01-24-2018 06:18 AM - edited 03-01-2019 06:22 PM
Customer find his PI is extremely slow, in the GUI and in the CLI.
Exact PI version is 3.2.1
Investigation using top, shows a process is taking 160% CPU and 40% memory.
Load average is 5.20, 6.20 4.00
On other PI systems this process also takes a very large amount of memory (+-40%), but the CPU usage is much lower 5-10%.
Server meets memory CPU and disk requirements.
Checking the IOPS shows the machine exceeds the minimal 200MB/s
vm-cpi/admin# ncs run test iops
Testing disk write speed ...
8388608+0 records in
8388608+0 records out
8589934592 bytes (8.6 GB) copied, 32.1145 s, 267 MB/s
Checking the pid using ps -ef | grep <pid> shows the process causing the issue:
prime <pid> 1 99 16:15 ? 00:39:45 /opt/CSCOlumos/bin/pi-java -XX:InitialCodeCacheSize=64m -XX:ReservedCodeCacheSize=256m -Dinstall.dir=/opt/CSCOlumos -Dapplication.name=ncs-server -Daes.dir=/opt/CSCOlumos -Dmaps.dir=/opt/CSCOlumos/domainmaps -Dheatmap.dir=/opt/CSCOlumos/domainmaps/heatmaps -Dmibs.root=/opt/CSCOlumos/resources/wireless/mibs -Daxis.ClientConfigFile=/opt/CSCOlumos/conf/rfm/classes/com/cisco/common/ha/config/client-config.wsdd -Daxis.socketSecureFactory=org.apache.axis.components.net.SunFakeTrustSocketFactory -Dreport.server=false -Dcom.cisco.xmp.inventory.collector.logging.level=info -Dcom.cisco.xmp.xde.logging.level=info -Dcom.cisco.xmp.nice.logging.level=info -Dcom.cisco.xmp.topology.logging.level=info -Dcom.cisco.xmp.job.manager.logging.level=info -Dcom.cisco.xmp.preferences.logging.level=info -Dcom.cisco.nm.discovery.logging.level=info -Dcom.cisco.xmp.usermgmt.logging.level=info -Dcom.cisco.xmp.grouping.logging.level=info -Dcom.cisco.xmp.persistence.logging.level=info -Dcom.cisco.xmp.config.logging.level=info -Dcom.cisco.xmp.poller.logging.level=info -Dcom.cisco.xmp.audit.logging.level=info -Dcom.cisco.xmp.grt.logging.level=info -Dcom.cisco.xmp.persistence.dbmigration.logging.level=info -Dcom.cisco.xmp.datacenter.eventcorrelation.logging.level=info -Dcom.cisco.xmp.existence.inventory.logging.level=info -Dxmp.default.home=/opt/CSCOlumos/ -Dxmp.updates.home=/opt/CSCOlumos/updates -Dxmp.updates.outofboxtechpacks.home=/opt/CSCOlumos/updates/staging/outofboxtechpacks -Dxmp.update.config.file=/opt/CSCOlumos/conf/update.properties -Dxmp.update.log.dir=/opt/CSCOlumos/logs -Dxmp.update.mdfid=/opt/CSCOlumos/conf/update.mdfid.properties -Dxmp.java.process.name=StartXMP -Djava.system.class.loader=com.cisco.xmp.classLoader.XMPSystemClassLoader -Dcom.cisco.xmp.classLoader.XMPSystemClassLoader.verbose=true -Dcom.cisco.xmp.acdr.devpackloader.DevicePackageLoader.packagesDir=/opt/CSCOlumos/xmp_inventory/dar -Dcom.cisco.xmp.dpl.rest.xar.XarRepository.searchIndexDir=/opt/CSCOlumos/xar-search-index -Dcom.cisco.xmp.inventory.profile.directory=/opt/CSCOlumos/xmp_inventory/dar -Dcom.cisco.nm.workflow.XDEHome=/opt/CSCOlumos/xmp_inventory/xde-home -Dcom.cisco.xmp.xdeinit.loadXARFilesInParallel=false -Dxmp.inventory.config.file=/opt/CSCOlumos/conf/inventory.properties -Dxmp.existenceInventory.config.file=/opt/CSCOlumos/conf/existenceInventory.properties -Dxmp.topology.config.file=/opt/CSCOlumos/conf/topology.properties -Dxmp.persistence.config.file=/opt/CSCOlumos/conf/persistence_config.properties -Dxmp.persistence.optimizeddlmetadata.dir=/opt/CSCOlumos/conf/ddlmetadata -Dxmp.persistence.prunemetadata.dir=/opt/CSCOlumos/conf/prunemetadata -Dxmp.persistence.schemacreate.listeners.dir=/opt/CSCOlumos/conf/schemacreate_listeners -Dxmp.persistence.notificationmetadata.dir=/opt/CSCOlumos/conf/notificationmetadata -Dxmp.db.config.file=/opt/CSCOlumos/conf/jdbc.properties -Dderby.system.home=/opt/CSCOlumos/conf/xmp_data/derbydata -Djava.util.prefs.PreferencesFactory=com.cisco.xmp.prefs.db.impl.DBImplPreferencesFactory -Dxmp.persistence.tqexclusion.file=/opt/CSCOlumos/conf/TqNotRequired.txt -Ddecap.home=/opt/CSCOlumos/decap -Dcom.cisco.xmp.XMPHome=/opt/CSCOlumos -Dcom.cisco.xmp.PI_TYPE=primary -Dxmp.logging.dir=/opt/CSCOlumos/logs -Dxmp.mbc.config.file=/opt/CSCOlumos/conf/mbc_config.properties -Dxmp.mbc.globalobject.types.file=/opt/CSCOlumos/conf/mbc_globalobject_types.properties -Dxmp.mbc.customtags.file=/opt/CSCOlumos/conf/mbc_customtags.properties -Dcom.cisco.xmp.ModelHome=/opt/CSCOlumos/lib/model -Dxmp.conf.dir=/opt/CSCOlumos/conf -Dcom.tc.productkey.path=/opt/CSCOlumos/conf/license/terracotta-license.key -Dtp_persistent_path=/opt/CSCOlumos/licenses -Dxmp.nbi.notifications.config.file=/opt/CSCOlumos/conf/nbi_notifications.properties -Dxmp.nbi.notifications.requiredClasses.file=/opt/CSCOlumos/conf/nbi_notifications_required.txt -Dxmp.poller.config.dir=/opt/CSCOlumos/conf/pollingconfig -Dxmp.grouping.config.file=/opt/CSCOlumos/conf/grouping.properties -Dxmp.grt.config.file=/opt/CSCOlumos/conf/grt_config.properties -Dxmp.messaging.config.file=/opt/CSCOlumos/conf/messaging.properties -Dxmp.update.asd=/opt/CSCOlumos/conf/update.asd.properties -Dxmp.syslog.config.file=/opt/CSCOlumos/conf/syslog_config.properties -Dxmp.updates.home=/opt/CSCOlumos/updates -Dxmp.update.config.file=/opt/CSCOlumos/conf/update.properties -Dxmp.update.mdfid=/opt/CSCOlumos/conf/update.mdfid.properties -javaagent:/opt/CSCOlumos/lib/xmp-third-party/aspectjweaver-1.8.4.jar -Daj.weaving.loadersToSkip=com.cisco.nm.xms.xdi.util.prmgr.ZipLoader,org.apache.jasper.servlet.JasperLoader,net.sf.jasperreports.engine.util.JRClassLoader -Djava.security.egd=file:///dev/urandom -Djavax.net.ssl.trustStore=/opt/CSCOlumos/conf/truststore -XX:CompileCommand=quiet -XX:CompileCommand=exclude,org/snmp4j/Snmp.send -XX:-ReduceInitialCardMarks -Dnet.sf.ehcache.enableShutdownHook=true -Dxmp.jobapprover.config.file=/opt/CSCOlumos/conf/jobapprover.properties -Dxmp.credential_mgmt.config.file=/opt/CSCOlumos/conf/credential_notifications.properties -Dxmp.topology.device.config.file=/opt/CSCOlumos/conf/DeviceImagesMapFile.xml -Dorg.apache.cxf.io.CachedOutputStream.OutputDirectory=/opt/CSCOlumos/tmp -Dxmp.audit.classes.file=/opt/CSCOlumos/conf/ClassesToAudit.xml -Dxmp.usermgmt.config.file=/opt/CSCOlumos/conf/usermgmt.properties -Dhttps.protocols=TLSv1,TLSv1.1,TLSv1.2 -Djava.awt.headless=true -Dsun.reflect.inflationThreshold=2147483647 -Doracle.jdbc.autoCommitSpecCompliant=false -Djdk.tls.ephemeralDHKeySize=2048 -Dxmp.scheduler.config.file=/opt/CSCOlumos/conf/scheduler.properties -Xmx4096m -XX:MetaspaceSize=512m -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/opt/CSCOlumos/crash -XX:ThreadStackSize=512 -XX:+UseConcMarkSweepGC -XX:ParallelGCThreads=5 -XX:NewRatio=2 -XX:+UnlockDiagnosticVMOptions -XX:-UseLoopPredicate -DsolrHome=/opt/CSCOlumos/conf/ifm/ciscoSolrHome2 -DsolrCore=core0 -Difm_config_dir=/opt/CSCOlumos/conf/ifm/ -DLogConfigurationArchiveStackTraces=false -DLogConfigurationArchiveDetailedDebugMsgs=false -DperfPoller.batchSize=1000 -XX:-UseCompressedClassPointers -XX:-UseCompressedOops -XX:MetaspaceSize=1280m -Xms2048m -Xmx6656m -Djava.io.tmpdir=/opt/CSCOlumos/tmp/temp/reports -Duser.timezone=CET -Dnet.sf.ehcache.skipUpdateCheck=true -cp /opt/CSCOlumos/classloader-conf:/opt/CSCOlumos/lib/xmp/XMPClassLoader-11.0.1.jar -Djava.system.class.loader=com.cisco.xmp.classLoader.XMPParallelSystemClassLoader com.cisco.xmp.startup.StartXMP -platformConfFile=xmp-server.properties
Possibly this process is called StartXMP
Does anyone have a clue what this process does?
Why it can suddely take so much CPU ?
Thanks for aany feedback or suggestion
Michel
Solved! Go to Solution.
01-25-2018 06:42 AM - edited 02-01-2018 05:51 AM
Restarting the systems did not resolve the slowness, but it appears that now after a restore of a backup prime is responding 'normal' again.
This perhaps means that the CPU usage (160%) of the mentioned process,.... is 'normal'.
Unfortunately I do not have access to sufficient other PI systems to confirm this.
01-24-2018 07:07 AM
>Server meets memory CPU and disk requirements
- How is this determined and or is this a VM or an appliance ? ; if it's a VM check cpu and memory resources through vcenter statistics (e.g.)
01-24-2018 10:23 PM
It's a VM. The validation is just from the server definition in Vsphere which matches with: https://www.cisco.com/c/en/us/td/docs/net_mgmt/prime/infrastructure/3-2/quickstart/guide/cpi_qsg.html#pgfId-121806
16 vCPU, 16 GB and 900GB disk.
It appears there are no jobs running (I assume they should have an In_Progress state in the job dashboard when they are running)
I'm trying to understand what this process is and what it does.
And of course I'm looking what could have changed.
01-25-2018 06:42 AM - edited 02-01-2018 05:51 AM
Restarting the systems did not resolve the slowness, but it appears that now after a restore of a backup prime is responding 'normal' again.
This perhaps means that the CPU usage (160%) of the mentioned process,.... is 'normal'.
Unfortunately I do not have access to sufficient other PI systems to confirm this.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide