Host view

The host view pages in the cluster management console give you all the information you need to keep an eye on your entire cluster and to quickly and easily spot those hosts that are experiencing any difficulties.

Healthy hosts mean that your workload is running optimally and you are making the best possible use of your resources. If your hosts are closed or unavailable, your workload suffers. The more hosts you have available to run workload units, the more robust your cluster becomes. The first step in keeping your cluster healthy is to monitor the health of your hosts.

On the host view pages:

Your primary host is always displayed as the last icon on your host view. For details, rest your cursor on the icon. For primary host properties, click the icon.
For details about the performance of a specific host:
- In the Host (Icon View) page, rest your cursor on the icon or click the icon to get host properties.
- In the Host (List View) page, click the host name to get host properties.
Use Filter Settings to customize the view of your hosts. For example, you can filter hosts by state to display only unavailable or closed hosts.
Customize the metrics you see in the Hosts (List View) by clicking Preferences.
Note: To change the maximum number of rows that show on each page, change the value for the EGOHostListMaxRowPerPage parameter in the pmc_conf.xml file, which is found by default at EGO_TOP/gui/conf/pmcconf/pmc_conf_ego.xml.
The default maximum number of rows on each page is 100.

How can I tell my hosts are in trouble?

Your cluster only works as well as your hosts are performing. Use the following hints to get an idea of how you can use the cluster management console to monitor your host health.

Pattern: Use the icon view to display your hosts and their health visually. It is the pattern of the health of the hosts that is important to maintaining a healthy cluster. There should be few hosts Unavailable or Closed compared to OK hosts.
Height: The height of the icons in the icon view gives you the CPU utilization of that host. Generally, in a healthy cluster, most hosts have between 7 and 9 bars in height, meaning they are using between 70% and 90% CPU utilization. Depending on the workload you are running, these thresholds may not be correct for you; they are general guidelines only.
The primary host: Keep an eye on your primary host and make sure it is not overloaded. Candidates are designed to back up your cluster should a primary host become unavailable for any reason, but it is recommended that you safeguard the health of your primary host as much as possible.