 |
» |
|
|
|
Limitations Headroom ratings are not
available for scenario targets. The capreport command does not support complexes or scenarios as targets. The Scenario Comparison
report does not support network I/O and disk metrics. The option that allows
you to choose whether to display data in percentages, absolute numbers,
or both is not available on the Scenario Comparison report. PMP data cannot be imported
from a virtual machine. You cannot edit more than
one scenario at a time. On a Windows CMS, the capreport command can only be executed by users who are
members of the Administrators group. Automatic IPM power calibration
on the Edit Power Settings pages requires that IPM and iLO2 Advanced
or iLO2 Select be properly licensed, configured, and discovered. There
is a time delay of up to a day before power and utilization data is
collected from iLO2 by IPM for use by Capacity Advisor. This data
collection can be started manually from the IPM user interface. Configuration
of IPM requires Administrator privileges.
Major Issues Capacity Advisor Extension
Pack licenses available with VSE Management Software Version 4.0 will
incorrectly have the value of “Free Trial” in the License Source field in the HP SIM License Manager graphical
interface, even though they have been purchased. If the Profile
Viewer or Scenario Editor displays
no data for a node (“Error: No data has been collected for the
specified workload”), and this occurs even after a successful
collection from that node (that is, data collection completes without
errors), then restarting Systems Insight Manager or rebooting the
CMS may correct the issue. Reporting algorithms are
not optimized to minimize CPU and memory consumption on the CMS. Workaround. For best results and minimal impact to system
performance, limit the number of systems included in any one report.
For example, with 512 MB maximum heap size and 2 GB RAM on the CMS
and a year of data, run the report on no more than 50 managed nodes
at a time. The cappmpextract command (command-line interface for HP PMP data import) reports
one physical CPU for machines that actually have dual-core processors,
as compared to data provided by the Utilization Provider, which would
report two CPU cores. Workaround. Do not include PMP and Utilization Provider data
within the same report. If necessary to do so, mark the PMP data as
invalid to obtain a more useful analysis. There is a substantial
difference in data reported by PMP agents and Utilization Provider
agents when one logical NIC is composed of two physical networks.
This difference affects the usefulness of network utilization reports
that contain both types of data. If you encounter the following
error message in a Capacity Advisor data collection output: ip-address: The system "hostname.hp.com" has no workload
defined. Make sure to select Tools->VSE Management... in HP-SIM
before running this command for the first time.
|
and the stated fix fails to correct the problem, you must
ensure that the DNS lookup returns the host name of the system rather
than its IP. For the IPM power history
of a managed node, the Refresh Data button
may appear not to work and a javascript error may be displayed in
the lower left corner of the page. When this happens, usually the
power data is collected. However, the graph will not update. To redraw
the graph, click Refresh Data and then the Draw Graph button after the Refresh Data operation has completed. The capovpaextract command does not generate a profile when running on a Solaris 10
system with two processors. Reports generated by the capreport command may be blocked by licensing constraints. The following issues exist
with reports generated from the Capacity Advisor user interface: It is possible to select both the “Scenario
Comparison” and “Power” report checkboxes. Running
such a report causes an exception to occur. On Utilization reports,
the table “Time spent at or above each percent of allocation” shows the wrong values, such as “0.00%”. When a report is based on a scenario, only the systems
in that scenario are available in the report. The Power report shows values that are too large. The Scenario Comparison report shows “N/A” in the power metrics.
Minor Issues If users import OVPA data
into Capacity Advisor, using capovpaextract, from
a managed node that was upgraded from an OVPA release prior to version
4.0 to a release later than 4.0, then the OVPA data of the release
prior to version 4.0 is kept. The imported
data (only memory) corresponding to the OVPA release prior to version
4.0 is incorrect. Workaround. You can use either of the following workarounds
for this situation: Import OVPA data in the timeframe of the new OVPA
agent. Import all data and after that invalidate it using
a capprofile command such as the following: capprofile –m –i –b begin_time –e end_time profileID
Capacity Advisor Extension
Pack licenses will not permit applying more than one license to a
single system after the initial term expires. If this is necessary,
contact your HP Support Representative. After a V2P migration
or after importing data (using either the OVPA or PMP import tools,
or using the capprofile command directly), it is
best to wait twenty-four hours before collecting data on the host,
in order to prevent potential profile data loss for the host in Capacity
Advisor. If Capacity Advisor data
collection succeeds for VMware ESX Guests, but no new data is displayed
for the guest in the Capacity Advisor Profile Viewer (that is, the “Data exists between”dates/times do not change), try restarting
the the VMM agent (Configure Virtual Machine ManagementRestart VMM Agent...) on the VMware ESX Host. After
restarting the VMM agent re-run the data collection for the guest. Help links are not working
for the capreport command when the CMS is localized
to Japanese. This only occurs for the Japanese Windows CMS and the
English Windows CMS localized to Japanese. The JS and CSS directories are not created when running
the capreport command. This only occurs for the
English Windows CMS localized to Japanese. Parked/Reassigned Workload Shows Incorrect Utilization. If you select a workload
in a scenario and choose What-if ActionPark Workloads... then later
select the same workload and choose What-if ActionMove Workloads... and move
the parked workload to a system, when you return to the System tab, the utilization shown for the system containing
the workload will be incorrect as shown. If you click on a utilization
bar to view the profile for the system, the correct utilization is
shown. Workaround. After parking and reassigning a workload, close
the scenario and reopen it. The correct utilization will be shown
for all systems and you can continue to edit the scenario. Monitored
workload definitions only support ASCII text. All fields in a monitored workload must be specified as ASCII (not
Unicode) text except for the description, which can contain non-ASCII
characters. The Smart Solver may produce
incorrect results if one or more workloads have data that is invalid
across the entire simulation interval for metrics that have utilization
limits specified. This issue affects Automated System Consolidation
to VMs, Automated Workload Stacking, Automated Load Balance of Servers, and Automated Load Balance of VMs. Workaround. Remove the workload(s) with the invalid data from
the list of workloads to be used for the Smart Solver simulation. If users attempt to move
workloads whose data is invalid across the entire simulation interval
for metrics that have utilization limits specified, then the systems
that the workloads are targeted to move to will display incorrect
headroom rating values. Workaround. Remove the workload(s) with the invalid data from
the list of workloads to be moved. If there are one or more
systems inside a scenario that are unauthorized or unidentifiable,
then reports can not be created for that scenario. In this case, the
following message is displayed: User does not have authorization for the scenario: <scenario name>. |
Workaround. To create a report for a scenario in this state,
remove the unauthorized or unidentifiable system in the scenario editor
and recreate this report. If an error occurs on
Step 2 of Wizard pages, the date that was set up by the user changes
to the default date. Under rare circumstances,
multiple errors may occur when collecting data from a very large set
of mixed physical and virtual systems. This can happen when using
the Collect Capacity Advisor Data... menu or when
using the capcollect command. The errors are reported
in the Capacity Advisor collection output (for example, a large number
of “Specified Vm Guest or Host [hostname] does not
exist within VMM” errors may be reported). Workaround. To avoid these errors, try collecting from all
physical systems and all virtual systems (VM guests and their hosts)
in two separate passes. Create new nightly collection tasks for collecting
from physical and virtual systems separately. These new collection
tasks replace the default “Collect ALL Capacity Advisor Data
Nightly” scheduled task. For information on creating nightly collection
tasks, see the Capacity Advisor User's Guide “Scheduling a Data Collection” section. If
you open the Capacity Advisor “Edit Scenario” page and
see this message: Error generating screen. Attempting to create second instance of
(Scenario_systemLayout_nameHeader_system_name_popupContent*) |
this means that the HP SIM identification of a VM Host is inconsistent
with the identification of the virtual machines associated with it. Workaround. You will need to re-identify the VM Host and its
virtual machines to correct this issue. You may then want to click
the “Refresh Data” link on the Virtualization Manager Workload tab to ensure the information is also up-to-date
for Capacity Advisor. If this problem persists, you may need to run vseassist on the VM Host to determine why HP SIM is not
correctly identifying the VM Host and its constituents. For VMware hosts/guests,
Capacity Advisor requires use of VMM 3.5 and assumes that the VMM
service is running on default port 40420. During VMM installation,
make sure to accept the default port settings. The Edit Power Settings
action will not work on unlicensed systems. On HP-UX 11i v1 with Perl
version 5.8.0 and kernel stack size parameter set to 128 MB, capovpaextract fails with the message “Out of Memory”. The following issues exist
with reports: Utilization data values are shown in seconds. Forecast date information is not shown in Utilization
reports. With command-line reports, in a configuration file
with two entries for the same default report name, only the first
report name is generated. This problem occurs only on a Windows CMS.
capcollect issues: Data collection from VMware and Microsoft Virtual
Server (MSVS) hosts and guests requires HP Virtual Machine Management
Pack 3.5 (VMM). VMware and MSVS hosts must be properly configured
and licensed within VMM, otherwise collection from VMware and MSVS
hosts and guests will fail. To verify proper installation and configuration
of VMM, consult the VMM user guide. A good indicator of whether VMM
is configured properly is to look at the main HP SIM display, and
locate the “VM status” column. Make sure there is a check
mark in a green circle, indicating normal status. For a VMware or MSVS guest, host names must be resolvable
from the CMS. If DNS is not correctly configured, the following error
message is displayed: The system "xxxxx" has no workload defined.
Make sure to select Tools->VSE Management... in HP-SIM
before running this command for the first time. |
Also, re-attempting the collection at a later time may
automatically correct the issue.
|
Printable version
