB/4 as you correctly highlighted. I wonder if it might be changed automatically since you have a lot more data ^-- Ignite persistence [used=57084MB] stored than the region capacity is - only 3GB. Also, this might be related to Direct Memory 使用，我认为 Java 没有计算在内堆使用。
无论如何，我认为最好明确检查 Ignite 内存指标，例如 data region and onheap 使用情况并详细检查它们。
在具有 16GB 内存的 Linux 主机上使用 Ignite 2.8.1-1 版本和默认配置（1GB 堆，20% 默认区域用于堆外存储和启用持久性）时，我注意到 ignite进程最多可以使用 11GB 的内存（通过检查顶部进程使用的内存常驻大小来验证，见附件）。当我检查日志中的指标时，消耗的内存（堆+堆外）加起来不接近 7GB。一种可能性是检查点缓冲区可以使用额外的内存，但默认情况下应该是默认区域的 1/4，即只有大约 0.25 * 0.2 * 16GB 的四分之一。
关于剩余内存的用途的任何提示？
谢谢！
Metrics for local node (to disable set 'metricsLogFrequency' to 0)
^-- Node [id=a52857f2, name=dummy-ignite-node, uptime=6 days, 10:41:45.541]
^-- H/N/C [hosts=98, nodes=221, CPUs=1481]
^-- CPU [cur=16.17%, avg=16.02%, GC=0%]
^-- PageMemory [pages=788278]
^-- Heap [used=1474MB, free=58.24%, comm=2020MB]
^-- Off-heap [used=3115MB, free=10.39%, comm=3376MB]
^-- sysMemPlc region [used=0MB, free=99.99%, comm=100MB]
^-- default region [used=3115MB, free=1.93%, comm=3176MB]
^-- metastoreMemPlc region [used=0MB, free=99.73%, comm=0MB]
^-- TxLog region [used=0MB, free=100%, comm=100MB]
^-- Ignite persistence [used=57084MB]
^-- sysMemPlc region [used=0MB]
^-- default region [used=57084MB]
^-- metastoreMemPlc region [used=0MB]
^-- TxLog region [used=0MB]
^-- Outbound messages queue [size=0]
^-- Public thread pool [active=0, idle=0, qSize=0]
^-- System thread pool [active=0, idle=8, qSize=0]
===一些关于设置的日志===
dsCfg=DataStorageConfiguration [sysRegionInitSize=41943040, sysRegionMaxSize=104857600, pageSize=0, concLvl=0, dfltDataRegConf=DataRegionConfiguration [name=default, *maxSize=3330655027, initSize=268435456,* swapPath=null, pageEvictionMode=DISABLED, evictionThreshold=0.9, emptyPagesPoolSize=100, metricsEnabled=false, metricsSubIntervalCount=5, metricsRateTimeInterval=60000, persistenceEnabled=true, checkpointPageBufSize=0, lazyMemoryAllocation=true], dataRegions=null, storagePath=null, checkpointFreq=180000, lockWaitTime=10000, checkpointThreads=4, checkpointWriteOrder=SEQUENTIAL, walHistSize=20, maxWalArchiveSize=1073741824, walSegments=10, walSegmentSize=67108864, walPath=db/wal, walArchivePath=db/wal/archive, metricsEnabled=false, walMode=FSYNC, walTlbSize=131072, walBuffSize=0, walFlushFreq=2000, walFsyncDelay=1000, walRecordIterBuffSize=67108864, alwaysWriteFullPages=false, fileIOFactory=org.apache.ignite.internal.processors.cache.persistence.file.AsyncFileIOFactory@285c08c8, metricsSubIntervalCnt=5, metricsRateTimeInterval=60000, walAutoArchiveAfterInactivity=-1, writeThrottlingEnabled=false, walCompactionEnabled=false, walCompactionLevel=1, checkpointReadLockTimeout=null, walPageCompression=DISABLED, walPageCompressionLevel=null], activeOnStart=true, autoActivation=true, longQryWarnTimeout=3000, sqlConnCfg=null, cliConnCfg=ClientConnectorConfiguration [host=null, port=10800, portRange=100, sockSndBufSize=0, sockRcvBufSize=0, tcpNoDelay=true, maxOpenCursorsPerConn=128, threadPoolSize=8, idleTimeout=0, handshakeTimeout=10000, jdbcEnabled=true, odbcEnabled=true, thinCliEnabled=true, sslEnabled=false, useIgniteSslCtxFactory=true, sslClientAuth=false, sslCtxFactory=null, thinCliCfg=ThinClientConfiguration [maxActiveTxPerConn=100]], mvccVacuumThreadCnt=2, mvccVacuumFreq=5000, authEnabled=false, failureHnd=RestartProcessFailureHandler [super=AbstractFailureHandler [ignoredFailureTypes=UnmodifiableSet []]], commFailureRslvr=DefaultCommunicationFailureResolver []]
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Daemon mode: on
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] OS: Linux 3.10.0-1160.49.1.el7.x86_64 amd64
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] OS user: root
[2022-01-27T16:41:44,227][INFO ][main][IgniteKernal%ignite-jetstream-prd1] PID: 7181
[2022-01-27T16:41:44,228][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Language runtime: Scala ver. 2.11.12
[2022-01-27T16:41:44,228][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM information: OpenJDK Runtime Environment 1.8.0_312-b07 Red Hat, Inc. OpenJDK 64-Bit Server VM 25.312-b07
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM total memory: 0.96GB
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Remote Management [restart: off, REST: off, JMX (remote: off)]
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] Logger: Log4J2Logger [quiet=true, config=config/ignite-log4j2.xml]
[2022-01-27T16:41:44,229][INFO ][main][IgniteKernal%ignite-jetstream-prd1] IGNITE_HOME=/usr/share/apache-ignite
[2022-01-27T16:41:44,230][INFO ][main][IgniteKernal%ignite-jetstream-prd1] VM arguments: [-XX:+AggressiveOpts, *-Xms1g, -Xmx1g, -XX:MaxPermSize=128M*, -Dfile.encoding=UTF-8, -DIGNITE_QUIET=true, -DIGNITE_HOME=/usr/share/apache-ignite, -DIGNITE_PROG_NAME=/usr/share/apache-ignite/bin/ignitevisorcmd.sh, -DIGNITE_DEPLOYMENT_MODE_OVERRIDE=ISOLATED]
Screen shot of top command output
是的，这里也考虑了检查点缓冲区大小，如果您没有覆盖默认值，它应该是 3GB/4 as you correctly highlighted. I wonder if it might be changed automatically since you have a lot more data ^-- Ignite persistence [used=57084MB] stored than the region capacity is - only 3GB. Also, this might be related to Direct Memory 使用，我认为 Java 没有计算在内堆使用。
无论如何，我认为最好明确检查 Ignite 内存指标，例如 data region and onheap 使用情况并详细检查它们。