Test 237: pSeries HACMP for AIX 5L

Section 1 - Planning and Design

1. Evaluate the potential HACMP nodes' ability to service applications (e.g., CPU, memory, I/O bandwidth, in a failed over condition)
2. Evaluate potential HACMP nodes for additional devices (e.g., tape, X.25, SNA, Async device)
3. Evaluate potential HACMP nodes for host names, uname, node names and service labels
4. Evaluate application compatibility with planned OS and HACMP versions
5. Evaluate network components such as the type of IP network, and the use of virtual IP addresses
6. Evaluate user connectivity to the node
7. Evaluate network components for single point of failure at the physical and logical layers
8. Evaluate requirements of a non-IP network
9. Evaluate the use of IPAT
10. Evaluate network security
11. Plan LVM (Logical Volume Manager) requirements (e.g., disk connections, FAStT limitations, volume groups, raw logical volumes, files system mount points, recovery type, sequential/parallel, root volume group mirrored)
12. Evaluate environmental points of failure (e.g., power, air conditioning, building)
13. Evaluate application scripts (e.g., application start scripts, application stop scripts, application monitor, application restart scripts, application cleanup scripts)
14. Evaluate resource requirements and behavior (e.g., DNP, sites, NFS, GMD, PPRC, GPFS, cron, time, communication links)
15. Assess HACMP suitability and implementation options (e.g., HACMP/XD: HAGEO, HACMP/XD: PPRC)

Section 2 - Installation/Configuration

1. Configure TCP/IP networks
2. Configure and test Non-IP networks
3. Create shared LVM configuration (e.g., manual and automatic configuration)
4. Determine which updates are necessary for an HACMP installation (e.g., firmware, adapter microcode, operating system)
5. Verify HACMP pre-reqrequisites (e.g., bos.adt, bos.data, RSCT)
6. Based on HACMP requirements, determine which components of HACMP to install and/or update
7. Define the cluster
8. Select communication devices (e.g., RS232, TMSSA, TMSCSI, Disk HB)
9. Configure cluster resources
10. Convert volume groups to enhanced concurrent access

Section 3 - Testing and Verification

1. Evaluate successful cluster start
2. Perform cluster shutdowns (e.g., graceful, takeover, forced, induced)
3. Verify successful node reintegration including forced stops
4. Verify the cluster behavior after IP failure (e.g., switch, cable, adapter, multiple adapter, etc.)
5. Verify the cluster behavior after non-IP failure (e.g., disk HB)
6. Emulate errlog entries
7. Determine whether application monitors function as designed
8. Validate rootvg mirror
9. Verify site failure
10. Determine cluster status (e.g., clstat, snmpinfo)
11. Verify that the client applications work
12. Verify nfs functionality

Section 4 - Post Implementation/Administration

1. Define custom cluster events
2. Configure pre and post events
3. Configure error notification
4. Implement application monitoring
5. Create and manage cluster snapshots
6. Employ C-SPOC for filesystem maintenance
7. Manage Clustered users and groups using C-SPOC
8. Determine which HA logs and/or system error logs should be used for problem determination
9. Manage resource groups
10. Start and stop clusters
11. Perform periodic cluster verification
12. Perform routine cluster monitoring
13. Plan and implement software updates (AIX, HACMP)
14. Plan and implement hardware replacement and changes
15. Determine and recover from an unsuccessful cluster event
16. Recover from a failed DARE operation

 

 

1. A planning session is setup to evaluate a customer's systems for the potential of a two-node cluster. During the evaluation, a nonstandard disk subsystem is discovered. The disk devices have a unique type in the ODM database. They require that specific commands be executed in order to make the disks available on system startup. Other commands are also available to identify IO rates, manage reservation locks and lock resets, manage disk capacity, and read the configuration from the disk. What limitations does HACMP impose on disk subsystem usage?
2. Only disks that can be identified as IBM or other authorized vendors can be used with HACMP.
3. For a disk to function under the control of HACMP, all disks must subscribe to the SCSI protocol regardless of physical connection type.
4. Unusual disk types are allowed as long as they can be identified in the CuDv with the PdDvLn field and have the necessary support programs.
5. There are no restrictions on the disk type that can be used with HACMP, as long as the device works correctly in a single system instance it will work with HACMP.

1. A customer is configuring the network components for an HACMP cluster. The plan is to use a four-port Ethernet adapter with IPAT via IP aliasing for the three service IP labels. Which of the following would be a concern with the proposed configuration?
2. The four-port adapter is not supported.
3. A single physical adapter is not supported.
4. The four-port adapter is a single point of failure.
5. There should be a "spare" port for each service IP label.

1. Which statement is true regarding hearbeats over IP aliases?
2. The Heartbeat IP aliases must be defined in the /etc/host file.
3. The service and non-service IP labels have no subnet restrictions.
4. The service and non-service IP labels must be on separate subnets.
5. The Heartbeat IP alias address can also be used to carry other traffic.

1. During a planning session, the customer expresses a concern about the security of the networks managed by HACMP. Specifically, the customer is concerned about the methods used by HACMP for synchronization of configuration information. How do HACMP nodes exchange configuration information?
2. On a public port with security records in /.rhosts
3. On a private port with security records in /.rhosts
4. On a public port with security records in /usr/es/sbin/cluster/etc/rhosts
5. On a private port with security records in /usr/es/sbin/cluster/etc/rhosts

1. The customer has a high volume application. GPFS was implemented across three nodes to provide the application with the required performance level. If a node fails, the customer wants the remaining nodes to assume the identify of the failed node. Which type of resource group can be used in this environment?
2. Rotating resource groups
3. Parallel resource groups
4. Cascading resource groups
5. Concurrent resource groups

1. A customer has an application that is 90% reads and 10% writes on a 35 Terabyte database. During a conversation with the customer, it was discovered that HACMP/XD:HAGeo was being considered but size of the database would make it too expensive. What facility of HACMP/XD:HAGeo needs to be clarified with the customer?
2. HACMP/XD:HAGeo only mirrors the writes across the geography after the initial synchronization.
3. HACMP/XD:HAGeo uses the all systems in the geography for reads the initial synchronization to increase performance.
4. HACMP/XD:HAGeo mirrors the read and writes across the geography after the initial synchronization for better performance.
5. HACMP/XD:HAGeo only mirrors the reads across the geography after the initial synchronization if the local system is slow.

1. A customer wants to install HACMP but has limited resources. The only system available to make a backup node is constrained by IO slots. There will only be space for one Ethernet adapter in the backup node. Why this configuration is problematic?
2. A single adapter node will not work.
3. A single adapter node will work if IPAT via IP aliasing is used.
4. A single adapter node will work if IPAT via IP replacement is used.
5. A single adapter node will work but will be a single point of failure.

1. A customer is moving two of four nodes in a cluster to a new computer room three floors down in the current building. The separation was prompted by a lack of physical space in the original computer room. The intention is to add separate power and air conditioning to the second computer room, making the cluster more highly available. Connection to the shared disk subsystems was not affected by the move, and the length of cables is within the disk subsystem specification. The RS232 heartbeat cable has exceeded the length, and RS232 line drivers have been used to extend the distance. Under what condition will the RS232 line drivers function properly with HACMP in this situation?
2. With no configuration adjustments
3. If the speed of the network is set in the Network Module
4. If the speed of the line driver device is set to 8,1,none and a speed of 9600
5. If the speed of the line driver device is set to 8,1,none and a speed of 38400

1. A customer has installed a cluster with all of the LPP's on the installation media that would install without error. The normal startup of the cluster includes all of the processes except the "Lock Manager". If the customer wishes to monitor the cluster from a "Windows" based system, what additional step needs to be performed?
2. The clstat application needs to be compiled in a "Windows" environment.
3. The clinfo application needs to be compiled in a "Windows" environment.
4. A native "Windows" SNMP client agent needs to be configured on the "Windows" client.
5. A Web server with the appropriate cgi application needs to be configured on a cluster node.

1. An administrator has a requirement to monitor the cluster status from a remote AIX client. It has been decided the cgi version of clstat will be available on the Web server cluster node. In order to monitor the cluster from the client, what cluster software needs to be installed?
2. clstat.jar
3. cluster.es.client
4. cluster.adt.client
5. none, just the Web browser

1. A customer is implementing a new two-node cluster and wants the most highly available solution including Fast Disk Takeover. The cluster has the following:
2. storage connected by a fibre switch

• asynchronous adapter with unused ports
• unused 10/100 Ethernet
• a pair of 1000 Ethernet adapters on each node

1. Best practice for the most robust heartbeat configuration would be to use the:}
2. async ports and disk heartbeating.
3. spare Ethernet connection and disk heartbeating.
4. spare Ethernet connection and async ports for heartbeating.
5. spare Ethernet connection and the 1000 Ethernet connection for heartbeating.

1. A customer has had a cluster for a long time and has upgraded it many times over the years. The cluster currently has 10 resource groups, each with several Volume Groups. The upgrading process has defaulted the resource groups to process sequentially. To speed up the completion time for acquiring resource groups, an investigation has started. The customer does not wish to use enhanced concurrent mode for the Volume Groups. What additional feature of HACMP may impact the resource acquisition time?
2. Nothing can be done to speed up the resource acquisition because the time is limited to the disk type.
3. The processing time can be reduced by changing the resource groups to use the fast disk takeover feature.
4. By changing the resource group processing order to not use an ordered list, the processing time should be reduced.
5. Decreasing the time in the network modules will facilitate faster detection of network failures and speed up the over all process.

1. Which of the following is true about cluster startup?
2. Cluster nodes are started in parallel.
3. Cluster nodes are started sequentially.
4. Cluster nodes started in parallel cannot use auto-restart.
5. Cluster nodes using settling timers cannot start sequentially.

1. The customer stopped a node in a cluster with the "shutdown -F" command. What will happen to the resource groups on the node that stopped?
2. The resource groups will start on the next available node after the settling timer has expired.
3. The resource groups will go off line unless the inactive takeover flag is set in the resource group.
4. The resource groups will be brought on line on the next available node as configured in the resource group.
5. The resource groups will go off line unless the inactive takeover flag is set in the HACMP run-time parameters.

1. A node on a running cluster recently re-integrated into the cluster. An administrator can determine if the resource groups have returned to their expected locations by using which command?
2. cldare
3. clRGinfo
4. cllistres
5. clshowres

1. An administrator is working on the network interfaces of a cluster node that uses IPAT via IP aliasing. The administrator changes one of the interfaces in use by HACMP from "up" to "down" with the ifconfig command. After noticing the interface is down, what will HACMP do?
2. Move any IP service label to another available interface.
3. Move any resource group using that interface to another node.
4. Issue a command to change the interface to an "up" condition, and, if successful, will not move any IP service labels.
5. Move any IP service label to another available interface on the same node, and then issue a command to change the interface to an "up" state.

1. A customer is testing a cluster before putting it into production. The customer wants to test the resource group fallover based on a volume group loss of quorum. The recommended method for testing resource group fallover for loss of quorum is to:
2. force vary off the volume groups.
3. emulate error notification function in HACMP.
4. disconnect the cables leading to the disk subsystem.
5. remove disk drives from the system until loss of quorum is reached.

1. To which logfile will the output of the custom application monitor be logged?
2. None
3. /tmp/hacmp.out
4. /var/hacmp/app_mon.log
5. /usr/es/adm/cluster.log

1. Which of the following commands should be used to verify that the rootvg is successfully mirrored?
2. lsvg -l rootvg
3. lspv -M rootvg
4. lsmirror -l rootvg}
5. lqueryvg -Mr rootvg

1. What utility is used to determine the status of the remote site in a HACMP/XD:HAGeo cluster?
2. clstat
3. clsite
4. clRGinfo
5. clsiteinfo

1. To provide additional cluster monitoring, a Tivoli Netview solution has been proposed. The administrator has been tasked with the configuration issues. One issue is the cluster must be modified to send traps to the Tivoli Management Console. On the cluster nodes, what file must be modified to use traps?
2. /etc/snmpd.conf
3. /etc/inetd.conf
4. /usr/es/sbin/cluster/etc/clhosts
5. /usr/es/sbin/cluster/etc/cltopinfo

1. An administrator wants to see the current state of a cluster and resource groups, 'rg1' and 'rg2'. Which of the following commands, executed at the command-line, will display the current state of the cluster and resource groups?
2. cldump
3. clfindres -v
4. cllistclstate -g
5. clgetrg -r rg1 and clgetrg -r rg2

1. In an attempt to improve a cluster client's application availability, the client node is being made cluster aware by installing the HACMP client software on an AIX client. To enable a client node to detect cluster state changes in a running cluster, which file must be configured?
2. /etc/hosts
3. /etc/snmpd.conf
4. /usr/es/sbin/cluster/etc/rhosts
5. /usr/es/sbin/cluster/etc/clhosts

1. A customer is heavily dependant on HACMP providing highly available NFS services. What command can be used to verify the NFS exports are correct?
2. lsnfs
3. showmount -e
4. cat /etc/exports
5. cat /usr/es/sbin/cluster/etc/exports

1. A two-node cluster has been installed with five resource groups. Only one resource group has a service IP label. If a network error occurs the resource group with the service IP label migrates to the backup node. The administrator would like to move all the resource groups when an event like this occurs. What is the best way to accomplish this?
2. Modify the crontab and inittab.
3. Configure pre or post event scripts.
4. Add event for hardware failure using smit.
5. Set the /usr/es/sbin/cluster/diag/cf_failover script.

1. Which of the following statements regarding Application Monitoring is specific to a custom resource monitor?
2. It can monitor more subtle aspects of application performance.
3. It may not be an appropriate option for all types of applications.
4. It can detect the death of one or more processes of an application using RSCT
5. It is easier to set up since it uses custom plugins for monitoring capability and requires no scripts.

1. To handle an application's growing demand for storage, a new volume group is being planned in an existing cluster environment. System outages are limited to one weekend every other month. How can HACMP be used to manage the downtime limitations?
2. Create the VG using the C-SPOC option Create a Shared Volume Group.
3. Create the VG using the C-SPOC option Import a Shared Volume Group.
4. Create the VG outside of HACMP and add it into HACMP during next outage.
5. Create the VG outside of HACMP let the HACMP Lazy Update feature find it.

1. A cluster node has been started and has not acquired the resources in the resource groups. Which log file will show the details of any event script processing that has occurred?
2. /tmp/hacmp.out
3. /tmp/clstrmgr.debug
4. /usr/es/adm/cluster.log
5. /var/adm/hacmp/hacmp.out

1. Which command can be used at the command line to move a resource group?
2. cldare
3. rgmove
4. rg_move
5. clRGmove

1. A stable cluster has not had an opportunity to fallover in several months. What would be a non-disruptive method of checking that the cluster is ready to function as expected?
2. Fail a node.
3. Run the clstat utility.
4. Run the clverify utility.
5. Perform cluster synchronization.

1. A customer with a multi-site HACMP cluster was changing resource groups to add another IP service label to a resource group. An error occurred during synchronization. Since the last node was added at the remote site, the cluster had previously synchronized correctly and had been running successfully for several months. What is the most likely cause of the synchronization error?
2. When using the site parameter, DARE is not allowed.
3. The service IP labels were not properly defined to HACMP
4. Changing the resource groups with the cluster running is not allowed.
5. The resource group can only have one service IP label for each resource group.

Answer Key:

1. C
2. C
3. B
4. D
5. C
6. A
7. D
8. B
9. D
10. D
11. A
12. C
13. B
14. B
15. B
16. A
17. B
18. A
19. A
20. A
21. A
22. A
23. D
24. B
25. B
26. A
27. A
28. A
29. D
30. C
31. A