Merge pull request #2086 from nebius/SCHED-845/increase-default-reconfiguration-period

itechdima · web-flow · commit e2d63b05f4f4 · 2026-01-26T16:05:11.000+01:00
increase default reconfiguration period
diff --git a/api/v1/slurmcluster_types.go b/api/v1/slurmcluster_types.go
@@ -343,7 +343,7 @@ type SConfigController struct {
 	// ReconfigureWaitTimeout defines the maximum time to wait for all nodes to restart during reconfiguration.
 	// Must be greater than ReconfigurePollInterval. Defaults to 1m
 	// +kubebuilder:validation:Optional
-	// +kubebuilder:default="1m"
+	// +kubebuilder:default="5m"
 	ReconfigureWaitTimeout *string `json:"reconfigureWaitTimeout,omitempty"`
 
 	// HostUsers controls if the pod containers can use the host user namespace
diff --git a/config/crd/bases/slurm.nebius.ai_slurmclusters.yaml b/config/crd/bases/slurm.nebius.ai_slurmclusters.yaml
@@ -3373,7 +3373,7 @@ spec:
                       Defaults to 20s
                     type: string
                   reconfigureWaitTimeout:
-                    default: 1m
+                    default: 5m
                     description: |-
                       ReconfigureWaitTimeout defines the maximum time to wait for all nodes to restart during reconfiguration.
                       Must be greater than ReconfigurePollInterval. Defaults to 1m
diff --git a/helm/slurm-cluster/values.yaml b/helm/slurm-cluster/values.yaml
@@ -525,7 +525,7 @@ sConfigController:
   runAsUid: 1001
   runAsGid: 1001
   reconfigurePollInterval: "20s"
-  reconfigureWaitTimeout: "1m"
+  reconfigureWaitTimeout: "5m"
   serviceMonitor:
     enabled: true
     jobLabel: "sconfigcontroller"
diff --git a/helm/soperator-crds/templates/slurmcluster-crd.yaml b/helm/soperator-crds/templates/slurmcluster-crd.yaml
@@ -29175,7 +29175,7 @@ spec:
                       Defaults to 20s
                     type: string
                   reconfigureWaitTimeout:
-                    default: 1m
+                    default: 5m
                     description: |-
                       ReconfigureWaitTimeout defines the maximum time to wait for all nodes to restart during reconfiguration.
                       Must be greater than ReconfigurePollInterval. Defaults to 1m
diff --git a/helm/soperator/crds/slurmcluster-crd.yaml b/helm/soperator/crds/slurmcluster-crd.yaml
@@ -29175,7 +29175,7 @@ spec:
                       Defaults to 20s
                     type: string
                   reconfigureWaitTimeout:
-                    default: 1m
+                    default: 5m
                     description: |-
                       ReconfigureWaitTimeout defines the maximum time to wait for all nodes to restart during reconfiguration.
                       Must be greater than ReconfigurePollInterval. Defaults to 1m
diff --git a/internal/controller/sconfigcontroller/jailedconfig_controller.go b/internal/controller/sconfigcontroller/jailedconfig_controller.go
@@ -53,7 +53,7 @@ import (
 const (
 	configMapField = ".spec.configMap.name"
 
-	defaultReconfigureWaitTimeout  = 1 * time.Minute
+	defaultReconfigureWaitTimeout  = 5 * time.Minute
 	defaultReconfigurePollInterval = 20 * time.Second
 )
 

Original file line number	Diff line number	Diff line change
`@@ -53,7 +53,7 @@ import (`
`53`	`53`	`const (`
`54`	`54`	`configMapField = ".spec.configMap.name"`
`55`	`55`
`56`		`- defaultReconfigureWaitTimeout = 1 * time.Minute`
	`56`	`+ defaultReconfigureWaitTimeout = 5 * time.Minute`
`57`	`57`	`defaultReconfigurePollInterval = 20 * time.Second`
`58`	`58`	`)`
`59`	`59`