Files
sglang/sgl-router/src/core/job_queue.rs

591 lines
21 KiB
Rust

//! Async job queue for control plane operations
//!
//! Provides non-blocking worker management by queuing operations and processing
//! them asynchronously in background worker tasks.
use std::{
collections::HashMap,
sync::{Arc, Weak},
time::{Duration, SystemTime},
};
use dashmap::DashMap;
use tokio::sync::mpsc;
use tracing::{debug, error, info, warn};
use crate::{
config::{RouterConfig, RoutingMode},
core::workflow::{
steps::WorkerRemovalRequest, WorkflowContext, WorkflowEngine, WorkflowId,
WorkflowInstanceId, WorkflowStatus,
},
metrics::RouterMetrics,
protocols::worker_spec::{JobStatus, WorkerConfigRequest},
server::AppContext,
};
/// Job types for control plane operations
#[derive(Debug, Clone)]
pub enum Job {
AddWorker { config: Box<WorkerConfigRequest> },
RemoveWorker { url: String },
InitializeWorkersFromConfig { router_config: Box<RouterConfig> },
}
impl Job {
/// Get job type as string for logging
pub fn job_type(&self) -> &str {
match self {
Job::AddWorker { .. } => "AddWorker",
Job::RemoveWorker { .. } => "RemoveWorker",
Job::InitializeWorkersFromConfig { .. } => "InitializeWorkersFromConfig",
}
}
/// Get worker URL for logging
pub fn worker_url(&self) -> &str {
match self {
Job::AddWorker { config } => &config.url,
Job::RemoveWorker { url } => url,
Job::InitializeWorkersFromConfig { .. } => "startup",
}
}
}
impl JobStatus {
fn pending(job_type: &str, worker_url: &str) -> Self {
Self {
job_type: job_type.to_string(),
worker_url: worker_url.to_string(),
status: "pending".to_string(),
message: None,
timestamp: SystemTime::now()
.duration_since(SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs(),
}
}
fn processing(job_type: &str, worker_url: &str) -> Self {
Self {
job_type: job_type.to_string(),
worker_url: worker_url.to_string(),
status: "processing".to_string(),
message: None,
timestamp: SystemTime::now()
.duration_since(SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs(),
}
}
fn failed(job_type: &str, worker_url: &str, error: String) -> Self {
Self {
job_type: job_type.to_string(),
worker_url: worker_url.to_string(),
status: "failed".to_string(),
message: Some(error),
timestamp: SystemTime::now()
.duration_since(SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs(),
}
}
}
/// Job queue configuration
#[derive(Clone, Debug)]
pub struct JobQueueConfig {
/// Maximum pending jobs in queue
pub queue_capacity: usize,
/// Number of worker tasks processing jobs
pub worker_count: usize,
}
impl Default for JobQueueConfig {
fn default() -> Self {
Self {
queue_capacity: 1000,
worker_count: 10,
}
}
}
/// Job queue manager for worker validation and removal operations
pub struct JobQueue {
/// Channel for submitting jobs
tx: mpsc::Sender<Job>,
/// Weak reference to AppContext to avoid circular dependencies
context: Weak<AppContext>,
/// Job status tracking by worker URL
status_map: Arc<DashMap<String, JobStatus>>,
}
impl std::fmt::Debug for JobQueue {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("JobQueue")
.field("status_count", &self.status_map.len())
.finish()
}
}
impl JobQueue {
/// Create a new job queue with background workers (spawns tasks)
///
/// Takes a Weak reference to AppContext to avoid circular strong references.
/// Spawns background worker tasks that will process jobs asynchronously.
pub fn new(config: JobQueueConfig, context: Weak<AppContext>) -> Arc<Self> {
let (tx, rx) = mpsc::channel(config.queue_capacity);
debug!(
"Initializing worker job queue: capacity={}, workers={}",
config.queue_capacity, config.worker_count
);
let rx = Arc::new(tokio::sync::Mutex::new(rx));
let status_map = Arc::new(DashMap::new());
let queue = Arc::new(Self {
tx,
context: context.clone(),
status_map: status_map.clone(),
});
for i in 0..config.worker_count {
let rx = Arc::clone(&rx);
let context = context.clone();
let status_map = status_map.clone();
tokio::spawn(async move {
Self::worker_loop(i, rx, context, status_map).await;
});
}
// Spawn cleanup task for old job statuses (TTL 5 minutes)
let cleanup_status_map = status_map.clone();
tokio::spawn(async move {
Self::cleanup_old_statuses(cleanup_status_map).await;
});
queue
}
/// Submit a job
pub async fn submit(&self, job: Job) -> Result<(), String> {
// Check if context is still alive before accepting jobs
if self.context.upgrade().is_none() {
RouterMetrics::record_job_shutdown_rejected();
return Err("Job queue shutting down: AppContext dropped".to_string());
}
// Extract values before moving job
let job_type = job.job_type().to_string();
let worker_url = job.worker_url().to_string();
// Record pending status
self.status_map.insert(
worker_url.clone(),
JobStatus::pending(&job_type, &worker_url),
);
match self.tx.send(job).await {
Ok(_) => {
let queue_depth = self.tx.max_capacity() - self.tx.capacity();
RouterMetrics::set_job_queue_depth(queue_depth);
debug!(
"Job submitted: type={}, worker={}, queue_depth={}",
job_type, worker_url, queue_depth
);
Ok(())
}
Err(_) => {
RouterMetrics::record_job_queue_full();
self.status_map.remove(&worker_url);
Err("Worker job queue full".to_string())
}
}
}
/// Get job status by worker URL
pub fn get_status(&self, worker_url: &str) -> Option<JobStatus> {
self.status_map.get(worker_url).map(|entry| entry.clone())
}
/// Remove job status (called when worker is deleted)
pub fn remove_status(&self, worker_url: &str) {
self.status_map.remove(worker_url);
}
/// Worker loop that processes jobs
async fn worker_loop(
worker_id: usize,
rx: Arc<tokio::sync::Mutex<mpsc::Receiver<Job>>>,
context: Weak<AppContext>,
status_map: Arc<DashMap<String, JobStatus>>,
) {
debug!("Worker job queue worker {} started", worker_id);
loop {
// Lock the receiver and try to receive a job
let job = {
let mut rx_guard = rx.lock().await;
rx_guard.recv().await
};
match job {
Some(job) => {
let job_type = job.job_type().to_string();
let worker_url = job.worker_url().to_string();
let start = std::time::Instant::now();
// Update status to processing
status_map.insert(
worker_url.clone(),
JobStatus::processing(&job_type, &worker_url),
);
debug!(
"Worker {} processing job: type={}, worker={}",
worker_id, job_type, worker_url
);
// Upgrade weak reference to process job
match context.upgrade() {
Some(ctx) => {
let result = Self::execute_job(&job, &ctx).await;
let duration = start.elapsed();
Self::record_job_completion(
&job_type,
&worker_url,
worker_id,
duration,
&result,
&status_map,
);
}
None => {
let error_msg = "AppContext dropped".to_string();
status_map.insert(
worker_url.clone(),
JobStatus::failed(&job_type, &worker_url, error_msg),
);
error!(
"Worker {}: AppContext dropped, cannot process job: type={}, worker={}",
worker_id, job_type, worker_url
);
break;
}
}
}
None => {
warn!(
"Worker job queue worker {} channel closed, stopping",
worker_id
);
break;
}
}
}
debug!("Worker job queue worker {} stopped", worker_id);
}
/// Execute a specific job
async fn execute_job(job: &Job, context: &Arc<AppContext>) -> Result<String, String> {
match job {
Job::AddWorker { config } => {
let engine = context
.workflow_engine
.get()
.ok_or_else(|| "Workflow engine not initialized".to_string())?;
let instance_id = Self::start_worker_workflow(engine, config, context).await?;
debug!(
"Started worker registration workflow for {} (instance: {})",
config.url, instance_id
);
let timeout_duration =
Duration::from_secs(context.router_config.worker_startup_timeout_secs + 30);
Self::wait_for_workflow_completion(
engine,
instance_id,
&config.url,
timeout_duration,
)
.await
}
Job::RemoveWorker { url } => {
let engine = context
.workflow_engine
.get()
.ok_or_else(|| "Workflow engine not initialized".to_string())?;
let instance_id = Self::start_worker_removal_workflow(engine, url, context).await?;
debug!(
"Started worker removal workflow for {} (instance: {})",
url, instance_id
);
let timeout_duration = Duration::from_secs(30);
let result =
Self::wait_for_workflow_completion(engine, instance_id, url, timeout_duration)
.await;
// Clean up job status when removing worker
if let Some(queue) = context.worker_job_queue.get() {
queue.remove_status(url);
}
result
}
Job::InitializeWorkersFromConfig { router_config } => {
let api_key = router_config.api_key.clone();
let mut worker_count = 0;
// Create iterator of (url, worker_type, bootstrap_port) tuples based on mode
let workers: Vec<(String, &str, Option<u16>)> = match &router_config.mode {
RoutingMode::Regular { worker_urls } => worker_urls
.iter()
.map(|url| (url.clone(), "regular", None))
.collect(),
RoutingMode::PrefillDecode {
prefill_urls,
decode_urls,
..
} => {
let prefill_workers = prefill_urls
.iter()
.map(|(url, port)| (url.clone(), "prefill", *port));
let decode_workers =
decode_urls.iter().map(|url| (url.clone(), "decode", None));
prefill_workers.chain(decode_workers).collect()
}
RoutingMode::OpenAI { .. } => {
info!("OpenAI mode: no workers to initialize");
return Ok("OpenAI mode: no workers to initialize".to_string());
}
};
debug!(
"Creating AddWorker jobs for {} workers from config",
workers.len()
);
// Process all workers with unified loop
for (url, worker_type, bootstrap_port) in workers {
let url_for_error = url.clone(); // Clone for error message
let config = WorkerConfigRequest {
url,
api_key: api_key.clone(),
worker_type: Some(worker_type.to_string()),
labels: HashMap::new(),
model_id: None,
priority: None,
cost: None,
tokenizer_path: None,
reasoning_parser: None,
tool_parser: None,
chat_template: None,
bootstrap_port,
health_check_timeout_secs: router_config.health_check.timeout_secs,
health_check_interval_secs: router_config.health_check.check_interval_secs,
health_success_threshold: router_config.health_check.success_threshold,
health_failure_threshold: router_config.health_check.failure_threshold,
max_connection_attempts: router_config.health_check.success_threshold * 10,
dp_aware: router_config.dp_aware,
};
let job = Job::AddWorker {
config: Box::new(config),
};
if let Some(queue) = context.worker_job_queue.get() {
queue.submit(job).await.map_err(|e| {
format!(
"Failed to submit AddWorker job for {} worker {}: {}",
worker_type, url_for_error, e
)
})?;
worker_count += 1;
} else {
return Err("JobQueue not available".to_string());
}
}
Ok(format!("Submitted {} AddWorker jobs", worker_count))
}
}
}
/// Start a workflow and return its instance ID
async fn start_worker_workflow(
engine: &Arc<WorkflowEngine>,
config: &WorkerConfigRequest,
context: &Arc<AppContext>,
) -> Result<WorkflowInstanceId, String> {
let mut workflow_context = WorkflowContext::new(WorkflowInstanceId::new());
workflow_context.set("worker_config", config.clone());
workflow_context.set_arc("app_context", Arc::clone(context));
engine
.start_workflow(WorkflowId::new("worker_registration"), workflow_context)
.await
.map_err(|e| format!("Failed to start worker registration workflow: {:?}", e))
}
/// Start worker removal workflow
async fn start_worker_removal_workflow(
engine: &Arc<WorkflowEngine>,
url: &str,
context: &Arc<AppContext>,
) -> Result<WorkflowInstanceId, String> {
let removal_request = WorkerRemovalRequest {
url: url.to_string(),
dp_aware: context.router_config.dp_aware,
};
let mut workflow_context = WorkflowContext::new(WorkflowInstanceId::new());
workflow_context.set("removal_request", removal_request);
workflow_context.set_arc("app_context", Arc::clone(context));
engine
.start_workflow(WorkflowId::new("worker_removal"), workflow_context)
.await
.map_err(|e| format!("Failed to start worker removal workflow: {:?}", e))
}
/// Wait for workflow completion with adaptive polling
async fn wait_for_workflow_completion(
engine: &Arc<WorkflowEngine>,
instance_id: WorkflowInstanceId,
worker_url: &str,
timeout_duration: Duration,
) -> Result<String, String> {
let start = std::time::Instant::now();
let mut poll_interval = Duration::from_millis(100);
let max_poll_interval = Duration::from_millis(2000);
let poll_backoff = Duration::from_millis(200);
loop {
// Check timeout
if start.elapsed() > timeout_duration {
return Err(format!(
"Workflow timeout after {}s for worker {}",
timeout_duration.as_secs(),
worker_url
));
}
// Get workflow status
let state = engine
.get_status(instance_id)
.map_err(|e| format!("Failed to get workflow status: {:?}", e))?;
let result = match state.status {
WorkflowStatus::Completed => Ok(format!(
"Worker {} registered and activated successfully via workflow",
worker_url
)),
WorkflowStatus::Failed => {
let current_step = state.current_step.as_ref();
let step_name = current_step
.map(|s| s.to_string())
.unwrap_or_else(|| "unknown".to_string());
let error_msg = current_step
.and_then(|step_id| state.step_states.get(step_id))
.and_then(|s| s.last_error.as_deref())
.unwrap_or("Unknown error");
Err(format!(
"Workflow failed at step {}: {}",
step_name, error_msg
))
}
WorkflowStatus::Cancelled => {
Err(format!("Workflow cancelled for worker {}", worker_url))
}
WorkflowStatus::Pending | WorkflowStatus::Paused | WorkflowStatus::Running => {
tokio::time::sleep(poll_interval).await;
poll_interval = (poll_interval + poll_backoff).min(max_poll_interval);
continue;
}
};
// Clean up terminal workflow states
engine.state_store().cleanup_if_terminal(instance_id);
return result;
}
}
/// Record job completion metrics and update status
fn record_job_completion(
job_type: &str,
worker_url: &str,
worker_id: usize,
duration: Duration,
result: &Result<String, String>,
status_map: &Arc<DashMap<String, JobStatus>>,
) {
RouterMetrics::record_job_duration(job_type, duration);
match result {
Ok(message) => {
RouterMetrics::record_job_success(job_type);
status_map.remove(worker_url);
debug!(
"Worker {} completed job: type={}, worker={}, duration={:.3}s, result={}",
worker_id,
job_type,
worker_url,
duration.as_secs_f64(),
message
);
}
Err(error) => {
RouterMetrics::record_job_failure(job_type);
status_map.insert(
worker_url.to_string(),
JobStatus::failed(job_type, worker_url, error.clone()),
);
warn!(
"Worker {} failed job: type={}, worker={}, duration={:.3}s, error={}",
worker_id,
job_type,
worker_url,
duration.as_secs_f64(),
error
);
}
}
}
/// Cleanup old job statuses (TTL 5 minutes)
async fn cleanup_old_statuses(status_map: Arc<DashMap<String, JobStatus>>) {
const CLEANUP_INTERVAL: Duration = Duration::from_secs(60); // Run every minute
const STATUS_TTL: u64 = 300; // 5 minutes in seconds
loop {
tokio::time::sleep(CLEANUP_INTERVAL).await;
let now = SystemTime::now()
.duration_since(SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs();
// Remove statuses older than TTL
status_map.retain(|_key, value| now - value.timestamp < STATUS_TTL);
debug!(
"Cleaned up old job statuses, remaining: {}",
status_map.len()
);
}
}
}