grid infrastructure management repository –everything you ...€¦ · database file management...

Copyright©2016, Oracleand/oritsaffiliates.Allrightsreserved.|

MarkV.Scardina- DirectorOracleQoS Management&OracleAutonomousHealthFramework

Oracle12cGridInfrastructureManagementRepository– EverythingYouWantedToKnow


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips

FrequentlyAskedQuestions

Q&A– FurtherInformation

1

2

3

4

3

5

6

Copyright©2016, Oracleand/oritsaffiliates.Allrightsreserved.| 4

CentralizedDatabaseforDiagnostic&PerformanceDataGIMRIntroduction– WhoaretheGIMR’sClients?

• Currentlyin12.1• ClusterHealthMonitor• RapidHomesProvisioning• EMCloudControl• TraceFileAnalyzer

• Newin12.2• ClusterActivityLog• ClusterHealthAdvisor• QoS Management


OracleAutonomousHealthFramework

• Integratesnextgenerationtoolsrunningascomponents- 24/7• DiscoversPotentialIssuesandNotifiesortakesCorrectiveActions• SpeedsIssueDiagnosisandRecovery• PreservesDatabaseandServerAvailabilityandPerformance• AutonomouslyMonitorsandManagesresourcestomaintainSLAs

5

WorkingforYouContinuously


CentralizedDatabaseforDiagnostic&PerformanceDataGIMRIntroduction– WhywasitImplemented?

• Longstandingrequirementfordiagnosticdatarepository

• NoavailablestorageinEMCCRepository• Proliferationoflocaldiskdatarepositories• GrowingscarcityoflocalspaceduetoDB

consolidation• ClientdependencyonOracleDBfunctionality• Needforinter-clientdatasharing• Requirementsforoff-clusterdataaccess


CentralizedDatabaseforDiagnostic&PerformanceDataGIMRIntroduction– WhatistheGIMR?

• InfrastructuredatabaseresidinginGIHome• SingleInstanceCDBwithsinglePDBwith

partitioning– Noseparatelicenserequired• Alwaysrunning– enabledbydefault• Clusterresourcewithrestartandfailover• UsesanASMDiskGroupbydefault• UsesaFixedsetofsystemresources• Automaticdatalifecyclemanagement


CentralizedDatabaseforDiagnostic&PerformanceDataGIMRIntroduction– WhyanOracleMultitenantDB?

• ClientDataInterconnectivity• CHM<->CHA• CHM<->QoS Management• TFA<->CHM• …

• EMCloudControlAccess• ClusterHealthMonitor• ClusterHealthAdvisor• …

• ClusterDomainSupport


Oracle12cDomainServicesCluster(DSC)

9

• HostsGIMRasaRepositoryService• EachclusterisassignedaPDB• Reduceslocalresourcefootprint• Centralizesmanagement• Speedsdeploymentandpatching• OptionalSharedStorageServices• Supportsmultipleversionsandplatformsgoingforward

DeployswithMinimumFootprintandMaximumManageability

ApplicationMemberCluster

DatabaseMemberCluster


OracleDomainServicesCluster




ORACLECLUSTERDOMAIN

Management Repository ServiceGrid Names ServiceStorage ServicesRapid Home Provisioning Service


OracleDomainServicesCluster

OracleClusterDomain

ASMIOService

ACFSServices

ASMService


UsesASMService


UsesIO&ASMServiceofDSC

MgmtRepository(GIMR)Service


GIonly


UseslocalASM

SharedASM

AdditionalOptionalServices

RapidHomeProvisioning

(RHP)Service

PrivateNetwork

SAN

NAS


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips



1

2

3

4

11

5

6


ListenerandNetworkingGIMRTechnicalDetails– Topologyfor12.1.0.2

SCANMGMTDB

CHM(ologgerd)

RHP(ghs oc4j)

TFA(tfactl)

EMCC(OMS)

Public(1521)

Private(1525)

NODE


ListenerandNetworkingGIMRTechnicalDetails– Topologyfor12.2.0.1

SCANMGMTDB

CHM(ologgerd)

RHP(ghserver)

TFA(tfactl)

EMCC(OMS)

Public(1521)Private

(1525)

CALOG(oc*)

CHA(ochad)

QoSM(qosmserver)


OracleDatabaseINITandCONFIGParametersandValuesGIMRTechnicalDetails- Configuration

PARAMETER 12.1.0.2 12.2.0.1STANDALONE 12.2.0.1DSC

INSTANCE_NAME -MGMTDB -MGMTDB -MGMTDB

DBNAME _MGMTDB _MGMTDB _MGMTDB

SGA_MAX_SIZE 752MB 1G 4GB

PGA_AGGREGATE_TARGET 352MB 500MB 2GB

PGA_AGREGATE_LIMIT 2GB 2GB 6GB

CPU_COUNT 2 2 8

PROCESSES 300 500 2000

SESSIONS 472 772 3024

USE_LARGE_PAGES TRUE TRUE TRUE


SRVCTL:GIMRLifecycleControlGIMRTechnicalDetails- Tools

srvctl start|stop mgmtdbsrvctl status mgmtdb

Database is enabledInstance -MGMTDB is running on node mysvr1

srvctl config mgmtdbDatabase unique name: _mgmtdbDatabase name:Oracle home: <CRS home>Oracle user: gridSpfile: +DATA/_MGMTDB/PARAMETERFILE/spfile.20150930124309Password file:Domain:Start options: openStop options: immediateDatabase role: PRIMARYManagement policy: AUTOMATICType: ManagementPDB name: myclusterPDB service: myclusterCluster name: myclusterDatabase instance: -MGMTDB

srvctl modify mgmtdb: DoNotUse!

srvctl start|stop mgmtlsnrsrvctl status mgmtlsnr

Listener MGMTLSNR is enabledListener MGMTLSNR is running on node(s): mysvr1

srvctl config mgmtlsnrName: MGMTLSNRType: Management ListenerOwner: gridHome: <CRS home>End points: TCP:1525Management listener is enabled.Management listener is individually enabled on nodes:Management listener is individually disabled on nodes:

srvctl modify mgmtlsnr -endpoints “TCP:1531”

12.2.0.1+PDB name: GIMR_DSCREP_##PDB service: GIMR_DSCREP_##


MGMTCA:User/PasswordManagementTechnicalDetails- Tools

FOR12.1• mgmtca

- ResetstheCHMdaemonGIMRpassword• mgmtca –em <password>

- For12.1setstheEMCCCHMuserpasswordforremoteGIMRaccess- For12.2passwordissetandsavedinEMCCdirectly

• mgmtca –gridhome- SetstheRHPserverGIMRpassword

FOR12.2• mgmtca –user CHMOS,CHA,CALOG,QOS,GRIDHOME

–allusers

PasswordsarestoredinOracleWalletsinOCRfordaemonaccess


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips



1

2

3

4

17

5

6


DatabaseDeploymentManagementGIMR– BestPractices

• Monitoring• DonotconfigureMGMTDBorMGMTLSNRasanEMCCtarget!• Hiddenincurrent13.2EMCCreleases• DatabaseandListenerareautomaticallymonitoredbyCRS• Databaseismanagedbyitsclients• EMCCdoesnothaveproperaccess• EMCCwouldseeitasaSIDB,andlosetrackwhenitfailsover

• SecurityandPasswordManagement• Securebydefault– passwordsautomaticallygenerated• Daemonclientsmonitorexpirationandautomaticallyresetasrequired


DatabaseFileManagementGIMR– BestPractices

• Datalifecycleauto-managedbyclients• RecommendedClientRetention- 72hourminimum• ForCHM,useoclumon manage -repos checkretentiontime 86400

• ForCHM,useoclumon manage -repos changerepossize <#Mbytes>

• CreateDedicatedGIMRDiskGrouponInstallation• 12.1.0.2diskgroupcreation:startwithGIMRhostingdiskgroup• GIMRtypicallyonlyrequiresexternalredundancy• Clusterwarefileseasyaretorelocatelater• Usecrsctl replace votedisk <grid_dg>

• Use ocrconfig -add <grid_dg> and ocrconfig –delete <gimr_dg>


MGMTSeparateDiskGroupCreationonInstallation(Builtinto12.2)GIMR– BestPractices

DiskSpacefor72hrs*• 12.1.0.2:– 5.2GB(<5nodes)– 500MBeachadditionalnode

• 12.2.0.1:– 36GB(<5nodes)– 4.7GBeachadditionalnode

• 12.2DSC:– 188GB(<5memberclusters)– 35GBeachadditionalcluster

*ExternalRedundancy


DatabaseFileManagementGIMR– BestPractices

• MoveGIMRdatabasefilestoitsownDiskGroup• DownloadMDBUtil fromMOS2065175.1• CreateASMDiskGroup(ex:MGMT)withASMCA• MoveGIMRwithMDBUtil

• mdbutil.pl --mvmgmtdb --target=+MGMT

• Confirmsuccessfulrelocationandstart• srvctl config mgmtdb• oclumon dumpnodeview –allnodes

• ManageMGMTDBAuditFiles– SeeMOS2202044.1• Location:<GRID_BASE>/admin/_mgmtdb/adump• Retentioncanbeautomaticallymanagedin12.1.0.2and12.2


MDBUtil inOperationPostInstallationGIMR– BestPractices

[grid@mysvr tmp]$ /tmp/mdbutil.pl --mvmgmtdb --target=+MGMTMoving MGMTDB, it will be stopped, are you sure (Y/N)? y2015-10-12 09:24:53: I Checking for the required paths under +MGMT2015-10-12 09:24:54: I Creating new path +MGMT/_MGMTDB/PARAMETERFILE2015-10-12 09:24:56: I Creating new path +MGMT/_MGMTDB/CONTROLFILE2015-10-12 09:24:59: I Creating new path +MGMT/_MGMTDB/ONLINELOG2015-10-12 09:25:01: I Creating new path +MGMT/_MGMTDB/DATAFILES2015-10-12 09:25:04: I Creating new path +MGMT/_MGMTDB/TEMPFILE2015-10-12 09:25:06: I Creating new path +MGMT/_MGMTDB/DATAFILES/mydb_c2015-10-12 09:25:08: I Creating new path +MGMT/_MGMTDB/TEMPFILE/mydb_c2015-10-12 09:25:08: I Getting MGMTDB Database files location2015-10-12 09:25:09: I Getting MGMTDB Temp files location2015-10-12 09:25:09: I Getting MGMTDB PDB mydb_c files location2015-10-12 09:25:09: I Getting MGMTDB PDB mydb_c Temp files location2015-10-12 09:25:10: I Creating temporary PFILE2015-10-12 09:25:10: I Creating target SPFILE2015-10-12 09:25:16: I Stopping mgmtdb2015-10-12 09:25:36: I Copying MGMTDB DBFiles to +MGMT2015-10-12 09:25:52: I Copying MGMTDB mydb_c PDB DBFiles to +MGMT2015-10-12 09:26:33: I Creating the CTRL File2015-10-12 09:26:59: I The CTRL File has been created and MGMTDB is now running from +MGMT2015-10-12 09:26:59: I Setting MGMTDB SPFile location2015-10-12 09:27:00: I Modifying the init parameter2015-10-12 09:27:00: I Removing old MGMTDB2015-10-12 09:27:02: I Restarting MGMTDB using target SPFile2015-10-12 09:27:47: I MGMTDB Successfully moved to +MGMT!


GridInfrastructureUpgradesandPatchesGIMR– BestPractices

• EnsureGIMRisfullyoperational• Executesrvctl status mgmtdb

• Executesrvctl status mgmtlsnr

• Executeoclumon dumpnodeview –all

• OptionallysaveexistingCHMdata• Ex: oclumon dumpnodeview -last “72:00:00” >> /tmp/gimr.sav• Ex: tfactl diagcollect –chmos –since 3d


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips



1

2

3

4

24

5

6


GIMR– TroubleshootingTips

• EnsureboththeMGMTDBanditslistener,MGMTLSNRareupandrunningonthesame node.

• Alertlogsandtracefilesareco-locatedwithuserdatabasesunder$GRID_BASE/admin.

• IftheGIMRiscorruptedin12.1,itcanbedeletedandre-createdusingMDBUtil• mdbutil.pl --addmdb --target=+MGMT

• IfEMCCCHMpagein12.1cannotauthenticatewiththedbsnmpuser,runmgmtca• mgmtca –em <mypassword>


GIMR– TroubleshootingTips

• IftheGIMRfailstobecreatedduringa12.1upgrade,itcanbecreatedpostupgradeusingMDBUtil.• mdbutil.pl --addmdb --target=+MGMT

• IftheGIMRfailsduetheMGMTLSNRunabletostart,changeitsportusingsrvctl.• srvctl modify mgmtlsnr -endpoints “TCP:1541”

• IfyouneedSQLPlus access,youmayuseOSauthentication.1. export ORACLE_SID=\-MGMTDB2. sqlplus / as sysdba

• NOTE:ThisshouldbedoneonlyunderOracleSupportServicedirection.


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips



1

2

3

4

27

5

6


GIMR– FrequentlyAskedQuestions

• CanIdisabletheGIMR?• No,itisnotsupportedtorun12.1.0.2+clustersonTierOneplatformswithout

theGIMRenabledandrunningasitsdataisrequiredbyOSS.• WillIlosemyclusterordatabaseavailabilityiftheGIMRgoesdown?• No,theGIMRclientsaredesignedtolocallycachedataiftheGIMRisdownfor

aperiodoftime.ShouldthishappenCRSwillrestartorfailitovertoanothernode.

• WhydoestheGIMRusehugepages?• TheGIMRonlyusesasmallquantityofhugepages (376in12.1)ifavailableto

preventitsSGAfromswappingsincesomeofitsclientshavetimingwindows.



• DoIneedtoseparatelypatchtheGIMR?• No,anypatchesfortheGIMRwillbeincludedintheGIPSUandapplied

duringtheGIpatchingprocess.• WillIlosemyGIMRdatawhenupgradingorapplyingapatch?• YouwillloseyourCHMdatabutnotyourRHPdataduringanupgrade.

WhetherithappensforaPSUwilldependuponthelevelofGIMRpatch.• DoIneedtoregularlybackuptheGIMR?• Itisoptionalatthistime,asitsdataisregularlywindowedthrough

droppingpartitions.Youcanuseoclumon toregularlyarchivedata.• For12.2:clientsmaintain/migratetheirdataonupgrade.



• DoestheGIMRgetconfiguredinaOracleRestartsingleserverinstall?• No,asitdoesnotcurrentlyhaveclientsinthatdeploymenttype.

• CanIuseDBCAtomakechangesorcreate/deletetheGIMR?• No,astheGIMRincorporatesaprotectedstartingcharacter“_”initsDB_NAME

itcannotbeoperatedonbyDBCA.MGMTCAoritsclientutilitiesmanagetheGIMR.

• IfEMCCisnotsupposedtomonitorit,whyisitdiscoveredasatarget?• ThediscoveryoftheGIMRhasbeenmaskedinanthelatestEMrelease.

• HowdoIcd intotheGIMR’straceorlogdirectory?• Remembertoescapethehyphen:cd ./-MGMTDB


Agenda

Introduction

TechnicalDetails

BestPractices

TroubleshootingTips



1

2

3

4

31

5

6


ForFurtherInformation

32

• Oracle12cClusterware AdminGuide

• OracleClusterware onOTN

• MDBUtil:GIManagementRepositoryconfigurationtool(DocID2065175.1)

• FAQ:12cGridInfrastructureManagementRepository(GIMR)(DocID1568402.1)

• HowToAutomateGIMROSAuditLogLifecycleManagement(DocID2202044.1willbepublishedshortly)

grid infrastructure management repository –everything you ...€¦ · database file management...

Documents