AIX LPAR Statistics Test

When applications executing on one/more LPARs experience sudden or sporadic slowdowns, administrators need to know how each of these LPARs is utilizing the processing capacity (i.e., CPU resources) assigned to it so that, they can accurately isolate that LPAR that is responsible for the processing bottleneck. By periodically reporting the CPU consumption of each LPAR on a pSeries server, this test facilitates such an analysis. Besides turning the spot light on resource-hungry LPARs, the test also reveals where such LPARs spend maximum time - whether on user-level processing, system-level processing, by being idle, or by being in waiting. In addition, you can optionally configure this test to report the number of hypervisor calls and the time spent by the LPAR's processor on these calls so that, the CPU-efficiency of such calls can also be determined.

Target of the test : An IBM pSeries server

Agent deploying the test : A remote agent

Outputs of the test : One set of results for each LPAR configured on an IBM pSeries server being monitored.

Configurable parameters for the test
Parameter Description

Test period

How often should the test be executed

Host

The IP address of the host for which this test is to be configured.

Port

Indicate the port at which the specified Host listens. By default, this is NULL.

Real ServerName

Specify the name of the pSeries server in the Real ServerName text box. If the target pSeries server has been auto-discovered using HMC, the server name will be set automatically in the Real ServerName text box. However, while configuring this test for a pSeries server that has been manually added, you have to explicitly provide the server name in the Real ServerName text box.

Note:

To obtain the real server name, a user can login to the target pSeries server as a valid pSeries user, go to the shell prompt of the server, and execute the following command: lssyscfg - r sys -F name

Is Managed By

By selecting an option from the Is Managed By list, indicate whether the target pSeries server is managed using an HMC server or an IVM (Integrated Virtual Manager) server. If the target server has been auto-discovered via an HMC server, the HMC option will be automatically chosen from this list.

Management Server, Management User, Management Password

This test connects to an HMC/IVM server to perform LPAR discovery and to collect host-level and "outside view" metrics from the pSeries server. To enable this communication, first, provide the IP address/host name of the HMC/IVM server in the Management Server text box. If the eG manager had automatically discovered the target pSeries server by connecting to an HMC server in the environment, then, the IP address/host name and user credentials pertaining to that HMC server will be automatically displayed in the Management Server, Management User, and Management Password text boxes.

However, if the pSeries server being monitored was manually added to the eG Enterprise system (and not auto-discovered via the HMC server), then, you will have to explicitly indicate whether the target pSeries server is managed by an HMC server or an IVM server by selecting an option from the Is Managed By list. If the HMC option is chosen, then, you will have to provide the IP address of the HMC server that manages the target pSeries server in the Management Server text box. In such a case, in the Management User and Management Password text boxes, you will have to provide the credentials of an HMC user who is assigned the hmcviewer role.

On the other hand, if the IVM option is chosen from the Is Managed By list, it implies that the IP address/host name and user credentials pertaining to that IVM server has to be explicitly provided in the Management Server, Management User, and Management Password text boxes.

Confirm Password

Confirm the HMC Password by retyping it here

Domain

Set the Domain parameter to none.

Admin User, Admin Password, Confirm Password

The eG agent remotely communicates with each discovered LPAR on the pSeries server to obtain their "inside view". For this, the eG agent will have to be configured with the credentials of a valid user with access rights to each LPAR. If a single user is authorized to access all the LPARs on the pSeries server, provide the name and password of the user in the Admin User and Admin Passsword text boxes, and confirm the password by retyping it in the Confirm Password text box. On the other hand, if the test needs to communicate with different LPARs using different user accounts, then, multiple user names and passwords will have to be provided. To help administrators provide these multiple user details quickly and easily, the eG administrative interface embeds a special configuration page. To know how to use this page, refer to Configuring Users for LPAR Monitoring.

SSL

By default, the HMC/IVM server (as the case may be) is not SSL-enabled. This indicates that by default, the eG agent communicates with the HMC/IVM server using HTTP. Accordingly, the SSL flag is set to No by default. If you configure the HMC/IVM server to use SSL, then make sure that the SSL flag is set to Yes, so that the eG agent communicates with the HMC/IVM server using HTTPS.

Ignore VMs Inside View

Administrators of some high security LPAR environments might not have permissions to internally monitor one/more LPARs. The eG agent can be configured to not obtain the 'inside view' of such ‘inaccessible’ VMs using the Ignore VMs Inside View parameter.  Against this parameter, you can provide a comma-separated list of LPAR names, or LPAR name patterns, for which the inside view need not be obtained. For instance, your Ignore VMs Inside View specification can be: *lp,aixlp*,lin*. Here, the * (asterisk) is used to denote leading and trailing spaces (as the case may be). By default, this parameter is set to none indicating that the eG agent obtains the inside view of all LPARs on a pSeries server by default.

Note:

While performing LPAR discovery, the eG agent will not discover the operating system of the LPARs configured in the Ignore VMs Inside View text box.

Exclude VMs

Administrators of some virtualized environments may not want to monitor some of their less-critical LPARs both from 'outside' and from 'inside'. The eG agent in this case can be configured to completely exclude such LPARs from its monitoring purview. To achieve this, provide a comma-separated list of LPARs to be excluded from monitoring in the Exclude VMs text box. Instead of LPARs, LPAR name patterns can also be provided here in a comma-separated list. For example, your Exclude VMs pecification can be: *lp,aixlp*,lin*. Here, the * (asterisk) is used to denote leading and trailing spaces (as the case may be). By default, this parameter is set to none indicating that the eG agent obtains the inside and outside views of all AIX LPARs on a virtual host by default. By providing a comma-separated list of LPARs/LPAR name patterns in the Exclude VMs text box, you can make sure the eG agent stops collecting 'inside' and 'outside' view metrics for a configured set of LPARs.

UseSudo and Sudocmd

By default, the UseSudo flag is set to No, indicating that the test does not collect hypervisor-related metrics by default. If the flag is set to Yes instead, the test uses the sudo command (by default) to collect the hypervisor metrics. This is why, the Sudocmd parameter is set to sudo by default. To use a command other than the sudo command, specify the command in the Sudocmd text box.

Intervals and Count

By default, the Intervals parameter is set to 1 and the Count parameter is set to 2. This implies that every time this test executes, it will collect metrics twice (because, the default Count is 2), and each set of metrics will be collected at the end of every second (because the default Interval is 1). Also, note that the agent will report only the last set of metrics to the eG manager.

Detailed Diagnosis

To make diagnosis more efficient and accurate, the eG Enterprise suite embeds an optional detailed diagnostic capability. With this capability, the eG agents can be configured to run detailed, more elaborate tests as and when specific problems are detected. To enable the detailed diagnosis capability of this test for a particular server, choose the On option. To disable the capability, click on the Off option.

The option to selectively enable/disable the detailed diagnosis capability will be available only if the following conditions are fulfilled:

  • The eG manager license should allow the detailed diagnosis capability
  • Both the normal and abnormal frequencies configured for the detailed diagnosis measures should not be 0.
Measurements made by the test
Measurement Description Measurement Unit Interpretation

Entitled processing capacity

Indicates the number of processing units this LPAR is entitled to receive - i.e., the number of processing units reseved for usage by this LPAR.

Number

For dedicated partitions, the entitled processing capacity is the number of physical processors.

User level entitled processing capacity used

Indicates the percentage of the entitled processing capacity used by this LPAR while executing at the user level.

Percent

A high value for this measure indicates that the user is executing CPU-intensive processes/applications on the LPAR. Comparing the value of this measure with the Kernel level entitled processing capacity used, Idle entitled processing capacity, and Waiting entitled processing capacity measures will reveal where the LPAR spent the maximum time.

Kernel level entitled processing capacity used

Indicates the percentage of the entitled processing capacity used by this LPAR while executing system-level processes.

Percent

A high value for this measure indicates that system-level processing is CPU-intensive. Comparing the value of this measure with the User level entitled processing capacity used, Idle entitled processing capacity, and Waiting entitled processing capacity measures will reveal where the LPAR spent the maximum time.

Idle entitled processing capacity idle

Indicates the percentage of the entitled processing capacity unused while this partition was idle and did not have any outstanding disk I/O request(s).

Percent

Idle processing capacity in the shared resources pool can be used by any shared LPAR that requires additional processing power. While capped partitions can utilize as much processing capacity from the pool as allowed by their maximum processing capacity configuration, the uncapped partitions can draw beyond this maximum threshold, but not beyond the total idle processing capacity in the pool.

Comparing the value of this measure with the User level entitled processing capacity used, Kernel level entitled processing capacity, and Waiting entitled processing capacity measures will reveal where the LPAR spent the maximum time.

Waiting entitled processing capacity

Indicates the percentage of the entitled processing capacity unused while this partition was idle and had outstanding disk I/O request(s).

Percent

A high value for this measure could indicate a processing bottleneck.

Comparing the value of this measure with the User level entitled processing capacity used, Kernel level entitled processing capacity, and Idle entitled processing capacity measures will reveal where the LPAR spent the maximum time.

Physical processors consumed

Indicates the number of physical processing units, of the total entitled processing capacity, that were consumed by this LPAR.

Number

CPU usage measured in absolute terms (i.e., processing units) helps administrators to effectively assess the resource impact of migrating an LPAR to another server. Note that this measure is displayed only when the partition type is "shared" or "dedicated-donating".

Entitled processors consumed

Indicates the percentage of the entitled capacity consumed.

Percent

Because the time base over which this data is computed can vary, the entitled capacity percentage can sometimes exceed 100%. This excess is noticeable only with small sampling intervals. Note that this measure is displayed only when the partition type is "shared".

Logical processors utilization

Indicates the percentage of logical processor(s) that was utilized by this LPAR while executing at the user and system level.

Percent

A very high value is indicative of excessive logical processor utilization by the LPAR. Note that this measure is displayed only when the partition type is "shared".

Available physical processors in the shared pool

Indicates the number of physical processors available in the shared resource pool that is used by this LPAR.

Number

Note that this measure is displayed only when the partition type is "shared".

Virtual context cpu switches

Indicates the number of virtual context switches that are virtual-processor hardware preemptions.

Number

Note that this measure is displayed only when the partition type is "shared" or "dedicated-donating".

Phantom interruptions received

Indicates the number of phantom (targeted to another shared partition in this pool) interruptions received.

MB

A phantom interrupt is an interrupt targeted to another partition that shares the same physical processor. For example, one partition starts an I/O operation. While the partition is waiting for the I/O to complete, it cedes the physical processor to another partition. The I/O operation completes and the controller sends an interrupt to the requesting processor, but as the interrupted partition running is not the intended destination, the partition says “not for me” and the interrupt is queued by the Power hypervisor. They do not have a a big impact on performance.

Time spent in hypervisor

Indicates the percentage of physical processor capacity spent making hypervisor calls.

Percent

A high value for this measure is indicative of too much time and CPU resources spent on hypervisor calls. Note that this measure will be available only if the HYPER CALLS flag is set to Yes for this test, and the ADMIN USER configured for this test is either a 'root user' or a 'root equivalent user'.

Hypervisor calls executed

Indicates the average number of hypervisor calls that were started.

Number

Note that this measure will be available only if the HYPER CALLS flag is set to Yes for this test, and the ADMIN USER configured for this test is either a 'root user' or a 'root equivalent user'.