feat(orch): add ScallignDown status

packages/api/internal/orchestrator/nodemanager/client.go

@@ -20,6 +20,7 @@ var OrchestratorToApiNodeStateMapper = map[orchestratorinfo.ServiceInfoStatus]ap
 	orchestratorinfo.ServiceInfoStatus_Healthy:   api.NodeStatusReady,
 	orchestratorinfo.ServiceInfoStatus_Draining:  api.NodeStatusDraining,
 	orchestratorinfo.ServiceInfoStatus_Unhealthy: api.NodeStatusUnhealthy,
+	orchestratorinfo.ServiceInfoStatus_Standby:   api.NodeStatusStandby,
 }
 
 func NewClient(tracerProvider trace.TracerProvider, meterProvider metric.MeterProvider, host string) (*clusters.GRPCClient, error) {

packages/api/internal/orchestrator/nodemanager/status.go

@@ -16,6 +16,7 @@ var ApiNodeToOrchestratorStateMapper = map[api.NodeStatus]orchestratorinfo.Servi
 	api.NodeStatusReady:     orchestratorinfo.ServiceInfoStatus_Healthy,
 	api.NodeStatusDraining:  orchestratorinfo.ServiceInfoStatus_Draining,
 	api.NodeStatusUnhealthy: orchestratorinfo.ServiceInfoStatus_Unhealthy,
+	api.NodeStatusStandby:   orchestratorinfo.ServiceInfoStatus_Standby,
 }
 
 func (n *Node) Status() api.NodeStatus {

packages/orchestrator/info.proto

@@ -8,8 +8,11 @@ option go_package = "https://github.com/e2b-dev/infra/orchestrator";
 // needs to be different from the enumeration in the template manager
 enum ServiceInfoStatus {
   Healthy = 0;
+  // Draining means the node is bound to be shut down. It will not accept new sandboxes and will stop once all existing sandboxes are done.
   Draining = 1;
   Unhealthy = 2;
+  // Standby means the node is not actively used, but it can return to Healthy and continues serving traffic.
+  Standby = 3;
 }
 
 enum ServiceInfoRole {

packages/orchestrator/internal/healthcheck/healthcheck.go

@@ -39,7 +39,7 @@ func (h *Healthcheck) getStatus() e2bHealth.Status {
 	switch h.info.GetStatus() {
 	case e2borchestratorinfo.ServiceInfoStatus_Healthy:
 		return e2bHealth.Healthy
-	case e2borchestratorinfo.ServiceInfoStatus_Draining:
+	case e2borchestratorinfo.ServiceInfoStatus_Draining, e2borchestratorinfo.ServiceInfoStatus_Standby:
 		return e2bHealth.Draining
 	}
 

packages/orchestrator/main.go

@@ -673,7 +673,7 @@ func run(config cfg.Config) (success bool) {
 	// Mark service draining if not already.
 	// If service stats was previously changed via API, we don't want to override it.
 	logger.L().Info(ctx, "Starting drain phase", zap.Int("sandbox_count", sandboxes.Count()))
-	if serviceInfo.GetStatus() == orchestratorinfo.ServiceInfoStatus_Healthy {
+	if status := serviceInfo.GetStatus(); status == orchestratorinfo.ServiceInfoStatus_Healthy || status == orchestratorinfo.ServiceInfoStatus_Standby {
 		serviceInfo.SetStatus(ctx, orchestratorinfo.ServiceInfoStatus_Draining)
 
 		// Wait for draining state to propagate to all consumers

spec/openapi-edge.yml

@@ -78,6 +78,7 @@ components:
         - healthy
         - draining
         - unhealthy
+        - standby
 
     ClusterOrchestratorRole:
       type: string

spec/openapi.yml

@@ -1294,17 +1294,22 @@ components:
 
     NodeStatus:
       type: string
-      description: Status of the node
+      description: |
+        Status of the node.
+        - draining: the node is bound to be shut down. It will not accept new sandboxes and will stop once all existing sandboxes are done.
+        - standby: the node is not actively used, but it can return to ready and continue serving traffic.
       enum:
         - ready
         - draining
         - connecting
         - unhealthy
+        - standby
       x-enum-varnames:
         - NodeStatusReady
         - NodeStatusDraining
         - NodeStatusConnecting
         - NodeStatusUnhealthy
+        - NodeStatusStandby
 
     NodeStatusChange:
       required: