Hello! Yes, data is kept in off-heap. But if you do a large group by or order by without using index, the intermediate results may overwhelm your heap.
I have no idea whether increasing heap to 6G will help or not. The recommendation is to test every query on development cluster and never experiment with group by/order by queries live in production. Regards, -- Ilya Kasnacheev вт, 19 мар. 2019 г. в 03:33, James Wang 王升平 (edvance CN) < james.w...@edvancesecurity.com>: > Hi Ilya Kasnacheev, > > > > On Ingite 2.7 > > No code is using affinityCall() > > Haven’t enabled cache statistics > > > > Could you please answer my questions in red below: > > > > Try to increase Xmx up to 4Gb or higher or avoid G1GC usage on small heaps > as it is very sensitive to free heap memory. > > James>> Yes, it helps. > > As I understand all the data is kept in non-heap memory, But could you > advise when the heap memory will be used, like “order”, “group by” > operation ? > > Looks like you have Visor node (or may be web-console) in grid. Is OOM > happened only when Visor attached to grid? > > James>> Most like to occur if execute a complicate query. > > Another question is: if I set Xmx=6GB, is it possible to avoide > out-of-memory failure. I regard this very critical because it make node > halt. > > > > Best Regards, > > James Wang > > M/WeChat: +86 135 1215 1134 > > > > *From:* Ilya Kasnacheev <ilya.kasnach...@gmail.com> > *Sent:* Tuesday, March 5, 2019 09:44 PM > *To:* user@ignite.apache.org > *Subject:* Re: java.lang.OutOfMemoryError: GC overhead limit exceeded > > > > Hello! > > > > Is there a chance you were using affinityCall() and you're on Ignite 2.7? > > > > We have a bug which can cause spurious OOMs here: > https://issues.apache.org/jira/browse/IGNITE-10925 - disabling cache > statistics will help > > > > Regards, > > -- > > Ilya Kasnacheev > > > > > > сб, 2 мар. 2019 г. в 18:13, James Wang 王升平 (edvance CN) < > james.w...@edvancesecurity.com>: > > Hi Support, > > > > Try to increase Xmx up to 4Gb or higher or avoid G1GC usage on small heaps > as it is very sensitive to free heap memory. > > James>> will increase heap memory and retry. > > As I understand all the data is kept in non-heap memory, But could you > advise when the heap memory will be used, like “order”, “group by” > operation ? > > > > Looks like you have Visor node (or may be web-console) in grid. Is OOM > happened only when Visor attached to grid? > > James>> Most like to occur if execute a complicate query. > > > > Another question is: if I set Xmx=6GB, is it possible to avoide > out-of-memory failure. I regard this very critical because it make node > halt. > > > > Thank you. > > > > Best Regards, > > James Wang > > M/WeChat: +86 135 1215 1134 > > > > *From:* Andrey Mashenkov <andrey.mashen...@gmail.com> > *Sent:* Friday, March 1, 2019 06:50 PM > *To:* user@ignite.apache.org > *Subject:* Re: java.lang.OutOfMemoryError: GC overhead limit exceeded > > > > Hi, > > > > Most likely heap size is too low. > > Try to increase Xmx up to 4Gb or higher or avoid G1GC usage on small heaps > as it is very sensitive to free heap memory. > > > > Looks like you have Visor node (or may be web-console) in grid. Is OOM > happened only when Visor attached to grid? > > > > On Fri, Mar 1, 2019 at 7:17 AM James Wang 王升平 (edvance CN) < > james.w...@edvancesecurity.com> wrote: > > OS: 4C +8GB > > Data Region = 4GB > > > > start command: > > nohup $IGNITE_HOME/bin/ignite.sh -Xmx1024m -XX:+UseG1GC > -XX:MaxDirectMemorySize=1G grtip-config.xml > ignite.log 2>&1 & > > How to adjust the the memory tunning. > > > > [21:29:22,777][SEVERE][mgmt-#33519%234-236-237-241%][GridJobWorker] > Runtime error caught during grid runnable execution: GridJobWorker > [createTime=1551360450243, startTime=1551360453071, > finishTime=1551360483944, taskNode=TcpDiscoveryNode > [id=a7839266-6396-4f7c-9ef7-c3a4b2355782, addrs=[0:0:0:0:0:0:0:1%lo, > 127.0.0.1, 192.168.1.236], sockAddrs=[/192.168.1.236:47500, > /0:0:0:0:0:0:0:1%lo:47500, /127.0.0.1:47500], discPort=47500, order=56, > intOrder=31, lastExchangeTime=1551151457630, loc=false, > ver=2.7.0#20181201-sha1:256ae401, isClient=false], internal=true, > marsh=BinaryMarshaller [], ses=GridJobSessionImpl [ses=GridTaskSessionImpl > [taskName=o.a.i.i.v.service.VisorServiceTask, dep=GridDeployment > [ts=1551094729510, depMode=SHARED, > clsLdr=sun.misc.Launcher$AppClassLoader@764c12b6, > clsLdrId=3aef2742961-5a63d018-0eb4-493e-91a4-be6d41caff85, userVer=0, > loc=true, sampleClsName=o.a.i.i.processors.cache.CachesRegistry, > pendingUndeploy=false, undeployed=false, usage=0], > taskClsName=o.a.i.i.v.service.VisorServiceTask, > sesId=8a35b792961-a7839266-6396-4f7c-9ef7-c3a4b2355782, > startTime=1551360450899, endTime=9223372036854775807, > taskNodeId=a7839266-6396-4f7c-9ef7-c3a4b2355782, > clsLdr=sun.misc.Launcher$AppClassLoader@764c12b6, closed=false, > cpSpi=null, failSpi=null, loadSpi=null, usage=1, fullSup=false, > internal=true, topPred=ContainsNodeIdsPredicate [], > subjId=a7839266-6396-4f7c-9ef7-c3a4b2355782, mapFut=IgniteFuture > [orig=GridFutureAdapter [ignoreInterrupts=false, state=INIT, res=null, > hash=327805292]], execName=null], > jobId=9a35b792961-a7839266-6396-4f7c-9ef7-c3a4b2355782], > jobCtx=GridJobContextImpl > [jobId=9a35b792961-a7839266-6396-4f7c-9ef7-c3a4b2355782, timeoutObj=null, > attrs={}], dep=GridDeployment [ts=1551094729510, depMode=SHARED, > clsLdr=sun.misc.Launcher$AppClassLoader@764c12b6, > clsLdrId=3aef2742961-5a63d018-0eb4-493e-91a4-be6d41caff85, userVer=0, > loc=true, sampleClsName=o.a.i.i.processors.cache.CachesRegistry, > pendingUndeploy=false, undeployed=false, usage=0], finishing=true, > masterLeaveGuard=false, timedOut=false, sysCancelled=false, > sysStopping=false, isStarted=true, job=VisorServiceJob [], held=0, > partsReservation=null, reqTopVer=null, execName=null] > > java.lang.OutOfMemoryError: GC overhead limit exceeded > > [21:29:22,781][SEVERE][nio-acceptor-tcp-comm-#28%234-236-237-241%][] JVM > will be halted immediately due to the failure: [failureCtx=FailureContext > [type=CRITICAL_ERROR, err=java.lang.OutOfMemoryError: GC overhead limit > exceeded]] > > [21:29:22,814][SEVERE][grid-timeout-worker-#23%234-236-237-241%][GridTimeoutProcessor] > Error when executing timeout callback: CancelableTask > [id=94ef2742961-0742c515-5b96-4aad-b07a-9f1ec60af5f3, > endTime=1551360514795, period=3000, cancel=false, task=MetricsUpdater > [prevGcTime=38806, prevCpuTime=2195727, > super=o.a.i.i.managers.discovery.GridDiscoveryManager$MetricsUpdater@73766331 > ]] > > java.lang.OutOfMemoryError: GC overhead limit exceeded > > > > > > James Wang / edvance > Mobile/WeChat: +86 135 1215 1134 > > This message contains information that is deemed confidential and > privileged. Unless you are the addressee (or authorized to receive for the > addressee), you may not use, copy or disclose to anyone the message or any > information contained in the message. If you have received the message in > error, please advise the sender by reply e-mail and delete the message. > > > > > -- > > Best regards, > Andrey V. Mashenkov > > This message contains information that is deemed confidential and > privileged. Unless you are the addressee (or authorized to receive for the > addressee), you may not use, copy or disclose to anyone the message or any > information contained in the message. If you have received the message in > error, please advise the sender by reply e-mail and delete the message. > > This message contains information that is deemed confidential and > privileged. Unless you are the addressee (or authorized to receive for the > addressee), you may not use, copy or disclose to anyone the message or any > information contained in the message. If you have received the message in > error, please advise the sender by reply e-mail and delete the message. >
