docs: add BinPack, Gang, Priority plugin documentation

config/_default/menus.toml

@@ -13,24 +13,20 @@
   url = "#about"
   weight = 2
 
-
+[[main]]
+  name = "Why Volcano"
+  url = "#required"
+  weight = 3
 
 [[main]]
   name = "Blog"
   url = "#posts"
   weight = 4
 
-
 [[main]]
-  name = "Why Volcano"
-  url = "#required"
-  weight = 3
-
-#[[main]]
-  #name = "Documentation"
-  #url = "docs/"
-  #weight = 6
-
+  name = "Documentation"
+  url = "/docs/"
+  weight = 6
 
 [[main_right]]
   name = "<i class=\"fab fa-github\" style=\"color: #000000; font-size: 1rem; line-height: 1.25\"></i>"
@@ -45,7 +41,6 @@
   weight = 20
 
 [[main_right]]
-
   name = "<img src=\"/img/icon_slack.svg\" alt=\"slack\" style=\" font-size: 1rem; line-height: 1.25; width:20px; height:20px; margin-top: 9px;\">"
   post = ""
   url = "https://cloud-native.slack.com/archives/C011GJDQS0N"
@@ -74,10 +69,10 @@
   identifier = "features"
 
 [[docs]]
-name = "User Guide"
-weight = 5
-identifier = "user-guide"
-url = "/en/docs/user-guide/"
+  name = "User Guide"
+  weight = 5
+  identifier = "user-guide"
+  url = "/en/docs/user-guide/"
 
 [[docs]]
   name = "Ecosystem"
@@ -123,25 +118,40 @@ url = "/en/docs/user-guide/"
   identifier = "features"
 
 [[v1-12-0]]
-  name = "Ecosystem"
+  name = "User Guide"
   weight = 5
-  identifier = "ecosystem"
+  identifier = "user-guide"
+  url = "/en/docs/v1-12-0/user-guide/"
 
+# Plugins section (top-level, collapsible container)
 [[v1-12-0]]
-  name = "Scheduler"
+  name = "Plugins"
   weight = 6
+  identifier = "plugins"
+
+# Scheduler section
+[[v1-12-0]]
+  name = "Scheduler"
+  weight = 7
   identifier = "scheduler"
 
+# Continue with remaining sections
+[[v1-12-0]]
+  name = "Ecosystem"
+  weight = 8
+  identifier = "ecosystem"
+
 [[v1-12-0]]
   name = "CLI"
-  weight = 7
+  weight = 9
   identifier = "cli"
 
 [[v1-12-0]]
   name = "Contribution"
-  weight = 8
+  weight = 10
   identifier = "contribution"
 
+
 # Documentation version v1.11.0
 
 [[v1-11-0]]
@@ -344,4 +354,4 @@ url = "/en/docs/user-guide/"
 [[posts]]
   name = "Customers"
   weight = 10
-  identifier = "customers"
+  identifier = "customers"

content/en/docs/v1-12-0/binpack.md

@@ -0,0 +1,83 @@
++++
+title = "BinPack"
+date = 2024-01-01
+lastmod = 2024-01-01
+draft = false
+toc = true
+type = "docs"
+
+# Add menu entry to sidebar.
+linktitle = "BinPack"
+[menu.v1-12-0]
+  parent = "plugins"
+  weight = 1
++++
+
+
+The **BinPack Plugin** in Volcano aims to fill nodes as efficiently as possible, trying not to leave nodes idle. It scores nodes based on resource utilization and schedules pods to maximize usage, which is especially useful for small jobs or high-concurrency workloads.
+
+## Plugin Details
+
+- **Plugin Name:** `binpack`
+- **Location in source:** `pkg/scheduler/plugins/binpack/binpack.go`
+- **Purpose:** Optimizes node usage by packing pods onto fewer nodes, improving cluster resource utilization.
+
+## Features
+
+1. **Node Scoring**
+   - Scores nodes according to available resources.
+   - Higher scores indicate better utilization of node resources.
+
+2. **Weighted Resource Consideration**
+   - Considers CPU, memory, and other resources with configurable weights.
+   - Different resources can be prioritized in node selection.
+
+3. **Pod Placement**
+   - Fills existing nodes as much as possible before using new nodes.
+   - Reduces fragmentation and improves autoscaling efficiency.
+
+4. **Integration with Scheduler**
+   - Injected as a plugin into Volcano's scheduling process.
+   - Works at the pod node-selection stage.
+
+## Source Overview
+
+```go
+// Example snippet from binpack.go
+func New(arguments framework.Arguments) framework.Plugin {
+    return &binPackPlugin{pluginArguments: arguments}
+}
+```
+The full implementation includes functions for scoring nodes, handling resource weights, and integrating with the scheduling session.
+
+## Example Configuration
+
+To enable the BinPack plugin, add it to the scheduler configuration:
+
+```yaml
+# scheduler configuration
+actions: "enqueue, allocate, preempt"
+tiers:
+- plugins:
+  - name: binpack
+    arguments:
+      binpack.cpu: 5
+      binpack.memory: 1
+```
+### Configuration Parameters
+
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `binpack.cpu` | Weight for CPU resources in scoring | 5 |
+| `binpack.memory` | Weight for memory resources in scoring | 1 |
+| `binpack.gpu` | Weight for GPU resources in scoring | 0 |
+
+## Usage Scenarios
+
+BinPack is particularly useful for:
+
+- **Batch processing jobs** where maximizing node utilization reduces cost
+- **Single query jobs** that run quickly and benefit from efficient packing
+- **High-concurrency services** with many small pods that can be packed onto fewer nodes
+- **Cloud cost optimization** where reducing node count minimizes infrastructure costs
+

content/en/docs/v1-12-0/gang.md

@@ -0,0 +1,84 @@
++++
+title = "Gang"
+date = 2024-01-01
+lastmod = 2024-01-01
+draft = false
+toc = true
+type = "docs"
+
+# Add menu entry to sidebar.
+linktitle = "Gang"
+[menu.v1-12-0]
+  parent = "plugins"
+  weight = 2
++++
+
+
+
+{{<figure library="1" src="gang.png" title="Gang Plugin">}}
+
+
+The **Gang Plugin** in Volcano enforces gang scheduling constraints, ensuring that a group of related tasks (a "gang") is scheduled together. This plugin is responsible for validating jobs, managing preemption, pipelining, and detecting job starvation.
+
+## Plugin Details
+
+- **Plugin Name:** `gang`
+- **Location in source:** `pkg/scheduler/plugins/gang/gang.go`
+- **Purpose:** Ensures gang scheduling constraints and job readiness for tasks and sub-jobs.
+
+## Features
+
+1. **Job Validation**
+   - Validates tasks and sub-jobs for gang scheduling.
+   - Checks minimum required tasks (`MinAvailable`) before scheduling.
+
+2. **Preemption**
+   - Determines which tasks can be preempted without violating gang scheduling constraints.
+   - Tracks ready task counts per job to maintain scheduling guarantees.
+
+3. **Job Ordering**
+   - Orders jobs and sub-jobs based on readiness.
+   - Ready jobs have higher priority over non-ready jobs.
+
+4. **Pipelining**
+   - Allows pipelined scheduling when tasks and sub-jobs meet pipelined conditions.
+
+5. **Starvation Detection**
+   - Detects jobs that cannot be scheduled due to insufficient resources.
+
+## Source Overview
+
+```go
+// Example snippet from gang.go
+func New(arguments framework.Arguments) framework.Plugin {
+    return &gangPlugin{pluginArguments: arguments}
+}
+```
+The full implementation includes functions for `OnSessionOpen`, `OnSessionClose`, job validation, preemption, pipelining, and readiness checks.
+
+## Example Configuration
+
+To enable the Gang plugin, add it to the scheduler configuration:
+
+```yaml
+# scheduler configuration
+    arguments:
+      timeout: 600
+```
+
+### Configuration Parameters
+
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `gang.schedule.timeout` | Timeout in seconds for gang scheduling | 600 |
+| `gang.schedule.retry` | Number of retry attempts for gang scheduling | 5 |
+
+## Usage Scenarios
+
+Gang scheduling is typically used for:
+
+- **Distributed training jobs** where all workers must start together
+- **MPI jobs** that require all processes to be running simultaneously
+- **Batch processing workloads** with task dependencies
+- **Big data analytics** where partial execution is not meaningful
+