Documentation
¶
Overview ¶
Package processor provides client functionality for the Agones allocation processor system
The allocation processor system enables batched game server allocation requests This package contains the client implementation that connects to the processor service batches allocation requests, and handles the stream-based communication protocol
Key components: - Client: Manages connection lifecycle and request batching - Config: Configuration for processor client behavior - Batch handling: Accumulates requests and sends them in batches to the processor
The client establishes a bidirectional gRPC stream with the processor service, registers itself, and then handles pull requests (to send batched allocations) and batch responses (containing allocation results)
Flow diagram:
┌─────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐
│ Client │ │ Processor Client │ │ Processor Server │
│ (Allocator/ │ │ (this package) │ │ │
│ Extension) │ │ │ │ │
└─────────────────┘ └─────────────────────┘ └─────────────────────┘
│ │ │
│ │ 1. Connect & Register │
│ │ (bidirectional stream)│
│ ├──────────────────────────►
│ │ │
│ 2. Allocate(request) │ │
├────────────────────────► │
│ │ │
│ │ 3. Add to hotBatch │
│ │ (accumulate) │
│ │ │
│ │ 4. Pull Request │
│ │ (to all clients) │
│ ◄──────────────────────────┤
│ │ 5. Send BatchRequest │
│ │ (hotBatch) │
│ ├──────────────────────────►
│ │ │
│ │ │ 6. Process
│ │ │ allocations
│ │ │
│ │ 7. BatchResponse │
│ │ (results) │
│ ◄──────────────────────────┤
│ 8. Return result │ │
◄────────────────────────┤ │
│ │ │
Note: Multiple Processor Clients can connect to one Processor Server
The server sends pull requests to all connected clients
Legend:
- Client: Makes allocation requests (allocator, extensions, etc.)
- Processor Client: Batches requests and manages communication
- Processor Server: Processes batched allocations from multiple clients
- Pull Request: Server asks all connected clients for pending requests
- BatchRequest: Client sends accumulated allocation requests
- BatchResponse: Server returns allocation results
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Client ¶
type Client interface {
// Run starts the processor client
Run(ctx context.Context) error
// Allocate performs a batch allocation request
Allocate(ctx context.Context, req *allocationpb.AllocationRequest) (*allocationpb.AllocationResponse, error)
}
Client interface for allocation operations Provides methods to run the processor client and perform allocation requests
type Config ¶
type Config struct {
// ClientID is a unique identifier for this processor client instance
ClientID string
// ProcessorAddress specifies the address of the processor service to connect to
ProcessorAddress string
// MaxBatchSize determines the maximum number of allocation requests to batch together
MaxBatchSize int
// AllocationTimeout is the maximum duration to wait for an allocation response
AllocationTimeout time.Duration
// ReconnectInterval is the time to wait before retrying a failed connection
ReconnectInterval time.Duration
}
Config holds the processor client configuration
Source Files