Instance Type Management

Instance type management provides the capabilities of managing a cluster of the computation resources by instances management such as create, delete, edit instances and of permission-control which allows only specified-groups to use the instances.

Creating New Instance Types

Click Add to add an Instance Type, which will pop up the editing screen of Instance Types.

You need to fill in the fields of Basic Info in the above picture:

• Name Only lowercase letters, numbers, hyphen - and a dot . can be filled in.

• Display name The display name of this Instance Type, and will be seen by users.

• Description The description of this Instance Type, and will be seen by users.

• CPU Limit define how many CPU are allowed to use by this Instance Type. The value is also applied to CPU Request when CPU Request is disabled.

• Memory Limit define how many memory are allowed to use by this Instance Type. The value also applied to Memory Request when Memory Request is disabled.

• GPU Limit define how many GPU can be used by this Instance Type. GPU can only be integer.

• GPU Resource Name define which type of GPU is going to be used, you can specify amd.com/gpu or nvidia.com/gpu. The GPU Limit must be larger than 1 to enable this field.

Overcommitting (advanced feature)

• CPU Request define how many CPU are requested to use by this Instance Type initially. Once it is enabled, instances are guaranteed to get the amount of CPU they request. If CPU Request < CPU Limit, the system will try to overcommit CPU resources within the limit if more resources are available.

• Memory Request define how many Memory are requested to use by this Instance Type initially. Once it is enabled, instances are guaranteed to get the amount of Memory they request. If Memory Request < Memory Limit, the system will try to overcommit Memory resources within the limit if more resources are available.

For details, please see Quality of Service for Pods and Resource QoS.

• Global If it is turned on, this Instance Type can be chosen by everyone. You can grant permission to specific groups in the Edit Groups.

Finally, click Confirm to complete the addition.

Toleration

When a node has a taint for certain reasons, an instance won't be scheduled to run on that node until an instance has a specific toleration to tolerate specified taint. Here we will show you how to add a toleration only.

For more detail and usage of kubectl taint, please refer to Taints and Tolerations concept.

In Toleration tab, you can add a Toleration by clicking on Add button.

Filling in the fields in popup to create a toleration.

• Key The key of taint which you want to tolerate.

• Value The value of taint is required if Equal from Operator is selected.

• Operator (Mandatary) Select Equal, Exists.

• Effect Select None, NoSchedule, PreferNoSchedule, NoExecute

Regarding options of Operator and Effect, please refer to Taints and Tolerations.

Click OK to add a toleration.

Finally, click Confirm to complete the addition.

For a use case of toleration, please see Toleration Use Case.

Node Selector

You can constrain an instance to only be able to run on specific nodes which have specific labels. A label is a map of key-value pair. Here we will show you how to add a nodeSelector only.

For more detail, please refer to Assigning Pods to Nodes for usage kubectl label.

In Node Selector tab, you can add a NodeSelector by clicking on + Add field button.

Filling in key/value with key-value of a label you want to specify.

Click Confirm to complete the addition.

For a use case, please see Node Selector Use Case.

Edit Groups

If Global is disabled, please click Edit Groups under the edit Instance Type screen to select the groups that have permission to use the Instance Type.