The Akri Agent executes on all worker Nodes in the cluster. It is primarily tasked with:
- 1.Handling resource availability changes
- 2.Enabling resource sharing
These two tasks enable Akri to find configured resources (leaf devices), expose them to the Kubernetes cluster for workload scheduling, and allow resources to be shared by multiple Nodes.
The first step in handling resource availability is determining what resources (leaf devices) to look for. This is accomplished by finding existing Configurations and watching for changes to them.
For each resource that is found:
- 1.An Instance is created and uploaded to etcd
- 2.A connection with the kubelet is established according to the Kubernetes Device Plugin framework. This connection is used to convey availability changes to the kubelet. The kubelet will, in turn, expose these availability changes to the Kubernetes scheduler.
Each protocol will periodically reassess what resources are visible and update both the Instance and the kubelet with the current availability.
This process allows Akri to dynamically represent resources that appear and disappear.
To enable resource sharing, the Akri Agent creates and updates the
Instance.deviceUsagemap and communicates with kubelet. The
Instance.deviceUsagemap is used to coordinate between Nodes. The kubelet communication allows Akri Agent to communicate any resource availability changes to the Kubernetes scheduler.
The Agent discovers resources via Discovery Handlers (DHs). A Discovery Handler is anything that implements the
DiscoveryHandlerservice defined in
discovery.proto. In order to be utilized, a DH must register with the Agent, which hosts the
Registrationservice defined in
discovery.proto. The Agent maintains a list of registered DHs and their connectivity statuses, which is either
Offline(Instant). When registered, a DH's status is
Waiting. Once a Configuration requesting resources discovered by a DH is applied to the Akri-enabled cluster, the Agent will create a connection with the DH requested in the Configuration and set the status of the DH to
Active. If the Agent is unable to connect or loses a connection with a DH, its status is set to
Instantmarks the time at which the DH became unresponsive. If the DH has been offline for more than 5 minutes, it is removed from the Agent's list of registered Discovery Handlers. If a Configuration is deleted, the Agent drops the connection it made with all DHs for that Configuration and marks the DHs' statuses as
Waiting. Note, while probably not commonplace, the Agent allows for multiple DHs to be registered for the same protocol. IE: you could have two udev DHs running on a node on different sockets.
The Agent's registration service defaults to running on the socket
/var/lib/akri/agent-registration.sockbut can be Configured with Helm. While Discovery Handlers must register with this service over UDS, the Discovery Handler's service can run over UDS or an IP based endpoint.
In addition to the
discoveryDetailsin Configuration that sets details for narrowing the Discovery Handlers' search, the
discoveryPropertiescan be used to pass additional information to Discovery Handler. One of scenarios that can leverage
discoveryPropertiesis to pass credential data to Discovery Handlers to perform authenticated resource discovery. It is common for a device to require authentication in order to access its properties. The Discovery Handler then need these credentials to properly discover and filter the device. The credential data can be placed in
discoverProperties, if it is specified in Configuration, Agent reads the content and generate a list of string key-value pair properties and pass the list to Discovery Handler along with
Agent supports plain text, K8s
configMapin the schema of
discoverProperies. An example below shows how each type of property is specified in
nameof property is required and needs to be in C_IDENTIFIER format. The value can be specified by
valueFrom. For value specified by
valueFrom, it can be from
optionalattribute is default to
false, it means if the data doesn't exist (in the
configMap), the Configuration deployment will fail. If
true, Agent will ignore the entry if the data doesn't exist, and pass all exist properties to Discovery Handler, the Configuration deployment will success.
- name: property_from_plain_text
value: “plain text data”
- name: property_from_secret
- name: property_from_configmap
For the example above, with the content of secret and configMap.
Agent read all properties and pass the string key-value pair list to Discovery Handle.
"property_from_plain_text": “plain text data”