Administration Guide

Monitoring HACMP Clusters

Scripts are provided for creating SP problem management (pman) events to monitor the DB2 HACMP ES configuration, in addition to those monitoring utilities already present in HACMP ES. To monitor HACMP status from the SP control workstation:

• Install the HACMP client code on the control workstation.
• Edit the /usr/sbin/cluster/etc/clhosts file, and include the SP ethernet IP addresses of the nodes that you want to monitor.
• Invoke the command startsrc -s clinfo to start monitoring the clusters.

HACMP supplies an interface for monitoring the clusters (/usr/sbin/cluster/clstat.

To use the problem management monitoring with SP Perspectives GUI for HACMP RS and user-defined events:

1. Invoke create_db2_events <nodelist>, where nodelist contains pcp or pexec style nodes. This script creates five pman events for monitoring by Perspectives.

   Note:

   The resource variables PSSP.pm.User_state12-16 are used in the creation of these events. If these resource variables are already being used for some other purpose, create_db2_events and update_db2_events must be updated to use different resource variables.

2. Start Perspectives on the control workstation. From the launch pad, choose the event perspective. You should see five events: db2_hacmp_recovery, db2_process_recovery, db2_paging_err, db2_nfs_err, and Errlog_PERM_entry.
3. Double-click on each event. On the screen that appears, register (within the Definition Table) a condition for the event. Click next to the down arrow by Name: "unnamed", and select the same name as the event you specify as the condition. Select the "Response Options" tab. Click on the button at the top of the display ("Send Message to Perspectives event session"). You can specify commands, errlog entries, as well as SNMP traps for these event occurrences. The event log displays are maintained only across Perspective sessions; therefore, you might want to create AIX error log entries for each. Select OK, and close the window.
4. From the Perspectives launch pad, select the hardware Perspective.
5. When the hardware frame GUI appears, select "View" and then "Monitor". You are provided with a list of events that can be monitored for your SP. Scrolling to the bottom of the list, you will find two additional events: one for HACMP DB2 recovery (db2_ha_ind), and the other for SP node PERM errors (Errlog_PERM_mon. Select those that you want to monitor. (When an event occurs, the node displays a red "X". If all monitored conditions are fine, the node display is green.) host_responds, switch_responds, and node_power_LED are typically used. You can also monitor the DB2 HACMP recovery, as well as PERM errors, on the node.

   haind:2:wait:/usr/bin/db2_update_events HAIND OFF 2>&1 >/dev/null

If you are planning to use NetView for your implementation, consider using HAVIEW (which is part of HACMP) for monitoring your configuration. For information about configuring that product, refer to the NetView documentation.