The Fujitsu ServerView PRO Pack integrates specific components for discovery, monitoring and PRO tips for Fujitsu PRIMERGY servers.
Read More...
The Performance and Resource Optimization (PRO) utilizes specific alerts from the System Center Operations Manager to initiate recovery and optimization actions in the System Center Virtual Machine Manager (SCVMM). For example when a hardware failure is detected then it would be useful to migrate virtual machines to a faultless server. PRO leverages the management pack infrastructure of Operations Manager. When a PRO monitor in a PRO-enabled management pack detects an opportunity for optimization then a PRO tip is generated. The PRO tip may include a recovery script which runs in the SCVMM. A SCVMM administrator can manually confirm and implement PRO tips or the SCVMM administrator can configure PRO to implement PRO tips automatically.
This product can automatically move running Virtual Machines from one PRIMERGY server to another server inside the same SCVMM host group if it detects a critical problem on the PRIMERGY server. Before you configure your environment for automatic PRO Tip implementation mode read the known hints and restrictions carefully to avoid negative impact of your Virtual Machine infrastructure.
Product supports SCOM 2012 R2 and SCVMM 2012 R2 running on Windows Server 2012 R2 and SCOM 2016 and SCVMM 2016 running on Windows Server 2016
System Center Operations Manager in one of the following versions:
System Center Virtual Machine Manager in one of the following versions:
Fujitsu ServerView Agents for Windows with version 5.10.5 or higher and ServerView RAID with version 5.0.0 or higher must be installed and running before Fujitsu ServerView PRO Pack for MS System Center is installed.
The algorithm to determine the new target host for a VM used in our product differs from the algorithm used on the SCVMM Console.
Our product do not show which server will be selected as new VM target host for a given VM. The customer can only check the result after the migration was done in the Jobs history of SCVMM.
This restriction is considered serious also because the execution of a PRO Tip may affect many running VMs that will be migrated to different target hosts in one PRO tip job and by the fact that our product provides the automatic execution of PRO Tips without intervention of an administrator.
A PRO Tip migrates all running VMs of a given server. The migration of the VM is done sequentially and the new target host for each VM is calculated just before the movement of the single event. Because one PRO Tip processes multiple VMs the new target serves for a single VM is not predictable for the customer.
By design the PRO Tip based migration of a VM can only be migrated to another VM target host that belongs to the same host group as the server where the PRO Tip event was generated.
Complex configuration like remote disaster recovery nodes are not considered
The product only uses the host rating result of the VMM PowerShell cmdlet “Get-SCVMHostRating”. More complex situations like a remote node used for disaster recovery only and is connected with lower bandwidth are not supported. It is up to the customer to configure its environment and the host groups accordingly.
In some situations a PRO Tip will be lost. This was recognized if the customer sets the server into maintenance mode in SCOM. After returning from the maintenance mode the PRO tip is lost and will not be shown again even if the maintenance was not started for this PRO Tip event or the maintenance has not fixed the problem.
PRO Tip related events happening during an interval where the VM host is set into maintenance in SCOM will not be lost. There is no notification about the happened event and still existing problem after the maintenance interval ends on the VM host server.
If a Quick Storage Migration is running at the moment a PRO Tip is (automatically) started this VM involved in the storage migration may not be migrated even if it is normally in running state.
The reason is that the FTS product only considers VMs that are in running state at the start of the PRO Tip implementation. During Quick Storage Migration the VM may be stopped at this moment where the recovery script of our product checks for running VMs and therefore missed. Running migrations are not considered by the product.
If a customer has associations between VMs like, that both VMs have to run on the same VM host server for performance reasons or two VMs shall not run on the same host (e.g. DNS Server and Backup DNS Server) the customer must take care using this product. The product does not take care about such relations and it is up to the customer to configure his environment in a way that ensures that the host rating used by the product does not influence the provided services negatively.
Customers running Hyper-V Replica must be aware that there may be situations after a PRO Tip failover where the replica no longer works as expected.
The product is restricted to the migration of VMs only. Storage migration is not supported even not in case where only a storage problem is detected (like RAID events).
Events from ETERNUS systems are not supported. Fiber Channel related events of the SAN subsystem are also not monitored. The Fiber Channel events are part of the standard Fujitsu Integration Pack but not included in the PRO Tip monitoring.
The support of CX250/270/420 does consider the housing location. Therefore a VM may be migrated to another CPU-module inside the same chassis even if the event like a PSU event relates to all CPU modules inside this chassis.
The host rating used by the product does not take virtual fibre channel inside a VM into account. Take special care about the host groups and the accessibility of FC on the VM-hosts to avoid loss of FC-connectivity by an PRO Tip implementation.
If you configure your PRO Monitor settings in SCVMM Console, the console shows you e.g. that the monitoring and remediation of a monitor is set to state “enabled”.
This new setting will be updated in SCOM management server also but this may delay up to a few minutes before the PRO Monitor become active and an event happened on a managed node will be processed as PRO Tip in SCVMM.
The recovery script used in our product has to find a new VM-host, There is no support to configure the used VMM Placement Option as “Load Balance” or “Consolidate”. The fix default used is “Consolidate”.
The product also discovers PRIMERGY servers configured as Network Load Balanced servers and events will generate PRO Tips with the migration strategy “Consolidate”. If you need the strategy “Load Balance” for your NLB servers you should explicitly exclude these servers from discovery and/or PRO monitoring.
The product does not check whether the target with the best host rating, that will be used for migration of a VM is a Network Load Balanced server. By this
The SCVMM allows specifying properties for clouds. These settings of a given cloud is not taken into account in the host rating used for the determination of a new VM-host.
E.g. If you work with Clouds in SCVMM be aware that your settings like Load Balancers or Port Classifications are not considered during the host rating calculation.
Customer PRO monitor configuration done in SCVMM is overridden periodically every 6 hours. This override enables all PRO monitors and therefore every PRO related event will be shown as PRO Tip in SCVMM even it is not really configured. This wrong configuration is active for a few minutes and than reconfigured automatically to the right settings from SCVMM in the background. The interval of the wrong configuration in the tests was between 6 and 15 minutes. This problem only occurs for disabled monitors.
If you dismiss a PRO Tip the according monitor in SCOM is not reset to healthy. You have to reset the monitor in SCOM manually because this unhealthy state of the monitor will prevent a new event managed by the same monitor to generate a new PRO Tip again.
Remark: In case of an implementation of a PRO Tip the monitor is reset to healthy automatically always.
The PRO monitors control a set of events. If a monitor already generated a PRO Tip that is not started or finished in SCVMM a further event (same or different event) monitored by the same monitor will not generate a second PRO Tip. An Alert in SCOM Active Alerts view is also not generated as long as this PRO monitor is not reset to healty.
Microsoft recommends that the PRO related monitors shall be disabled after import of the PRO related Management Pack. Due to the used design of the FTS MP the monitors are enabled after import. You should also import the provided override Management Pack during installation of the product to configure the recommended default behavior of disabled PRO related monitors.
If you select a PRO Alert in the SCVMM GUI window called “PRO” the text in the “Cause and Resolution” sections is shown is a PRO Tip specific text. This is a fix text. If you override the default recommended PRO Tip action in SCOM this static text does no longer reflect the chanced behavior that relates to the made override.
SCVMM supports customer properties that can be specified as “Must match”, “Must not match”, “Should match” and “Should not match”. Only the two “must” rules “Must match” and “Must not match” are considered in the host rating.
The customer can configure the number of concurrently moveable VMs. This parameter is ignored by the recovery script of our product.
If two or more PRO Tips for a given host group occur in short time these PRO Tips will be started simultaneously by SCVMM but the recovery script of our product will serialize the migration in a form that at any given time only one VM will be migrated. This restriction belongs at host group level and can increase the total time the migrations needs.
If a PRO Tip is configured for automatic implementation (remediation) and this PRO Tip implementation failed, this failure will not be reported and the customer may miss this failed migration.
There is no view in SCVMM or SCOM that allows detecting this problem. Neither this PRO Tip is shown as failed in the PRO window nor is the failure reported in the Job history view.
Also in SCOM the customer cannot detect it because the critical health state is reset at the end of the PRO Tip implementation.
The only possibility to detect this is a view into the Windows Event log for entries of the recovery script. But the occurrence of a PRO related event is not predictable and therefore this is considered as a not very practicable workaround.
The product considers VMs as “black box” and evaluates VM-host based parameters only with a few exceptions like the “Possible Owner” property.
Because the product is not aware of relations between two or more VMs and is also not aware of high-available configurations inside one or more VMs we recommend to exclude the product from such more advanced VM configurations. Examples that may be damaged by the product if not excluded are:
The user manual recommends for some problems that the customer shall exclude the VM from PRO management. This is done e.g. for VMs containing an SQL Cluster. You should be aware that this recommendation also stops all PRO enabled management of other PRO Packs from Microsoft or third party vendors.
Example: The Microsoft Management Pack “Virtual Machine Manager PRO V2 HyperV Host Performance” which is installed by default in SC2012 provides PRO base management. Excluding a VM from PRO may also interfere PRO Tips of this Management Pack as side effect.
The user manual recommends for some problems that the customer shall exclude the VM from PRO management. This recommendation is an exclusion of VMs from any PRO based management. This exclusion will result in an environment where PRO management is available for some VMs but excluded for other VMs even if they run on the same single VM-host. This partly availability is considered as difficult to manage and may not reflect customers desire. This restriction became even stronger because the recommended exclusion focus on complex VM configurations (like the restriction that two VMs must always reside on the same VM-host) and VMs designed for high availability (Cluster inside VMs or DNS server and Backup DNS Server that should never run on the same VM-host).
There may be situations where a VM in running state became unavailable by the PRO Tip remediation of the product. During the quality assurance tests this was seen in situations where the new VM-target host becomes unavailable during the migration of a VM. The repair mechanism of the recovery script is not able to restore this VM on the original or another host if the VM-target host is unavailable. The VM can be repaired and started again only manually by the user, after the VM-target is available again.
The health state of the PRO monitors does not show the current health state of the related component, even not if you press “recalculate health” in SCOM. Our product uses an own FTS specific interpretation of these health states. The health state has to be interpreted in a form like:
A warning or critical health state of an FTS PRO monitor indicates that a warning or critical event for this component has happened in the past and the related PRO Monitor is not closed in SCVMM yet.
As an example: If the customer views a FAN failure and the related PRO Tip and he replace this FAN immediately because it is a Customer Self-service Event the monitor will still show critical health-state as long as the PRO Tip is not implemented or dismissed in SCVMM.
SCVMM 2012 SP1 supports for standalone servers the migration option “Live VM and Storage migration (Live VSM)” and “Live migration” where only the VM is migrated.
If both options are available the recovery script always takes the Live VSM migration. This may conflict to the requirement where the “fastest” migration option should be selected always.
For the customer the difference may be seen in longer interval to complete the migration but for the user this difference is not visible because both options are live migrations with a minimal interruption of the offered service from this migrated VM anyway.
In situations where a fast evacuation is indicated this may result in a “lose of service” for the VM that could not be moved by this unnecessary delay.
The migration of a VM can result as ok, warning or error. A migration ends with warning in general can be started and often works without limitations because the warning also show restrictions in redundancy like “multiple I/O not configured”.
The recovery script does not analyze the reason of the warning and processes a warning equal to a failed migration. By this, VMs where the movement ends with a warning will be repaired and moved to another VM host of the host group. If no other VM-host is available as new target the VM will be restarted on the original the faulty VM host again.
One example shows this restriction is the setting “Enable virtual switch optimizations”. A customer may set this option to get optimal support. If necessary he can also migrate the VM to a VM-host without this optimization.
In this case the VM runs without this feature. By our product the systems without optimization will not be possible as new target even not in critical situations. As a consequence a VM may not be evacuated by this restriction.
There may be some issues together with maintenance mode and the reboot of the systems. This complexity was not tested. Therefore the reason of this restriction is that it is an untested configuration.
The product only migrate VMs that are in state “running” while a PRO Tip implementation starts. The SCVMM feature to migrate also stopped VMs to bring them to a healthy system also is not supported. It was the explicit requirement to migrate only running VMs.
The product only migrate VMs that are in state “running” while a PRO Tip implementation starts. The SCVMM feature to migrate also VMs in “saved state” to bring them to a healthy system also is not supported. It was the explicit requirement to migrate only running VMs.
On HyperV you can configure more than one network switch. If your VM using a network configured for “internal only” this condition is not checked during migration.
The behavior is equal to the SCVMM behavior on a manual migration. To avoid breaking your configuration ensure that all hosts in a host group have the same network name configured for the “internal only” network you are using in two or more of your VMs.
On HyperV each virtual switch configured as external connection type has a name and a mapping to a physical network card.
If different serves in your host group use the same name for the virtual switch but have this switch mapped to different network adapters that are configured for different subnets a migration may end successful but the communication of the VM is damaged.
The behavior is equal to the SCVMM behavior on a manual migration. To avoid breaking your configurations please ensure that all hosts in a host group have the same name for a virtual switch and that is always mapped to adapters for the same subnet.
In SCOM there are views for “Windows Server State” and “Windows Computer State”. While problems detected by the Fujitsu Basic integration Pack will also change the health state in this views the PRO monitor states are not included in this health state evaluation.
To get the PRO related health state of a PRIMERGY server please look into the Monitoring View “Fujitsu PRIMERGY PRO Monitoring ? PRO Monitored Servers”.
For manual initiated VM migration SCVMM allows to influence the host rating for proper new VM host target by specifying expected parameters like CPU usage or network usage.
This feature is not supported and by this the host rating of the product is solely based on the values reflect past performance data of the VM during the time it runs on the current (faulty) host as it is evaluated by default in SCVMM.
If you close an PRO Alert in SCOM the PRO Tip will be removed from SCVMM PRO window also. The PRO Monitor is not reset to healthy and prevents the detection of same events in the future.
You have to manually reset the related PRO monitor health state to healthy.
During a manual started migration of VMs the SCVMM provides the option to place each VHD to an individually assigned directory.
The recovery script does not support this feature and will store all VHDs of a VM to the first available virtual machine Path of the new VM target host.
The FTS PRO Tip implementation will use the fastest available migration type by default. This fastest one is calculated by the SCVMM host rating algorithm. But there is no specific view to determine the used migration type before a PRO Tip implementation is started. Therefore be aware that you may consider a downtime of your service that is longer as you would expect or is acceptable by pressing the iImplement button for the PRO Tip.
While SCVMM allows migrating a VM from a server outside a fail over cluster to cluster node this feature is not supported by the FTS product.
Windows Core installation are excluded by PM decision and therefore not tested. There is no special mechanism included to prevent installation of the product on Windows core systems. It may work but it is a not tested feature and therefore not supported.
Virtual machines can be created by SCVMM features but also directly inside a local HyperV-manager or inside the Fail Over Cluster Manager on a managed node.
Because SCVMM does not support all VM configurations available on HyperV or Fail Over Cluster Manager we restrict full support only to VMs that are completely managed by SCVMM.
VMs with unsupported configurations are included in the SCVMM Console views with SCVMM status andVM state “Running” but they are not migrated in case of a PRO Tip. The PRO Tip will end with result OK in such cases even if there are one or more of the above mentioned VMs are in running state on the faulty server.
Due to a Microsoft Framework problem, sometimes the connection between Microsoft System Center Operation Manager (SCOM) and Microsoft System Center Virtual Machine Manager (SCVMM) can be interrupted. In this case new settings for the PRO Monitor Configuration are not taken over and a PRO Tip do not appear also. To renew the connection between Microsoft System Center Operation Manager (SCOM) and Microsoft System Center Virtual Machine Manager (SCVMM) a manually refresh of the connection is necessary.
After PRO Pack installation the Parameter MigrationMaximumVMs is not included
in the config file by default. In this case a simultaneous migration is not
possible. The Parameter must be added manually in the SVIVMM-PROLog.ini file,
which is located in C:\Windows\Temp\sviscom\pro\.
The functions of this product are released for System Center configurations with “homogeneous host groups”.
To avoid problems with this product the customer should make sure that the product feature (PRO Tip) is only enabled inside host groups configured as “homogeneous host groups”.
“Homogeneous host groups” are defined as:
For Failover cluster the host group is configured in a way that the storage configuration needed for a VM is also available on the best SCVMM host-rated cluster node. This configuration must span all used storage connection of cluster nodes. Only by SCVMM supported storage connections are allowed.
The host group configuration warranties that the level of virtualization or cloud configuration does not result in a host rating where the best rated VM-host cannot run the VM properly.
If Microsoft Power Optimization is used together with PRO Pack the customer should ensure that a PRO Tip can migrate all VMs to new target hosts within PRO Tips default behavior. (Special delay for the boot time of a server is not part of PRO Pack).
Installation of the Integration Pack in a non default destination folder does not work correctly.
If a customer changes the destination folder during the installation (MSI installer), only the euly_en.pdf and readme_en.pdf is being copied into the new destination folder.
All other MP files are still being copied into the default destination path.
(C:\Program Files\Fujitsu\ServerView Suite\SCOM Integrations\..)
The ServerView Library Management Pack is used and referenced by all Fujitsu ServerView Integration Packs for System Center Operations Manager.
If a ServerView Integration Pack contains a newer version of the ServerView Library Management Pack this new Fujitsu ServerView Library MP can usually be imported into SCOM without impacts to any other Fujitsu ServerView Integration Management Packs.
In the rare case that a new version of the ServerView Library Management Pack is not compatible with the old version, it is necessary to uninstall all Fujitsu Management Packs including their Override Management Packs and reinstall all Fujitsu Management Packs from the folder ‘Management Packs’ together with the updated ServerView Library Management Pack.
The configuration dialog for Recovery script of PRO Pack 7.0 looks like:
You have the possibility to change the following settings:
The section VM migration order allows controlling the migration order of the VMs on a VM host:
The section Cluster related properties contains settings related to failover clusters.
| Management Pack | Version |
|---|---|
| Fujitsu.PRIMERGY.PRO | 7.0.2.0 |
| Fujitsu.PRIMERGY.PRO.Default.Override | 7.0.2.0 |
Home