Merge pull request #19 from kerthcet/feat/agent

Update docs

README.md

@@ -27,11 +27,11 @@ _Name Story: the inspiration of the name `Manta` is coming from Dota2, called [M
 
 ## Features Overview
 
-- **Preheat Models**: Models could be preloaded to the cluster, or even specified nodes to accelerate the model serving.
-- **Model Caching**: Once models are downloaded, origin access is no longer necessary, but from other node peers.
-- **Plugin Framework**: _Filter_ and _Score_ extension points could be extended with your own logic to pick up the best candidates in the form of plugin.
-- **Model LCM**: Manage the model lifecycles automatically with different configurations.
-- **Memory Management(WIP)**: Specify the maximum reserved memory for use, and GC with LRU algorithm.
+- **Model Preheat**: Models could be preloaded to clusters, to specified nodes to accelerate the model serving.
+- **Model Cache**: Models will be cached after downloading for faster model loading.
+- **Model Lifecycle Management**: Manage the model lifecycle automatically with different policies, like `Retain` or `Delete`.
+- **Plugin Framework**: _Filter_ and _Score_ plugins could be extended to pick up the best candidates.
+- **Memory Management(WIP)**: Manage the reserved memories for caching, together with LRU algorithm for GC.
 
 ## Quick Start
 
@@ -41,15 +41,14 @@ Read the [Installation](./docs//installation.md) for guidance.
 
 ### Preheat Models
 
-A toy sample to preload the `Qwen/Qwen2.5-0.5B-Instruct` model:
+A sample to preload the `Qwen/Qwen2.5-0.5B-Instruct` model:
 
 ```yaml
 apiVersion: manta.io/v1alpha1
 kind: Torrent
 metadata:
   name: torrent-sample
 spec:
-  replicas: 1
   hub:
     repoID: Qwen/Qwen2.5-0.5B-Instruct
 ```
@@ -62,7 +61,6 @@ kind: Torrent
 metadata:
   name: torrent-sample
 spec:
-  replicas: 1
   hub:
     repoID: Qwen/Qwen2.5-0.5B-Instruct
   nodeSelector:
@@ -79,20 +77,13 @@ kind: Torrent
 metadata:
   name: torrent-sample
 spec:
-  replicas: 1
   hub:
     repoID: Qwen/Qwen2.5-0.5B-Instruct
   reclaimPolicy: Delete
 ```
 
 More details refer to the [APIs](https://github.com/InftyAI/Manta/blob/main/api/v1alpha1/torrent_types.go).
 
-## Roadmap
-
-- Support GC policy with LRU algorithm
-- More integrations with serving projects
-- Support file chunking
-
 ## Community
 
 Join us for more discussions:

agent/deploy/daemonset.yaml

@@ -26,7 +26,7 @@ spec:
       containers:
       - name: agent
         # image: inftyai/manta-agent:v0.0.1
-        image: inftyai/test:manta-agent-110811
+        image: inftyai/test:manta-agent-111001
         ports:
         - containerPort: 9090
         resources:

api/v1alpha1/torrent_types.go

@@ -78,9 +78,7 @@ type TorrentSpec struct {
 	// URI *URIProtocol `json:"uri,omitempty"`
 
 	// Replicas represents the replication number of each object.
-	// The real Replicas number could be greater than the desired Replicas.
 	// +kubebuilder:default=1
-	// +kubebuilder:validation:Maximum=99
 	// +optional
 	Replicas *int32 `json:"replicas,omitempty"`
 	// ReclaimPolicy represents how to handle the file replicas when Torrent is deleted.

config/crd/bases/manta.io_torrents.yaml

@@ -96,11 +96,8 @@ spec:
                 type: string
               replicas:
                 default: 1
-                description: |-
-                  Replicas represents the replication number of each object.
-                  The real Replicas number could be greater than the desired Replicas.
+                description: Replicas represents the replication number of each object.
                 format: int32
-                maximum: 99
                 type: integer
             type: object
           status:

config/manager/kustomization.yaml

@@ -5,4 +5,4 @@ kind: Kustomization
 images:
 - name: controller
   newName: inftyai/manta
-  newTag: "110905"
+  newTag: "111001"

config/samples/_v1alpha1_torrent.yaml

@@ -9,7 +9,6 @@ metadata:
     app.kubernetes.io/created-by: controller
   name: torrent-sample
 spec:
-  replicas: 2
   reclaimPolicy: Delete
   # nodeSelector:
   #   zone: zone1