PRIMECLUSTER Installation and Administration Guide 4.1 (for Solaris(TM) Operating System)

6.7 Setting Up Patrol Diagnosis

The patrol diagnosis facility periodically diagnoses the following hardware units that are connected to the STANDBY node.

• Shared disk units

  The function diagnoses whether a shared disk unit has become unusable because the power is switched off, a cable is disconnected (adapter side or device side) or because of some other reason.

  If the diagnosis results indicate that an error was detected in a shared disk unit, a message is output to "syslogd(1M)" and the CRM main window.

• Network interface cards

  The function diagnoses whether any network interface card cannot communicate because a cable is disconnected or because of some other reason.

  If the patrol diagnosis results indicate that an error was detected in a network interface card, a message is output to "syslogd(1M)" and the CRM main window, and brings userApplication Faulted.

  If a cluster application uses a Global Link Services (GLS) resource, patrol diagnosis is disabled for the network interface card that is managed by the Gls resource. If a cluster application uses a Gls resource, use the GLS monitoring function. For details on the GLS monitoring function, see "Failure Monitoring Function" and "Standby Patrol Function" in the "PRIMECLUSTER Global Link Services Configuration and Administration Guide: Redundant Line Control Function."

When the Patrol Diagnosis Facility detects a hardware error, this error affects the Patrol Diagnosis Facility resource of a cluster application. This Patrol Diagnosis Facility resource is registered when the cluster application is created and a user enables the Patrol Diagnosis.

Set up patrol diagnosis according to the following flow:

1. Register the patrol diagnosis resource to userApplication.
2. Set up the patrol diagnosis configuration.

• Set up patrol diagnosis after initial setup of the cluster resource manager is completed. If the patrol diagnosis setup command (clspconfig) is executed before initial setup of the cluster resource manager, the following message is displayed:

  6615: The cluster configuration management facility is not running.

• If you do not configure the patrol diagnosis configuration, the patrol diagnosis facility will not diagnose the hardware.
• To view the manual pages of the individual commands, add "/etc/opt/FJSVcluster/man" to the MANPATH variable.

Registering a patrol diagnosis resource to a userApplication

To register a patrol diagnosis resource to userApplication, use the userApplication Configuration Wizard (GUI) or the RMS Wizard (CUI).

For instructions on resource registration with the userApplication Configuration Wizard (GUI), see "Creating Cluster Applications."

For instructions on resource registration with the RMS Wizard (CUI), see "/usr/opt/reliant/htdocs.solaris/wizards.en/wizards/CRM.htm."

Setting the patrol diagnosis configuration

Before starting RMS, set up the following configuration for patrol diagnosis by executing the "clspconfig(1M)" command on an arbitrary node.

For details on the "clspconfig(1M)" command, see the manual page for "clspconfig(1M)."

Specifying the target hardware for patrol diagnosis execution

Specify the hardware product that runs patrol diagnosis if either of the following conditions applies. This hardware is used by the specified userApplication.

• If the cluster application uses a Gds (Global Disk Services) resource
• If the cluster application uses an Ipaddress (takeover network) resource

Operation Procedure: (When the cluster application uses Gds resources)

If the cluster application is using Gds resources, specify the physical disks that were defined to the Gds resources as hardware for patrol diagnosis execution. The physical disks can be obtained by the procedure below.

1. Confirmation of the disk class

   Check the disk class to be used by the cluster application. The disk class is the class that was specified in "Creating Gds Resources."

   Example) class0001

2. Confirmation of the physical disks

   Check the physical disks that were defined in the disk class. The physical disks are those that were defined in "Setting Up Shared Disks." Use the class configuration setup screen or the "sdxinfo(1M)" command to check the physical disks. For information on the class configuration setup screen, see "Setting Up Shared Disks." For details on the "sdxinfo(1M)" command, see the "PRIMECLUSTER Global Disk Services Configuration and Administration Guide."

   Example) c1t4d0, mplb2048

3. Setup of the target hardware for patrol diagnosis execution

   Set the physical disks that were defined in the disk class as hardware for patrol diagnosis execution.

   Example 1) For c1t4d0

   # /etc/opt/FJSVcluster/bin/clspconfig -u app1 TARGET_DISK=c1t4d0 TARGET_DISK=c1t4d1

   Example 2) For mplb2048

   # /etc/opt/FJSVcluster/bin/clspconfig -u app1 TARGET_DISK=mplb2048

Operation Procedure: (When the cluster application uses Ipaddress resources)

If the cluster application is using Ipaddress resources, specify the network interface card to be used in the takeover network as hardware for patrol diagnosis execution. Follow the procedure below to determine the network interface card.

1. Confirmation of the network interface card

   Check the network interface card to be used by the takeover network. The network interface card is the interface card that was selected in "Selecting an Interface" of "Creating Takeover Network Resources."

   Example) hme0

2. Setup of target hardware for patrol diagnosis execution

   Configure the network interface card to be used by the takeover network as hardware for patrol diagnosis execution.

   Example)

   # /etc/opt/FJSVcluster/bin/clspconfig -u app1 TARGET_LAN=hme0

Time interval for patrol diagnosis

Specify the time interval for patrol diagnosis. If this setting is not specified, patrol diagnosis runs at 360-minute intervals.

Example) To run patrol diagnosis at 60-minute intervals

# /etc/opt/FJSVcluster/bin/clspconfig -u app1 INTERVAL=60

You can also check the current setting.

Example) Checking the current setting (INTERVAL=360)

# /etc/opt/FJSVcluster/bin/clspconfig -u app1

ACTIVE=true

INTERVAL=360

TARGET_LAN=hme1

TARGET_DISK=c1t4d0

TARGET_DISK=c1t4d1

TARGET_DISK=c1t4d2

TARGET_DISK=c1t4d3

Stopping and restarting patrol diagnosis

Stop and restart patrol diagnosis by executing the "clspconfig(1M)" command. For details on the "clspconfig(1M)" command, see the manual page for "clspconfig(1M)."

To stop patrol diagnosis, specify "ACTIVE=false" as an operand in the "clspconfig(1M)" command.

Example)

# /etc/opt/FJSVcluster/bin/clspconfig -u app1 ACTIVE=false

To restart patrol diagnosis, specify "ACTIVE=true" as an operand in the "clspconfig(1M)" command.

Patrol diagnosis is restarted from the next time with the same configuration as when patrol diagnosis was stopped, including hardware for executing patrol diagnosis and the time interval for patrol diagnosis.

Example)

# /etc/opt/FJSVcluster/bin/clspconfig -u app1 ACTIVE=true