This is a new file for Dynamic Cluster configuration parameters.
Example:
<ResourceGroupConf>
<HypervisorResGrps>
<ResourceGroup>
<Name>KVMRedHat_Hosts</Name>
<Template>rhel55hv</Template>
<MembersAreAlsoPhysicalHosts>Yes</MembersAreAlsoPhysicalHosts>
</ResourceGroup>
</HypervisorResGrps>
</ResourceGroupConf> The following parameters are configured in the <ResourceGroupConf> section of the file.
Yes
Defines a host as a hypervisor which can also run physical machine workload on idle resources.
Not defined.
KVMRedHat_Hosts
The name of the resource group to which this hypervisors of the given template belong. The only valid value is KVMRedHat_Hosts.
Not defined.
template_name
The name of the template associated with the hypervisor resource group. This must match a template name defined in the Templates section.
Not defined.
Most parameters are configured as shown:
<ParametersConf>
<Parameter name="PARAMETER_1">
<Value>value_1</Value>
</Parameter>
<Parameter name="PARAMETER_2">
<Value>value_2</Value>
</Parameter>
</ParametersConf>The following parameters are configured in the <ParametersConf> section of the file.
seconds
Time to keep provision requests in memory for query purposes after they are completed. Before this period is completed, the requests appear in the output of the bdc action command. After this period expires, they only appear in the output of bdc hist. Specify the value in seconds.
1800
admin_user_name::directory_path
A connection string to specify the Platform Cluster Manager Advanced Edition administrator account name and installation directory
Admin::/opt/platform
The Platform Cluster Manager Advanced Edition administrator is Admin and the installation directory is /opt/platform.
integer
Maximum number of Dynamic Cluster event log files (dc.events) to keep.
DC_EVENT_FILES_MIN_SWITCH_PERIOD
10
seconds
The minimum elapsed time period before archiving the Dynamic Cluster event log. Define the value in seconds.
Works together with DC_PROVISION_REQUESTS_MAX_FINISHED to control how frequently Dynamic Cluster archives the file dc.events. The number of finished requests in the file is evaluated regularly, at the interval defined by this parameter. The file is archived if the number of requests has reached the threshold defined by DC_PROVISION_REQUESTS_MAX_FINISHED.
The event log file names are switched when a new file is archived. The new file is named dc.events, the archived file is named dc.events.1, and the previous dc.events.1 is renamed dc.events.2, and so on.
DC_PROVISION_REQUESTS_MAX_FINISHED
1800
seconds
The maximum amount of time that a VM can be down during a live migration. This is the amount of time from when the VM is stopped on the source hypervisor and started on the target hypervisor. If the live migration cannot guarantee this down time, the system will continue to retry the live migration until it can guarantee this maximum down time (or the DC_LIVEMIG_MAX_EXEC_TIME parameter value is reached). Specify the value in seconds, or specify 0 to use the hypervisor default for the down time.
0
seconds
The maximum amount of time that the system can attempt a live migration. If the live migration cannot guarantee the down time (as specified by the DC_LIVEMIG_MAX_DOWN_TIME parameter) within this amount of time, the live migration fails. Specify the value in seconds from 1 to 2147483646.
2147483646
minutes
Limits the lifetime of a dynamically created virtual machine. After the specified time period has elapsed since a VM’s creation, if the VM becomes idle, the system automatically powers off and destroy the VM.
This parameter is useful when propagating updates to a shared template. If a shared template is updated, all VM instances which were generated from this template will still run with the old template even if they were powered off.
Setting this parameter to a finite value will cause VMs to be uninstalled (deleted from disk) after the specified period and completing running workload. Any further requests for the shared template must install a new VM, which will then be based on the new version of the template. Therefore administrators can be sure that the template update has been propagated throughout the system after the specified time period.
Not defined. Infinite time.
minutes
Minimum time to live of the Dynamic Cluster host or virtual machine before it can be reprovisioned. This parameter is used to prevent system resources from being reprovisioned too often and generating unnecessary load on the infrastructure. For example, if the value is set to 60, any freshly provisioned machine will not be reprovisioned in less than 60 minutes.
0
integer
Number of finished provisioning requests in the Dynamic Cluster event log before it is archived. Works together with DC_EVENT_FILES_MIN_SWITCH_PERIOD to control how frequently Dynamic Cluster archives the file dc.events. The number of jobs in the file is evaluated regularly, at the interval defined by DC_EVENT_FILES_MIN_SWITCH_PERIOD. The file is archived if the number of jobs has reached or exceeded the threshold defined by DC_PROVISION_REQUESTS_MAX_FINISHED.
The event log file names are switched when a new file is archived. The new file is named dc.events, the archived file is named dc.events.1, the previous dc.events.1 is renamed dc.events.2, and so on.
DC_EVENT_FILES_MIN_SWITCH_PERIOD
5000
seconds
After each job finishes, allow a grace period before the machine can accept another provision request. Specify the value in seconds.
By default, when a job completes, the machine it was running on becomes eligible for reprovisioning. However, some jobs have post-execution processes that may be interrupted if the host is reprovisioned too quickly. This parameter configures a grace period after job termination during which the host cannot be reprovisioned, which gives these processes a chance to complete.
0
minutes
Time period to uninstall (delete from storage) a dynamically created VM in the off state. Specify the value in minutes.
A virtual machine can be created to meet peak workload demands. However, after peak loads pass, those virtual machines will be powered off and stored in storage. Those dynamic virtual machines can be configured to be deleted if they remain off for a long time.
1440
integer
Multiple values allowed.
This parameter wraps each value in <memsize> instead of <Value>. For example:
<memsize> integer </memsize>
Specify one or more choices for VM memory size. Specify the value in MB. Separate values in a list with space.
The memory size of any new VM is the smallest of all the choices that satisfy the job’s resource requirement.
For example, if a job requires 500 MB memory, and this parameter is set to "1024 1536", the VM is created with 1024 MB memory. If a job requires 1025 MB memory, a VM is created with 1536 MB memory.
Using this feature helps prevent the hypervisor hosts from being fragmented with multiple VMs of different size. When the VMs have standardized memory size, they can easily be reused for jobs with similar memory requirements.
Dynamic Cluster only.
If this parameter is used, DC_VM_MEMSIZE_STEP in lsb.params is ignored.
512 minimum
Not defined
string
Prefix for naming a new VM that is created by Dynamic Cluster. Specify a text string.
You can specify more than one name using multiple <Value/> entries, but only the first value is used for dynamically creating new VMs. However, all VMs named with any of these prefixes will be treated as dynamically created VMs even if they were manually created. They will be subject to DC_VM_UNINSTALL_PERIOD, DC_MACHINE_MAX_LIFETIME.
vm_lsf_dyn
resource_group_name
Specify names of Dynamic Cluster resource groups which are allowed to create new VMs. The only valid value is KVMRedHat_Hosts.
directory_path
Dynamic Cluster working directory. This is the location where the dc.events file will be stored.
/opt/lsf/work/cluster_name/dc
For more information, see the setup instructions.