Converse agora com nosso suporte
Chat com o suporte

Foglight for VMware 6.1.0 - Release Notes

Known issues

Known issues

This 6.1.0 release of Foglight for VMWare includes the following known issues at the time of this release.

Defect ID

Known Issue

VM-6572

There are no VMware tags services displayed on the Foglight Management Server with FIPS-compliant mode enabled.

VFOG-11752

The Virtual Machines Network Performance - Service report is missing units when exported as an Microsoft Excel spreadsheet.

VFOG-11219

Too many vCenter sessions are detected, due to the agents not disconnecting properly when done. This leaves many idle sessions on the vCenter.

VFOG-10849

When forwarded, the vSphere Alarm Virtual machine CPU usage displays null values.

VFOG-10796

 vCenter alarms are not collected by default. To enable their collection, set the Alarms Collection Enabled property of the VMware Performance Agent to True.

VFOG-10217

A number of Warning messages are found in the log, indicating a big time difference between the monitored vCenter and Management Server.

VFOG-10171

Incorrect virtual CPU ratio values are found in custom reports.

VFOG-9390

The table on the Top CPU Consuming Virtual Machines view cannot be sorted.

VFOG-9067

VMware cartridge creates duplicate VMware Performance agents with the ID -1.

VFOG-9042
VFOG-6727

The vCenter Certificate's public key length has to be 1024-bits or higher and could not use MD5. This is because vCenters are configured to use a 512-bit long certificate keys. The Foglight Agent Manager JRE is updated to version 1.7.0.40 (1.7u40) or higher, which restricts the certificate key length to 1024 bits or higher.

Workaround:

  1. Locate the following entry in the <fglam_home>/jre/lib/security/java.security file: 

    jdk.certpath.disabledAlgorithms=MD2, RSA keySize < 1024

  2. Modify it as follows: 

    jdk.certpath.disabledAlgorithms=MD2, RSA keySize < 512

  3. Deactivate and activate the VMware Performance Agent.

VFOG-6302

The IP address of ESX host cannot be collected and presented in the dashboard.

VFOG-8973

If there are multiple virtual machines with the same name running in different virtual centers, their performance data may be incorrect.

VFOG-8515

Upgrading to Foglight for Virtualization, Enterprise Edition 8.0 causes the Virtual Environment Overview tiles to show zero objects.

Workaround: Update the VMware Performance agent on the VMware Environment Administration tab, as instructed. For upgrade instructions, see Upgrade and Compatibility.

VFOG-3421

When the IP or host name of the monitoring host changes, this results in two Virtual Center instances, one that collects data, and another that does not.
Workaround: Using the Data Management dashboard, navigate to VMware Model > Virtual Centers, and delete the Virtual Center host that no longer collects data.

VFOG-2198

After exporting the CPU Utilization graph to Excel, incorrect labels appear in the resulting file.

N/A

For large virtual infrastructure environments, the performance of the browser interface can be affected by the amount of memory allocated to the Java Virtual Machine (JVM) on the Management Server.

Management Server dashboards

Defect ID

Known Issue

VFOG-9887

On the All Hosts dashboard, accessible by clicking Homes > Hosts in the browser interface, Disk Current and Network Current columns for any ESX hosts or virtual machines show "n/a" instead of the actual metric values. This behavior does not apply to physical hosts collecting OS metric data. This table shows disk and network utilization percentages for ESX hosts or virtual machines which are not collected by  Foglight for VMware. Hovering over these columns shows the related disk and network throughputs, which appear in units other than percentages.

Workaround: There is no workaround available for this issue.

 


Third party known issues

Third party known issues

The following is an issue known to exist at the time of this release.

Defect ID

Known Issue

VFOG-11632

Performance metrics cannot be collected for VSAN 5.5 because of its limited API.
For more information, visit: http://pubs.vmware.com/vsphere-60/index.jsp?topic=%2Fcom.vmware.wssdk.apiref.doc%2Fvim.host.VsanInternalSystem.html

Workaround: Upgrade to VSAN version 6.0.

VFOG-11599

A known issue in vCenter 5.1, 5.5 and 6.0 leads to wrong packet loss values, causing the VMW Virtual Machine Packet Loss rule alarms to be triggered. For more information, visit:
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2052917

Workaround: If you monitor version 5.1/5.5, install the latest patch. At the moment, there is no solution for version 6.0.

VFOG-10311

If a monitored VM is using a vmxnet3 adapter, networking data cannot be collected. This is a known issue in vCenter 6. For more information, visit:
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2119264

VFOG-10301

The VM Network Performance data was lost when moving a virtual machine between vCenters.

VFOG-9632

The VMware Explorer displays logical disk information for RHEL 6 virtual machines, but not for RHEL 7 virtual machines. This is due to the limitation of the default open-vm-tools version 9.4.0, included with RHEL 7 and CentOS 7, which does not support the XFS file system.

VFOG-8579

vSwitch topology views display incorrect utilization and packet loss values, showing data for unused and standby NICs.
Workaround: Since NICs are connected physically, they receive broadcast messages, which results in the display of data transmission rates rate for these NICs. However there is no actual loss of data transmit. To prevent this data from appearing, physically remove such NICs.

VFOG-8486

The total command latency is collected for monitored datastores, but cannot be collected for NFS (network file system) storage.
Workaround: The total command latency is based on disk-based calculations, and is only available for a disk/LUN that uses SCSI commands. Not all types of datastores are disk-based (NFS for example is not), and sometimes a disk can create multiple datastores. As an alternative, the datastore total latency metric can be used.

VFOG-8455

The virtual machine guest OS boot time is only collected in vCenter 5 or higher.

VFOG-8278

Cluster CPU utilization is currently calculated using Hypervisor CPU usage. vCenter calculates the same values using only the available memory for virtual machines.

VFOG-5139
VFOG-5438

Collection of traffic type utilization metrics is only possible if vCenter network usage counters are available and configured properly. These metrics illustrate how different types of data traffic are represented in the overall network traffic managed by a virtual switch. In a default installation network usage counters (listed below) are set to 3. To enable the collection of traffic type utilization metrics, each of these counters must be set to 1:

throughput.usage.vm
throughput.usage.vmotion
throughput.usage.ft
throughput.usage.iscsi
throughput.usage.nfs
throughput.usage.hbr

On ESXi version 4.1 network usage counters are not available, so this feature is not supported.
ESXi versions 5.0 and 5.1 includes network usage counters, and they can be configured as required.
The counters were removed from ESXi versions 5.5, 5.5.0a, and 5.5.0b, so traffic type utilization metrics are not collected.

For more information, see the following article: http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2083590.

 


Upgrade and compatibility

Upgrade and compatibility

Foglight Evolve 6.1.x include a new version of Foglight for VMware 6.1.x. This new version of Foglight for VMware provides new capabilities to Foglight Evolve.

Foglight for VMware supports VMware vSphere and VMware vCenter Server versions 5.x and 6.x.

Important: The VMware Performance Agent uses a Virtual Center account to log in to the monitored Virtual Center. The administrative role associated with that account must be granted the Performance > Modify intervals privilege. When collecting traffic-type network metrics from the monitored ESX hosts, this privilege enables the VMware Performance Agent to change the performance metric counter levels in order to collect this data.

 

vCenters in Linked Mode: Linked Mode is supported, however an agent monitoring one vCenter does not gather data from other linked vCenters. Each vCenter requires its own dedicated agent.

The following version of the Cisco Nexus 1000V Switch for VMware vSphere is supported:

  • On vSphere 5.x, version 4.2(1)SV2(2.2) is supported. At the time of this release, this is the latest version of the Cisco Nexus 1000V Switch supporting vSphere 5.0, 5.1, and 5.5.

Note: The extended support for VirtualCenter 2.5 ended on May 21, 2013. Foglight for VMware is no longer tested with or supported for VirtualCenter 2.5. 

Foglight for VMware is not compatible with the VirtualCenter for VMware Server.

Important: If you install or upgrade to this version of Foglight for VMware on the Management Server (outside of the Foglight Evolve or Foglight for Storage Management product families), and you also have the Foglight for Storage Management cartridge installed on the server, you must ensure that the Foglight for Storage Management cartridge is at least version 3.1.4. Using an earlier version will prevent the Storage Management data collections and dashboards from working as expected.

The following Foglight product versions and platforms are compatible with this release.

Product Name

Product Version

Platform

Foglight Management Server

6.1.0

All platforms supported by this version of the Foglight Management Server

Foglight Agent Manager

6.1.0

All platforms supported by this version of the Foglight Agent Manager

Foglight Evolve

6.1.0

All platforms supported by this version of the Foglight Evolve

To upgrade to the latest version of Foglight for VMware:
  1. Deactivate all VMware Performance agent instances.
  2. On the Cartridge Inventory dashboard, install the following cartridge files. Do not remove the older version of these cartridges. Install version 6.0.0 over the older version.
    • vUsage-Feedback-6_1_0.car
    • DRP-6_1_0.car
    • Virtual-VMware-6_1_0.car
    • Virtual-VMware-Admin-6_1_0.car
  3. On the VMware Environment dashboard, on the Administration tab, update all of the existing VMware Performance agent instances.
  4. Important. If you increased the JVM memory allocation in a previous cartridge installation (as described in System Requirements), you need to repeat this process after deploying the new agent package and before activating the agents.
  5. Activate all VMware Performance agent instances.

Note: Extra minutes will be needed if you upgrade to the latest version in the environment where is deployed with more than 4,000 virtual machines.

For more information, see the Foglight Upgrade Guide.

 


System requirements

System requirements

Before installing Foglight for VMware, ensure your system meets the following minimum hardware and software requirements:

Operating System

Foglight for VMware can be installed on any supported Foglight Management Server configuration.

Platform, Memory and Hard Disk Space The platform, memory, and hard disk space requirements for the monitored servers must match the Agent Manager requirements. For complete information, see the System Requirements and Platform Support Guide.

Note: 32-bit Windows and Linux Agent Manager installers are not included in the Foglight Evolve installer. New installations should use 64-bit Agent Managers. If you are upgrading 32-bit agent hosts, you can obtain 32-bit Agent Manager Installers from Quest Support.

For a complete list of system requirements, refer to the Installing the Virtualization Management Capabilities guide.

VMware Performance Agent and Agent Manager configuration

On 64-bit hosts meeting the minimum system requirements, the embedded Agent Manager can be used to run VMware Performance Agents to monitor up to1000 virtual machines. If the total number of virtual machines to be monitored from a single agent host is greater than 4000, an Agent Manager should be installed on a separate host.

Important: Foglight Evolve Virtual Appliance comes pre-configured to support up to 4,000 virtual machines. If you are using this product, there is no need to follow the configuration procedure described in this section.

If additional cartridges and agents are added to the environment, product performance should be monitored and agents moved off of the embedded Agent Manager to reduce the load.

Foglight VMware Performance Agent host system requirements

  Minimum CPU Minimum Memory Minimum Host Memory Total Monitored VMs
Windows 64-bit

2*

2 GB

4 GB

1000

Linux 64-bit

2*

2 GB

4 GB

1000

* Additional CPUs may be required for larger environments.

Monitoring more than 4000 virtual machines from a single agent host

The Agent Manager JVM usually requires additional memory to monitor more than 4000 virtual machines. The total number of virtual machines is the total from all vCenters that will be monitored from all VMware Performance agents running on the Agent Manager.

The following calculations are guidelines, not hard and fast rules. Memory requirements can vary greatly from installation to installation with similar VM counts. If insufficient memory is configured, the failure mode is easily recognizable: all agents on the Agent Manager host will go into a broken state after the agent(s) were activated for a short period of time, usually within 24 hours. In addition, the Agent Manager log will contain a line similar to the following:

Caused by: java.lang.OutOfMemoryError: Java heap space

If this is the case, add memory greater than what is shown in the calculations below, in increments of 512 MB until the agents stabilize.

JVM memory requirements for VMware Performance agents are calculated using the following formula:

512 MB + 0.5 MB/VM

According to the above formula, monitoring 4000 virtual machines requires 2512 MB of memory:

512 MB + 2000 MB = 2512 MB (round up to 3 GB)

The default setting for agents deployed on 64-bit systems is 2560 MB, which suffice this environment.

Similarly, monitoring 8192 virtual machines requires 4608 MB of memory:

512 MB + 4096 MB = 4608 MB

Foglight Evolve Virtual Appliance comes pre-configured to support up to 4,000 virtual machines. This requires the default Agent Manager settings to be changed.

To change the JVM memory settings:
  1. Determine the amount of additional memory required. This will be the total from the last step above minus the default value of 2560 MB. In the example above, this is 4608 – 2560 = 2048 MB.
  2. On the agent machine, open the baseline.jvmargs.config file for editing. The file is located in the <Agent_Manager_home>/state/default/config directory.
  3. Add the following lines in the Memory Settings section:

    vmparameter.0 = "-Xms2048m";
    vmparameter.1 = "-Xmx2048m";

    Note: If this file has been previously edited, increment the numeric parameters accordingly.
  4. Delete the existing deployed negotiation configuration settings directory:

    <Agent_Manager_home>/state/default/config/deployments
  5. Restart the Agent Manager for these settings to take effect.
Host system recommendations
Dedication

When monitoring larger vCenters, the Agent Manager machine hosting the VMware Performance agents should be dedicated to this task. No other Foglight agent types should run on the host and the host should not run any other applications.

Memory

Regardless of the values set for JVM memory above, Agent Manager never allocates more than 80% of system memory. So the machine hosting the Agent Manager and VMware Performance Agents must have sufficient memory. Beyond the memory requirements of the Agent Manager, a minimum of 2 GB should be free for the operating system. In the example above for 8000 VMs and an Agent Manager memory requirement of 4608 MB, the host should have a minimum of 6656 MB - the greater of:

1.25 x 4608 MB = 5760 MB

Or:

2048 + 4608 MB = 6656 MB

If the Foglight Agent Manager is configured on a virtual machine, it is recommended that the VM use a memory reservation to ensure maximum performance.

CPU

CPU usage on the Agent Manager host is relatively low most of the time. However, usage peaks dramatically during the performance metric collection. This is normal and expected. CPU utilization consistently over 50% is an indication that additional processing power is required. As with memory, usage can vary between different installations with similar numbers of virtual machines. The following guidelines should be followed:

Up to 1000 VMs

2 CPUs

1000 – 4000 VMs

4 CPUs

4000+ VMs

add 1 CPU per 1000 VMs – round up when necessary

 

When the Agent Manager and VMware Performance agents are running on a virtual machine, the VM should be configured with CPU reservation whenever possible to ensure best performance. Lack of processing power on the Agent Manager manifests in missed collections and gaps in the data usually noticeable in various graphs throughout the Foglight for VMware dashboards.

Foglight Management Server minimum memory requirements

# of virtual machines Memory*
Foglight JVM settings
1000

Xms|Xmx=4G

2500 Xms|Xmx=9G
5000 Xms|Xmx=17G
10000 Xms|Xmx=32G
20000 Xms|Xmx=62G

* Edit the Foglight Management Server JVM settings in the server.config file located under Quest\Foglight\config\. When running Foglight on a virtual machine, full memory needs be reserved at the hypervisor level.

 


Documentos relacionados

The document was helpful.

Selecione a classificação

I easily found the information I needed.

Selecione a classificação