Tuesday, September 9, 2008

Frequently Asking questions in HACMP

a. What characters should a hostname contain for HACMP configuration?
The hostname cannot have following characters: -, _, * or other special characters.

b. Can Service IP and Boot IP be in same subnet?

No. The service IP address and Boot IP address cannot be in same subnet. This is the basic requirement for HACMP cluster configuration. The verification process does not allow the IP addresses to be in same subnet and cluster will not start.

c. Can multiple Service IP addresses be configured on single Ethernet cards?

Yes. Using SMIT menu, it can be configured to have multiple Service IP addresses running on single Ethernet card. It only requires selecting same network name for specific Service IP addresses in SMIT menu.

d. What happens when a NIC having Service IP goes down?

When a NIC card running the Service IP address goes down, the HACMP detects the failure and fails over the service IP address to available standby NIC on same node or to another node in the cluster.

e. Can Multiple Oracle Database instances be configured on single node of HACMP cluster?

Yes. Multiple Database instances can be configured on single node of HACMP cluster. For this one needs to have separate Service IP addresses over which the listeners for every Oracle Database will run. Hence one can have separate Resource groups which will own each Oracle instance. This configuration will be useful if there is a failure of single Oracle Database instance on one node to be failed over to another node without disturbing other running Oracle instances.

f. Can HACMP be configured in Active - Passive configuration?

Yes. For Active - In Passive cluster configuration, do not configure any Service IP on the passive node. Also for all the resource groups on the Active node please specify the passive node as the next node in the priority to take over in the event of failure of active node.

g. Can file system mounted over NFS protocol be used for Disk Heartbeat?

No. The Volume mounted over NFS protocol is a file system for AIX, and since disk device is required for Enhanced concurrent capable volume group for disk heartbeat the NFS file system cannot be used for configuring the disk heartbeat. One needs to provide disk device to AIX hosts over FCP or iSCSI protocol.

h. Which are the HACMP log files available for troubleshooting?

Following are log files which can be used for troubleshooting:
1. /var/hacmp/clverify/current//* contains logs from current execution of cluster verification.
2. /var/hacmp/clverify/pass//* contains logs from the last time verification passed.
3. /var/hacmp/clverify/fail//* contains logs from the last time verification failed.
4. /tmp/hacmp.out file records the output generated by the event scripts of HACMP as they execute.
5. /tmp/clstmgr.debug file contains time-stamped messages generated by HACMP clstrmgrES activity.
6. /tmp/cspoc.log file contains messages generated by HACMP C-SPOC commands.
7. /usr/es/adm/cluster.log file is the main HACMP log file. HACMP error messages and messages about HACMP related events are appended to this log.
8. /var/adm/clavan.log file keeps track of when each application that is managed by HACMP is started or stopped and when the node stops on which an application is running.
9. /var/hacmp/clcomd/clcomd.log file contains messages generated by HACMP cluster communication daemon.
10. /var/ha/log/grpsvcs. file tracks the execution of internal activities of the grpsvcs daemon.
11. /var/ha/log/topsvcs. file tracks the execution of internal activities of the topsvcs daemon.
12. /var/ha/log/grpglsm file tracks the execution of internal activities of grpglsm daemon.


People Interested in Detailed study on HACMP Please read :

http://www.redbooks.ibm.com/redbooks/pdfs/sg246375.pdf

No comments: