HP Serviceguard manual How the Package Manager Works, Package Types, Non-failover Packages

Page 50

IMPORTANT: During Step 1, while the nodes are using a strict majority quorum, node failures can cause the cluster to go down unexpectedly if the cluster has been using a quorum device before the configuration change. For example, suppose you change the Quorum Server polling interval while a two-node cluster is running. If a node fails during Step 1, the cluster will lose quorum and go down, because a strict majority of prior cluster members (two out of two in this case) is required. The duration of Step 1 is typically around a second, so the chance of a node failure occurring during that time is very small.

In order to keep the time interval as short as possible, make sure you are changing only the quorum configuration, and nothing else, when you apply the change.

If this slight risk of a node failure leading to cluster failure is unacceptable, halt the cluster before you make the quorum configuration change.

How the Package Manager Works

Packages are the means by which Serviceguard starts and halts configured applications. A package is a collection of services, disk volumes, IP addresses, and generic resources that are managed by Serviceguard to ensure they are available.

Each node in the cluster runs an instance of the package manager; the package manager residing on the cluster coordinator is known as the package coordinator.

The package coordinator does the following:

Decides when and where to run, halt, or move packages. The package manager on all nodes does the following:

Executes the control scripts that run and halt packages and their services.

Reacts to changes in the status of monitored resources.

Package Types

Three different types of packages can run in the cluster; the most common is the failover package. There are also special-purpose packages that run on more than one node at a time, and so do not failover. They are typically used to manage resources of certain failover packages.

Non-failover Packages

There are two types of special-purpose packages that do not fail over and that can run on more than one node at the same time: the system multi-node package, which runs on all nodes in the cluster, and the multi-node package, which can be configured to run on all or some of the nodes in the cluster. System multi-node packages are reserved for use by HP-supplied applications, such as Veritas Cluster Volume Manager (CVM) and Cluster File System (CFS).

The rest of this section describes failover packages.

Failover Packages

A failover package starts up on an appropriate node (see node_name on (page 235)) when the cluster starts. A package failover takes place when the package coordinator initiates the start of a package on a new node. A package failover involves both halting the existing package (in the case of a service, network, or resource failure), and starting the new instance of the package.

Failover is shown in the following figure:

50 Understanding Serviceguard Software Components

Image 50
Contents Managing Serviceguard Twentieth Edition Legal Notices Contents Contents Planning and Documenting an HA Cluster Contents Building an HA Cluster Configuration 158 Contents Configuring Packages and Their Services 227 Contents Cluster and Package Maintenance 261 279 Troubleshooting Your Cluster 321 Contents 349 Integrating HA Applications with Serviceguard 354Software Upgrades 357 377 Blank Planning Worksheets 369Migrating from LVM to VxVM Data Storage 374 IPv6 Network Support 378Index 398 Using Serviceguard Manager 385396 Publishing History Publishing HistoryPreface Securing Serviceguard and other Serviceguard white papers Related PublicationsServiceguard at a Glance What is Serviceguard?Failover Shows what happens in a failover situationAbout Veritas CFS and CVM from Symantec Typical Cluster After FailoverUsing Serviceguard Manager Using SAMUsr/sbin/sam -w What are the Distributed Systems Administration Utilities?Roadmap for Configuring Clusters and Packages Tasks in Configuring a Serviceguard Cluster Understanding Serviceguard Hardware Configurations Redundancy of Cluster ComponentsRedundant Network Components Rules and RestrictionsRedundant Ethernet Configuration Redundant LANsFor legacy packages, see Configuring Cross-Subnet Failover Cross-Subnet ConfigurationsConfiguration Tasks RestrictionsFor More Information Supported Disk Interfaces Redundant Disk StorageReplacing Failed Network Cards Disk Mirroring Disk Arrays using RAID Levels and Multiple Data PathsData Protection About MultipathingReplacing Failed Disk Mechanisms Monitoring LVM Disks Through Event Monitoring ServiceMonitoring VxVM and CVM Disks Replacing Failed I/O CardsSample Scsi Disk Configurations Mirrored Disks Connected for High AvailabilitySample Fibre Channel Disk Configuration Cluster with High Availability Disk ArrayRedundant Power Supplies Larger ClustersActive/Standby Model Point to Point Connections to Storage DevicesEight-Node Cluster with XP or EMC Disk Array Serviceguard Daemons Understanding Serviceguard Software ComponentsServiceguard Architecture Configuration Daemon cmclconfd Cluster Daemon cmcldSyslog Log Daemon cmlogd Cluster Logical Volume Manager Daemon cmlvmdFile Management Daemon cmfileassistd Cluster Object Manager Daemon cmomdNetwork Manager Daemon cmnetd Service Assistant Daemon cmservicedQuorum Server Daemon qs Lock LUN Daemon cmdisklockdProxy Daemon cmproxyd Configuring the ClusterHow the Cluster Manager Works CFS ComponentsHeartbeat Messages Manual Startup of Entire ClusterCluster Quorum to Prevent Split-Brain Syndrome Automatic Cluster StartupDynamic Cluster Re-formation Cluster LockUse of a Lock LUN or LVM Lock Disk as the Cluster Lock Lock RequirementsDual Lock Disk Use of the Quorum Server as the Cluster LockSingle Lock Disk or LUN No Cluster Lock Quorum Server OperationNon-failover Packages How the Package Manager WorksPackage Types Failover PackagesDeciding When and Where to Run and Halt Failover Packages Failover Packages’ Switching BehaviorBefore Package Switching Package Configuration Data Automatic Rotating StandbyRotating Standby Configuration before Failover Configurednode Policy Packages after Failover Failback PolicyAutomatic Failback Configuration After Failover Using Older Package Configuration Files Using the Generic Resources Monitoring ServiceUnderstanding Serviceguard Software Components See also Using Generic Resources to Monitor Volume Groups Using the Event Monitoring ServiceUsing the EMS HA Monitors See also Using EMS to Monitor Volume GroupsHow Packages Run What Makes a Package Run?Before the Control Script Starts Legacy Package Time Line Showing Important EventsDuring Run Script Execution Package Time Line Legacy PackageNormal and Abnormal Exits from the Run Script Service Startup with cmrunservWhile Services are Running When a Package is Halted with a Command During Halt Script ExecutionLegacy Package Time Line for Halt Script Execution Normal and Abnormal Exits from the Halt Script Package Control Script Error and Exit ConditionsError Conditions and Package Movement for Failover Packages How the Network Manager Works Stationary and Relocatable IP AddressesAdding and Deleting Relocatable IP Addresses Types of IP AddressesMonitoring LAN Interfaces and Detecting Failure Link Level Load SharingLocal Switching Cluster Before Local Network Switching Where interface is the primary interface Cmmodnet -e interfaceRemote Switching Monitoring LAN Interfaces and Detecting Failure IP LevelReasons To Use IP Monitoring How the IP Monitor WorksFailure and Recovery Detection Times Constraints and Limitations Example 1 If Local Switching is ConfiguredReporting Link-Level and IP-Level Failures See also Reporting Link-Level and IP-Level FailuresCmmodnet -e lan2 Example 2 If There Is No Local SwitchingAutomatic Port Aggregation Support for HP-UX Vlan Vlan ConfigurationsWhat is VLAN? Types of Redundant Storage Volume Managers for Data StorageConfiguration Restrictions Additional Heartbeat RequirementsWhite papers About Device File Names Device Special FilesExamples of Mirrored Storage Physical Disks Within Shared Storage Units Multiple Devices Configured in Volume Groups Examples of Storage on Disk ArraysMultiple Paths to LUNs Veritas Volume Manager VxVM Types of Volume ManagerHP-UX Logical Volume Manager LVM Propagation of Disk Groups in VxVMPropagation of Disk Groups with CVM Veritas Cluster Volume Manager CVMCluster Startup Time with CVM For heartbeat requirements, see Redundant Heartbeat SubnetsComparison of Volume Managers Redundant Heartbeat SubnetsPros and Cons of Volume Managers with Serviceguard Responses to Failures System Reset When a Node FailsWhat Happens when a Node Times Out ExampleResponses to Hardware Failures Responses to Package and Service Failures Responses to Package and Generic Resources FailuresService Restarts Network Communication FailureGeneral Planning Serviceguard Memory RequirementsPlanning and Documenting an HA Cluster Planning for ExpansionSample Cluster Configuration Hardware PlanningLAN Information SPU InformationNetwork Information Nnn.nnn.nnn.nnnUnder Cluster Configuration Parameters Scsi Addressing in Cluster ConfigurationDisk I/O Information DiskinfoPower Supply Planning Hardware Configuration WorksheetPower Supply Configuration Worksheet Cluster Lock PlanningQuorum Server Worksheet Using a Quorum ServerCluster Lock Disk and Re-formation Time Using Generic Resources to Monitor Volume Groups LVM PlanningUsing EMS to Monitor Volume Groups LVM Worksheet For more information, see Using the EMS HA MonitorsCVM and VxVM Planning Cluster Configuration Planning CVM and VxVM WorksheetPoints To Note Where cDSFs ResideAbout Cluster-wide Device Special Files cDSFs Limitations of cDSFs LVM Commands and cDSFsAbout Easy Deployment Limitations of Easy Deployment Heartbeat Subnet and Cluster Re-formation TimeAdvantages of Easy Deployment Rules and Restrictions for IPv6-Only Mode What Is IPv4-only Mode?What Is IPv6-Only Mode? Localhost ipv6-localhost ipv6-loopbackIPV6 or ANY What Is Mixed Mode?Recommendations for IPv6-Only Mode Cluster configuration file Cluster Configuration ParametersRules and Restrictions for Mixed Mode Name of the cluster as it will appear in the outputPlanning and Documenting an HA Cluster IPv4-Only,IPv6-Only, and Mixed Mode page 106 for Happens when You Change the Quorum ConfigurationGo/hpux-serviceguard-docs under HP Serviceguard 99 and Specifying a Quorum ServerSee also About Hostname Address Families IPv4-Only Hpux-serviceguard-docs under HP ServiceguardIPv6-Only, and Mixed Mode page 106 for important Sitepreferred or Configuration file see Configuring PackagesTheir Services page 227 and these in turn must Configuration Planning page 125 must be specified Cluster Is RunningCluster Configuration Planning Protocols and services. RPC assumes that each network You cannot change the heartbeat configuration whileCVM/CFS on HP Serviceguard A.11.20 April To that LAN, to risk timeout without being servicedSee IPv6 Address Types Also What Happens when You Change the QuorumConfiguration Online page 49 for important information Lock LUN page 189 for more informationFailbackpolicy When You Change the Quorum Configuration OnlineCluster is running, see Updating the Cluster Lock Disk See About Package Weights page 144 for more Planning and Documenting an HA Cluster 69, Monitoring LAN Interfaces and Detecting See also What Happens when a Node Times Out88, Cluster Daemon cmcld page 41, IP-Level FailuresWhen a network interface card has failed Default isConfiguration file specifies one of two ways to decide How Serviceguard will handle the recovery of the primarySee Monitoring LAN Interfaces and Detecting Failure IP Planning and Documenting an HA Cluster Logical Volume and File System Planning Package Configuration PlanningCluster Configuration Next Step Access Control Policies also known as Role Based AccessCVM 4.1 and later without CFS CVM 4.1 and later with CFS About the Volume Monitor Using the Volume MonitorOr --log-file Or --helpOr --version Or --log-levelUsr/sbin/cmvolmond /dev/vg01/lvol1 /dev/vg01/lvol2 VolumepathPlanning for NFS-mounted File Systems Usr/sbin/cmvolmond -t 10 /dev/vg00/lvol1Package Configuration Planning Choosing Switching and Failover Behavior Package Failover BehaviorExtended generic resource Parameters for Configuring Generic ResourcesConfiguring a Generic Resource Cmmakepkg -i $SGCONF/pkg1/pkg1.conf -m sg/genericresourceCmcheckconf -v -P $SGCONF/pkg1/pkg1.conf Cmapplyconf -P $SGCONF/pkg1/pkg1.confCmgetresource -r sfmdisk Cmviewcl -v -f line -p pkg1 grep genericresourceCmrunpkg pkg1 Cmsetresource -r sfmdisk -s upParameters for Configuring EMS Resources Online Reconfiguration of Generic ResourcesRules for Simple Dependencies About Package DependenciesSimple Dependencies Assume that we want to make pkg1 depend on pkg2Planning and Documenting an HA Cluster Dragging Rules for Simple Dependencies Planning and Documenting an HA Cluster Extended Dependencies See Rules for differentnode and anynode Dependencies Rules for Exclusionary DependenciesWhat Happens when a Package Fails Rules for differentnode and anynode DependenciesConfiguring Weights and Capacities About Package WeightsPackage Weights and Node Capacities Cmmakepkg 1m manpageNodename node1 Capacityname packagelimit Weightname packagelimit weightvalueSimple Method For pkg2Defining Capacities Comprehensive MethodPoints to Keep in Mind Clustername cluster23 Nodename node1 Nodename node2Defining Weights Defining Default WeightsWeightname B Weightvalue Weightname a WeightvalueWeightname B Weightvalue Weightname a Rules and Guidelines Cmquerycl 1m manpageAbout External Scripts Pevmonitoringinterval Using Serviceguard Commands in an External Script Lasthaltfailed Determining Why a Package Has Shut DownAbout Cross-Subnet Failover Cmviewcl -v -f line displays a lasthaltfailed flagConfiguring a Package to Fail Over across Subnets Example Implications for Application DeploymentConfiguring monitoredsubnetaccess Configuring a Package Next StepsConfiguring nodename Configuring ipsubnetnodePlanning for Changes in Cluster Size Where Serviceguard Files Are Kept Building an HA Cluster ConfigurationInstalling and Updating Serviceguard Preparing Your Systems Configuring the ClusterCreating cDSFs for a Group of Nodes Creating Cluster-wide Device Special Files cDSFsBefore You Start Etc/cmcluster.confCmpreparecl -n nodename -n nodename Csshsetup -r node2Csshsetup -r -f /etc/cmcluster/sshhosts Cmpreparecl -n node1 -n node2 -n node3 -n node4Adding a Node to a cDSF Group Using Easy DeploymentDisplaying the cDSF Configuration Removing a Node from a cDSF GroupCmquerycl -N $SGCONF/mynetwork Using Easy Deployment Commands to Configure the ClusterFor example Preparing Your Systems Building an HA Cluster Configuration PVG bus1 /dev/cdisk/disk14 /dev/cdisk/disk15 Format for entries in cmclnodelist is as follows Configuring Root-Level AccessAllowing Root Access to an Unconfigured Node Ensuring that the Root User on Another Node Is Recognized About identdAny of the aliases. Examples Configuring Name ResolutionOfficial hostname, as defined by hosts 4, for example Safeguarding against Loss of Name Resolution Services Ensuring Consistency of Kernel Configuration For NIS, enter two linesTuning Network and Kernel Parameters Enabling the Network Time ProtocolCreating Mirrors of Root Logical Volumes Make the new disk a boot diskChoosing Cluster Lock Disks Backing Up Cluster Lock Disk InformationSetting Up a Lock LUN Usr/sbin/idisk -w -p -f partition.txt /dev/rdisk/disk12 Creating a Disk Partition on an HP Integrity SystemUsr/sbin/idisk -w -p -f partition.txt /dev/rdsk/c1t4d0 This will create three device files, for exampleExcluding Devices from Probing Defining the Lock LUNSetting Up and Running the Quorum Server Creating a Storage Infrastructure with LVMUsing the Generic Resources Disk Monitor Using the EMS Disk MonitorUsing Mirrored Individual Data Disks Creating Volume GroupsCreating File Systems Creating Logical VolumesSetting Logical Volume Timeouts Lvchange -t 60 /dev/vg01/lvol1Deactivating the Volume Group Distributing Volume Groups to Other NodesVerify the configuration Distributing the Volume GroupCreate a directory to mount the disk Deactivate the volume group on ftsys10Still on ftsys9, copy the map file to ftsys10 Creating a Storage Infrastructure with VxVM Making Physical Volume Group Files ConsistentCreating Additional Volume Groups Converting Disks from LVM to VxVMCreating Disk Groups Initializing Disks for VxVMInitializing Disks Previously Used by LVM Deporting Disk Groups Re-Importing Disk GroupsHere is an example of the command enter it all one line Configuring the ClusterClearimport at System Reboot Time Cmquerycl -v -C $SGCONF/clust1.conf -n ftsys9 -n ftsys10Specifying the Address Family for the Cluster Hostnames Cmquerycl OptionsSpeeding up the Process Specifying the Address Family for the HeartbeatGenerating a Network Template File Specifying a Lock DiskSpecifying the Cluster Lock Full Network ProbingVgchange -c y /dev/vglock Specifying a Lock LUNCmquerycl -v -n ftsys9 -n ftsys10 See also Choosing Cluster Lock DisksCmquerycl -q QSHost QSAddr -n ftsys9 -n ftsys10 -C Specifying a Quorum ServerObtaining Cross-Subnet Information Will produce the output such as the followingConfiguring the Cluster Controlling Access to the Cluster Specifying Maximum Number of Configured PackagesModifying the Membertimeout Parameter Identifying Heartbeat SubnetsHow Access Roles Work Access RolesLevels of Access Setting up Access-Control PoliciesMonitor Fulladmin Packageadmin Userrole must be one of these three valuesUsername root Username john Userhost bitRole Conflicts Package versus Cluster Roles Adding Volume GroupsVerifying the Cluster Configuration Distributing the Binary Configuration File Differences between Legacy CFS and Modular CFS Storing Volume Group and Cluster Lock Configuration DataModular CFS packages v/s Legacy CFS packages Operational commands for Legacy CFS and Modular CFS Delete a mount point, check point, or snapshot in a package Cfscluster status Cfscluster config -t 900 -sPreparing the Cluster and the System Multi-node Package Cfsdgadm add logdata all=sw Creating the Disk GroupsCreating the Disk Group Cluster Packages Cfsdgadm displayCfsdgadm showpackage logdata Creating VolumesUse the vxprint command to verify Vxprint logfilesCmmakepkg -m sg/cfsall /etc/cmcluster/cfspkg1.ascii Create a package configuration fileFor instructions on creating modular CFS packages, see Cmapplyconf -P /etc/cmcluster/cfspkg1.ascii Apply the package configuration fileCmcheckconf -P /etc/cmcluster/cfspkg1.ascii Cvmconcurrentdgoperations CmviewclBdf CfsconcurrentmountunmountoperationsSee the mountvxfs 1m manpage Cmmakepkg -m sg/cfsall /etc/cmcluster/ckpt1.asciiPackage. For more information, see the manpage Current primary, a primary migration is triggered toCmmakepkg -m sg/cfsall snap1.ascii Create a package configuration file for the snapshot imageVxassist -g cvmdg3 make vol1 100m vxvol -g cvmdg3 startall Information about the mount options, see Mount pointsSnapshotmountoptions Mountvxfs 1m manpageOnline reconfiguration of modular CFS package parameters Cmviewcl -v -f line -p cfspkg1 Cmcheckconf -P cfspkg1.asciiVerify the output Apply the configurationCmapplyconf -P cfspkg1.ascii Legacy Style of Packaging Modular Style of Packaging Managing Disk Groups and Mount Points Using Legacy Packages Associate it with the cluster and mount it Creating Checkpoint and Snapshot Packages for CFSFsckptadm -n create check2 /tmp/logdata/logfiles Cfsmount /tmp/checklogfilesAssociate it with the cluster It is persistentVxassist -g dg1 make vol1 100m vxvol -g dg1 startall Cfsmount /local/snap1 cmviewclYou need to do the tasks described in the following sections Identifying the Master Node Initializing the Veritas Volume ManagerPreparing the Cluster for Use with CVM Initializing Disks for CVMVxdg -s init logdata c0t3d2 Adding Disk Groups to the Package ConfigurationUsr/lib/vxvm/bin/vxdisksetup -i c4t3d4 Mirror Detachment Policies with CVMChecking Cluster Operation with Serviceguard Commands Using Dsau during ConfigurationChecking Cluster Operation with Serviceguard Manager Managing the Running ClusterPreventing Automatic Activation of LVM Volume Groups Setting up Autostart FeaturesManaging a Single-Node Cluster Here is an example of the /etc/rc.config.d/cmcluster fileChanging the System Message Single-Node Operation Deleting the Cluster ConfigurationDisabling identd Change the cmclconfd entry in /etc/inetd.conf toBuilding an HA Cluster Configuration Configuring Packages and Their Services Choosing Package Modules Types of Package Failover, Multi-Node, System Multi-NodeFailoverpolicy Failbackpolicy Ipsubnet Ipaddress Base Package Modules Differences between Failover and Multi-Node PackagesPackage Modules and Parameters Cmmakepkg -m sg/all $SGCONF/sg-allOptional Modules Optional Package ModulesBase Modules Locallanfailoverallowed Package Parameter Explanations ExternalscriptCmmakepkg $SGCONF/sg-all Autorun NodefailfastenabledNodename Runscripttimeout HaltscripttimeoutOperationsequence SuccessorhalttimeoutScriptlogfile LoglevelPriority DependencynameDependencycondition For more information, see About Package DependenciesDependencylocation Specifies where the dependencycondition must be metWeightname, weightvalue Monitoredsubnet MonitoredsubnetaccessLocallanfailoverallowed ClusterinterconnectsubnetIpsubnet Ipaddress New for A.11.18 for both modular and legacy packagesIpsubnet Ipsubnetnode ServicenameSee the package configuration file for more examples IpaddressServicefailfastenabled ServicecmdServicerestart ServicehalttimeoutGenericresourceupcriteria Defines when the status of a generic resource is evaluatedGenericresourceevaluationtype Resourcepollinginterval ResourcenameName of a resource to be monitored ResourcestartConcurrentvgchangeoperations EnablethreadedvgchangeResourceupvalue Cannot lock /etc/lvmconf//lvmlock still tryingVxvolcmd VgchangecmdCvmactivationcmd Vxvmdg KillprocessesaccessingrawdevicesCvmdg VxvmdgretryConcurrentmountandumountoperations Fsfsckopt -s Fstype vxfsConcurrentfsckoperations FsmountretrycountFsdirectory FsnameFsserver FstypeFsfsckopt FsmountoptFsumountopt PevUserrole UsernameUserhost Additional Parameters Used Only by Legacy PackagesCmmakepkg Examples Generating the Package Configuration FileBefore You Start Mkdir $SGCONF/pkg1Editing the Configuration File Next StepPackagetype. Enter failover, multinode, or systemmultinode See About Package Dependencies page 137 for more informationEditing the Configuration File Vg vg01 Vg vg02 Verifying and Applying the Package Configuration # vxdg -tfC import dg01 Adding the Package to the ClusterHow Control Scripts Manage VxVM Disk Groups Viewing Dependencies Cluster and Package MaintenanceReviewing Cluster and Package Status Cmviewcl -r A.11.16Cluster Status Viewing CFS Multi-Node InformationTypes of Cluster and Package States Node Status and StateReviewing Cluster and Package Status Unknown DownNormal Running Status Examples of Cluster and Package StatesFailover and Failback Policies Quorum Server Status CFS Package StatusStatus After Halting a Package Then run cmviewcl -v, we’ll see If we use the following commandStatus After Moving the Package to Another Node After we halt ftsys10 with the following command Status After Auto Run is EnabledOutput of the cmviewcl command is now as follows Status After Halting a NodeViewing Information about System Multi-Node Packages This output can be seen on both ftsys9 and ftsys10Viewing Information about Unowned Packages Cmviewcl -v -p SG-CFS-pkg Checking Status of the Cluster File System CFSStatus of the Packages in a Cluster File System Status of Legacy CVM Disk Group Packages Status of CFS Modular Disk Group and Mount Point PackagesCmviewcl -v -p mpdg1 Ftsys9 Sw swStatus of Legacy CFS Mount Point Packages Checking the Cluster Configuration and ComponentsFtsys10 Cfsmntadm display -v /tmp/logdata/logfilesUser-created files if you specify them Etc/nsswitch.conf Etc/servicesChecking Cluster Components Verifying Cluster Components Cmapplyconf 1mRun cmcheckconf -C Limitations Setting up Periodic Cluster VerificationManaging the Cluster and Nodes See the cron 1m manpage for more informationUsing Serviceguard Commands to Start the Cluster Starting the Cluster When all Nodes are DownAdding Previously Configured Nodes to a Running Cluster Cmruncl -v -n ftsys9 -n ftsys10Cmrunnode -v ftsys8 Removing Nodes from Participation in a Running ClusterHalting the Entire Cluster Cmhaltnode -f -v ftsys9Rules and Restrictions Automatically Restarting the ClusterWhat You Can Do Managing the Cluster and Nodes Additional Points To Note Halting the Cluster and Detaching its Packages Halting a Node and Detaching its PackagesHalting a Detached Package Cmrunnode node1 Managing Packages and ServicesStarting a Package Starting a Package that Has Dependencies Using Serviceguard Commands to Start a PackageHalting a Package Halting a Package that Has DependenciesChanging Package Switching with Serviceguard Commands Changing Package Switching BehaviorUsing Serviceguard Commands to Halt a Package Moving a Failover PackageCmmodpkg -d -n lptest3 pkg1 Maintaining a Package Maintenance ModeSee Performing Maintenance Using Maintenance Mode Cluster and Package Maintenance Performing Maintenance Using Maintenance Mode ProcedureExcluding Modules in Partial-Startup Maintenance Mode Cmrunpkg -m sg/packageip pkg1Cmrunpkg -m sg/services -e sg/packageip pkg1 Reconfiguring a ClusterCmrunpkg -e sg/service pkg1 Types of Changes to the Cluster ConfigurationChange to the Cluster Configuration Previewing the Effect of Cluster ChangesWhat You Can Preview Using Preview mode for Commands and in Serviceguard ManagerMode see Maintaining a Package Maintenance Mode Cmmodpkg -e -t pkg1You would see output something like this Using cmevalCmeval -v newstate.in Updating the Cluster Lock Disk Configuration Online Updating the Cluster Lock ConfigurationReconfiguring a Halted Cluster Updating the Cluster Lock LUN Configuration OnlineAdding Nodes to the Cluster While the Cluster is Running Reconfiguring a Running ClusterCmapplyconf -C clconfig.ascii Cmgetconf -c cluster1 temp.asciiCmquerycl -C clconfig.ascii -c cluster1 -n ftsys8 -n ftsys9 What You Can DoWhat You Must Keep in Mind Cmquerycl -c cluster1 -C clconfig.ascii Example Adding a Heartbeat LANCmgetconf clconfig.ascii Removing a LAN or Vlan Interface from a NodeSee also Replacing LAN or Fibre Channel Cards Changing the LVM Configuration while the Cluster is RunningChanging the VxVM or CVM Storage Configuration Cmgetconf -c clustername clconfig.ascii Configuring a Legacy PackageCreating the Legacy Package Configuration Mkdir /etc/cmcluster/pkg1 Configuring a Package in StagesEditing the Package Configuration File Cluster and Package Maintenance Creating the Package Control Script Cmmakepkg -s /etc/cmcluster/pkg1/pkg1.shCustomizing the Package Control Script Adding Serviceguard Commands in Customer Defined Functions Support for Additional ProductsCopying Package Control Scripts with HP-UX commands Verifying the Package ConfigurationDistributing the Configuration Cmcheckconf -v -P /etc/cmcluster/pkg1/pkg1.confConfiguring Cross-Subnet Failover Configuring nodenameCreating Subnet-Specific Package Control Scripts Reconfiguring a PackageConfiguring monitoredsubnetaccess IP0 = SUBNET0 IP1 = SUBNET1Cmgetconf -p pkg1 pkg1.conf Reconfiguring a Package on a Running ClusterMigrating a Legacy Package to a Modular Package Reconfiguring a Package on a Halted Cluster Adding a Package to a Running ClusterCmapplyconf -v -P app1.conf Deleting a Package from a Running ClusterCmhaltpkg mypkg Cmdeleteconf -p mypkg Unmount the shared file system cfsumount mount pointCmmodpkg -R -s myservice pkg1 Resetting the Service Restart CounterAllowable Package States During Reconfiguration Change servicerestart modular package Types of Changes to PackagesLocallanfailoverallowed Change vxvolcmd Cfsmountoptions Responding to Cluster Events Changes that Will Trigger WarningsSingle-Node Operation Disabling ServiceguardRemoving Serviceguard from a System Start the Cluster using Serviceguard Manager Troubleshooting Your ClusterTesting Cluster Operation Testing the Package ManagerMonitoring Hardware Testing the Cluster ManagerTesting the Network Manager Using EMS Event Monitoring Service Hardware Monitors Using System Fault Management ServiceUsing Event Monitoring Service Hardware Monitors and Persistence RequestsReplacing a Faulty Mechanism in an HA Enclosure Using HP Isee HP Instant Support Enterprise EditionReplacing a Faulty Array Mechanism Replacing DisksReplacing a Lock Disk Replacing a Lock LUNReplacing I/O Cards Online Hardware Maintenance with In-line Scsi TerminatorCmdisklock reset /dev/dsk/c0t1d1 Replacing Scsi Host Bus AdaptersReplacing LAN or Fibre Channel Cards Offline ReplacementOnline Replacement Replacing a Failed Quorum Server System After Replacing the CardReviewing Package IP Addresses Troubleshooting ApproachesUsing cmquerycl and cmcheckconf Using cmviewcl Reviewing the System Log FileFollowing is an example of a successful package starting Reviewing Object Manager Log FilesSample System Log Entries Cmreadlog /var/opt/cmom/cmomd.logUsing the cmcheckconf Command Reviewing Serviceguard Manager Log FilesReviewing Configuration Files Reviewing the System Multi-node Package FilesReviewing the LAN Configuration Solving ProblemsUsing the cmviewconf Command Serviceguard Command HangsNslookup ftsys9 Networking and Security Configuration ErrorsCluster Re-formations Caused by Temporary Conditions System Administration Errors Package Control Script Hangs or FailuresFuser -kulogical-volume umount logical-volume Problems with Cluster File System CFS Llt, gab Vxfen W cvm CfsNode and Network Failures Problems with VxVM Disk GroupsPackage Movement Errors Force Import and Deport After Node FailureTimeout Problems Troubleshooting the Quorum ServerAuthorization File Problems Access denied to quorum serverMessages Enterprise Cluster Master Toolkit Designing Highly Available Cluster Applications Automating Application OperationDefine Application Startup and Shutdown Controlling the Speed of Application FailoverInsulate Users from Outages Evaluate the Use of JFS Use Raw VolumesReplicate Non-Data File Systems Minimize Data LossDesign for Multiple Servers Use Restartable TransactionsUse Checkpoints Balance Checkpoint Frequency with PerformanceAvoid Node-Specific Information Designing Applications to Run on Multiple SystemsDesign for Replicated Data Sites Obtain Enough IP Addresses Avoid Using SPU IDs or MAC AddressesAssign Unique Names to Applications Allow Multiple Instances on Same SystemBind to Relocatable IP Addresses Use uname2 With CareBind to a Fixed Port Avoid File Locking Give Each Application its Own Volume GroupUse Multiple Destinations for SNA Applications Call bind before connectHelp menu for ndd -h ipstrongesmodel Etc/rc.config.d/nddconf as followsUsr/sbin/route add net default 128.17.17.1 1 source Usr/sbin/route delete net default 128.17.17.1 1 source Restoring Client ConnectionsBe Able to Monitor Applications Handling Application FailuresCreate Applications to be Failure Tolerant Minimizing Planned Downtime Reducing Time Needed for Application Upgrades and PatchesProvide for Rolling Upgrades Do Not Change the Data Layout Between ReleasesProviding Online Application Reconfiguration Documenting Maintenance OperationsDefining Baseline Application Behavior on a Single System Integrating HA Applications with ServiceguardChecklist for Integrating HA Applications Integrating HA Applications in Multiple Systems Testing the Cluster Move it backSpecial Considerations for Upgrade to Serviceguard A.11.19 Software UpgradesSpecial Considerations for Upgrade to Serviceguard A.11.20 Rolling Upgrade Types of UpgradeHow To Tell when the Cluster Re-formation Is Complete Rolling Upgrade Using DRDNon-Rolling Upgrade Guidelines for Rolling UpgradeRestrictions for DRD Upgrades Non-Rolling Upgrade Using DRDPerforming a Rolling Upgrade Limitations of Rolling UpgradesMigrating cmclnodelist entries from A.11.15 or earlier Running the Rolling UpgradeKeeping Kernels Consistent Performing a Rolling Upgrade Using DRD Running the Rolling Upgrade Using DRDHalt the first node, as follows Example of a Rolling UpgradeStep Running Cluster with Packages Moved to Node Node 1 Rejoining the Cluster Repeat the process on node 2. Halt the node, as followsGuidelines for Non-Rolling Upgrade Performing a Non-Rolling UpgradeSteps for a Non-Rolling Upgrade Using DRD Performing a Non-Rolling Upgrade Using DRDLimitations of Non-Rolling Upgrades using DRD Guidelines for Migrating a Cluster with Cold Install Checklist for MigrationWorksheet for Hardware Planning Power Supply WorksheetBlank Planning Worksheets LVM Volume Group and Physical Volume Worksheet Quorum Server WorksheetCluster Configuration Worksheet VxVM Disk Group and Disk WorksheetPackage Configuration Worksheet Package Configuration Worksheet Loading VxVM Migrating Volume GroupsMigrating from LVM to VxVM Data Storage Customizing Packages for VxVM Mntdg0202, respectivelyRestart the package Removing LVM Volume GroupsCustomizing Packages for CVM Migrating from Legacy CFS Packages to Modular CFS Packages Textual Representation of IPv6 Addresses IPv6 Network SupportIPv6 Address Types IPv4 and IPv6 Compatibility IPv6 Address PrefixUnicast Addresses IPv4 Compatible IPv6 AddressesSite-Local Addresses Aggregatable Global Unicast AddressesLink-Local Addresses Multicast AddressesNetwork Configuration Restrictions Ndd -get /dev/ip6 ip6nddadsolicitcount Example ConfigurationsLocal Primary/Standby LAN Patterns Ndd -set /dev/ip6 ip6nddadsolicitcountnExample Configurations 384 IPv6 Network Support Accessing Serviceguard Manager Using Serviceguard ManagerBefore Using HP Serviceguard Manager Setting Up About the Online Help SystemScenario 1 Single cluster management Launching Serviceguard ManagerAccessing Serviceguard Manager Opt/hpsmh/bin/hpsmh autostartSystem Management Homepage with Serviceguard Manager From the left-hand panel, expand Cluster by Type Expand HP Serviceguard, and click on a Serviceguard clusterSign Membertimeout Maximum and Minimum Values for ParametersLaunching Monitoring Scripts Monitoring Script for Generic ResourcesSample scripts Launching Monitoring Scripts Template of a Monitoring Script I L I T Y N C T I O N S Monitoring Script for Generic Resources Template of a Monitoring Script Migrating EMS Resources to Generic Resources Start the package Identify the equivalent SFM style resource monitorIndex APA399 Cvmactivationcmd Firstclusterlockpv LAN INONLYORINOUT, 69 Inout Pollingtarget defined Qsaddr Servicename Vxvmdg
Related manuals
Manual 32 pages 5.87 Kb