Help Center/ Cloud Container Engine/ User Guide/ Networking/ Services/ LoadBalancer/ Traffic and Performance Optimization Configuration/ Configuring Timeout for a LoadBalancer Service
Updated on 2026-06-16 GMT+08:00
View PDF
Share

Configuring Timeout for a LoadBalancer Service

LoadBalancer Services allow you to configure timeout, which is the maximum duration for keeping a connection if no request is received from the client. If no request is received during this period, the load balancer closes the connection and establishes a new one with the client when the next request arrives.

After timeout is configured, if you delete the timeout configuration on the CCE console or delete the target annotation from the YAML file, the configuration on the ELB will be retained.

Notes and Constraints

The following table lists the scenarios where timeout can be configured for a Service.

Timeout Type

Load Balancer Type

Constraint

Supported Cluster Version

Idle timeout

Dedicated

None
  • v1.19: v1.19.16-r30 or later
  • v1.21: v1.21.10-r10 or later
  • v1.23: v1.23.8-r10 or later
  • v1.25: v1.25.3-r10 or later
  • Other clusters of later versions

Idle timeout

Shared

UDP is not supported.
  • v1.23: v1.23.13-r0 or later
  • v1.25: v1.25.8-r0 or later
  • v1.27: v1.27.5-r0 or later
  • v1.28: v1.28.3-r0 or later
  • Other clusters of later versions

Request timeout

Dedicated and shared

Only HTTP and HTTPS are supported.

Response timeout

Dedicated and shared

Only HTTP and HTTPS are supported.

Creating a LoadBalancer Service and Configuring Timeout Durations

Use one of the following methods.

  1. Log in to the CCE console and click the cluster name to access the cluster console.
  2. In the navigation pane, choose Services & Ingresses. In the upper right corner, click Create Service.

    In this example, only mandatory parameters for configuring SNI are listed. Retain the default settings for other parameters. For details, see Using the CCE Console (New Version).

  3. Configure basic parameters.

    Parameter

    Description

    Example

    Service Type

    Select LoadBalancer.

    None

    Service Name

    Enter a name, which can be the same as the workload name.

    nginx

    Namespace

    Select the namespace that the workload belongs to.

    default

    Selector

    Add the key and value of a pod label. The Service will be associated with the workload pods based on the label and direct traffic to the pods with this label.

    You can also click Reference Workload Label to use the label of an existing workload. In the dialog box displayed, select a workload and click OK.

    app:nginx

  4. Configure load balancer parameters.

    Parameter

    Description

    Example

    Load Balancer
    Select a load balance type and how the load balancer will be created. To enable HTTP/HTTPS on the listener port of a dedicated load balancer, the type of the load balancer must be Application (HTTP/HTTPS) or Network (TCP/UDP/TLS) & Application (HTTP/HTTPS).
    • Use existing: Only the load balancers in the same VPC as the cluster can be selected. If no load balancer is available, click Create Load Balancer to create one on the ELB console.
    • Auto create: The load balancer will be created in the VPC that the cluster belongs to. For details, see Table 1.

    An existing Dedicated load balancer of the Network (TCP/UDP/TLS) & Application (HTTP/HTTPS) type

  5. Configure access parameters.

    Parameter

    Description

    Example

    Service Affinity
    Whether to route external traffic to a local node or a cluster-wide endpoint. For details, see Service Affinity (externalTrafficPolicy).
    • Cluster-level: The IP addresses and ports of all nodes in a cluster can access the workload associated with the Service. However, accessing the Service may result in decreased performance due to route redirection, and the client's source IP address may not be obtainable.
    • Node-level: Only the IP address and port of the node where the workload is located can access the workload associated with the Service. Accessing the Service will not result in a performance decrease due to route redirection, and the client's source IP address can be obtained.

    Cluster-level

    Port
    • Protocol: the protocol used by the Service. According to the Kubernetes implementation, if a Service uses a load balancer with a non-UDP protocol, this parameter must be set to TCP and the corresponding listener frontend protocol must be selected. For details, see Protocols for Services.
    • Container Port: the port that the workload listens on. For example, Nginx uses port 80 by default.
    • Service Port: the port used by the Service.
      • Listen on a port: The port ranges from 1 to 65535.
      • Listen on ports: ELB allows you to create listeners that listen on ports within specified ranges. Each listener can support up to 10 non-overlapping port ranges.

        To configure port ranges for load balancer listeners, ensure the following conditions are met:

        • The cluster version must be v1.23.18-r0, v1.25.13-r0, v1.27.10-r0, v1.28.8-r0, v1.29.4-r0, v1.30.1-r0, or later.
        • A dedicated load balancer must be used with TCP/UDP/TLS selected.
        • This function requires ELB. Before using this function, check whether ELB supports full-port listening and forwarding for layer-4 protocols in the current region.
    • Frontend Protocol: Set the protocol of the load balancer listener for establishing connections with clients. When a dedicated load balancer is selected, HTTP/HTTPS can be configured only when Application (HTTP/HTTPS) is selected and TLS can be configured only when Network (TCP/UDP/TLS) is selected.
    NOTE:

    When a LoadBalancer Service is created, a random node port number (NodePort) is automatically generated.
    • Protocol: TCP
    • Container Port: 80
    • Service Port: 80
    • Frontend Protocol: HTTP

  6. Configure listener parameters.

    Parameter

    Description

    Constraint

    Example

    Idle Timeout

    The duration for which a client connection can remain idle before being terminated. If there are no requests reaching the load balancer during this period, the load balancer will disconnect the connection from the client and establish a new connection when there is a new request.

    Value:
    • For TCP listeners, the value ranges from 10 to 4000 (in seconds). The default value is 300.
    • For HTTP, TERMINATED_HTTPS, or HTTPS listeners, the value ranges from 0 to 4000 (in seconds). The default value is 60.
    • For UDP listeners (supported only by dedicated load balancers), the value ranges from 10 to 4000 (in seconds). The default value is 300.

    This configuration is not supported when UDP is used on a shared load balancer.

    60s

    Request Timeout

    The duration within which a request from a client must be received. There are two cases:

    • If the client fails to send a request header to the load balancer during this period, the request will be interrupted.
    • If the interval between two consecutive request bodies reaching the load balancer exceeds this period, the connection will be disconnected.

    The value ranges from 1 to 300 (in seconds). The default value is 60.

    This parameter is available only when the listener frontend protocol is HTTP or HTTPS.

    60s

    Response Timeout

    Timeout for backend server response. If the backend server does not respond within this period after receiving a request, the load balancer will stop waiting and return an HTTP 504 error.

    The value ranges from 1 to 300 (in seconds). The default value is 60.

    This parameter is available only when the listener frontend protocol is HTTP or HTTPS.

    60s

  7. Click Create.
  1. Log in to the CCE console and click the cluster name to access the cluster console.
  2. In the navigation pane, choose Services & Ingresses. In the upper right corner, click Create Service.
  3. Configure Service parameters. In this example, only mandatory parameters required for configuring timeout are listed. For details about how to configure other parameters, see Using the CCE Console (Old Version).

    • Service Name: Specify a Service name, which can be the same as the workload name.
    • Service Type: Select LoadBalancer.
    • Selector: Add a label and click Confirm. The Service will use this label to select pods. You can also click Reference Workload Label to use the label of an existing workload. In the dialog box that is displayed, select a workload and click OK.
    • Load Balancer: Select a load balancer type and creation mode.
      • A load balancer can be dedicated or shared.
      • This section uses an existing load balancer as an example. For details about the parameters for automatically creating a load balancer, see Table 4.
    • Port
      • Protocol: protocol that the load balancer complies with. Timeout cannot be configured for a UDP-compliant shared load balancer.
      • Service Port: the port used by the Service. The port ranges from 1 to 65535.
      • Container Port: the port that the workload listens on. For example, Nginx uses port 80 by default.
      • Frontend Protocol: Select a protocol for the listener. If HTTP/HTTPS is not enabled, only the idle timeout can be configured.
    • Listener
      • Advanced Options: Select a proper option.

        Configuration

        Description

        Constraint

        Idle Timeout

        The duration for which a client connection can remain idle before being terminated. If there are no requests reaching the load balancer during this period, the load balancer will disconnect the connection from the client and establish a new connection when there is a new request.

        Value:
        • For TCP listeners, the value ranges from 10 to 4000 (in seconds). The default value is 300.
        • For HTTP, TERMINATED_HTTPS, or HTTPS listeners, the value ranges from 0 to 4000 (in seconds). The default value is 60.
        • For UDP listeners (supported only by dedicated load balancers), the value ranges from 10 to 4000 (in seconds). The default value is 300.

        This configuration is not supported when UDP is used on a shared load balancer.

        Request Timeout

        The duration within which a request from a client must be received. There are two cases:

        • If the client fails to send a request header to the load balancer during this period, the request will be interrupted.
        • If the interval between two consecutive request bodies reaching the load balancer exceeds this period, the connection will be disconnected.

        The value ranges from 1 to 300 (in seconds). The default value is 60.

        This parameter is available only after HTTP/HTTPS is enabled on ports.

        Response Timeout

        Timeout for backend server response. If the backend server does not respond within this period after receiving a request, the load balancer will stop waiting and return an HTTP 504 error.

        The value ranges from 1 to 300 (in seconds). The default value is 60.

        This parameter is available only after HTTP/HTTPS is enabled on ports.
    Figure 1 Configuring timeout

  4. Click OK.
Use annotations to configure timeout. The following shows an example: 
apiVersion: v1 
kind: Service 
metadata: 
  annotations:
    kubernetes.io/elb.id: <your_elb_id>    # In this example, an existing dedicated load balancer is used. Replace its ID with the ID of your dedicated load balancer.
    kubernetes.io/elb.class: performance  # Load balancer type
    kubernetes.io/elb.protocol-port: http:80 # The HTTP port 80 is used.
    kubernetes.io/elb.keepalive_timeout: '300'  # Timeout setting for client connections
    kubernetes.io/elb.client_timeout: '60'      # Timeout for waiting for a request from a client
    kubernetes.io/elb.member_timeout: '60'      # Timeout for waiting for a response from a backend server
  name: nginx 
spec: 
  ports: 
  - name: service0 
    port: 80
    protocol: TCP 
    targetPort: 80
  selector: 
    app: nginx 
  type: LoadBalancer
Table 1 Key annotation parameters

Parameter

Mandatory

Type

Description

kubernetes.io/elb.keepalive_timeout

No

String

Timeout for client connections. If there are no requests reaching the load balancer during this period, the load balancer will disconnect the connection from the client and establish a new connection when there is a new request.

Value:
  • For TCP listeners, the value ranges from 10 to 4000 (in seconds). The default value is 300.
  • For HTTP, TERMINATED_HTTPS, or HTTPS listeners, the value ranges from 0 to 4000 (in seconds). The default value is 60.
  • For UDP listeners (supported only by dedicated load balancers), the value ranges from 10 to 4000 (in seconds). The default value is 300.

kubernetes.io/elb.client_timeout

No

String

Timeout for waiting for a request from a client. There are two cases:

  • If the client fails to send a request header to the load balancer during this period, the request will be interrupted.
  • If the interval between two consecutive request bodies reaching the load balancer exceeds this period, the connection will be disconnected.

The value ranges from 1 to 300 (in seconds). The default value is 60.

kubernetes.io/elb.member_timeout

No

String

Timeout for backend server response. After a request is forwarded to the backend server, if the backend server does not respond within the duration specified by member_timeout, the load balancer will stop waiting and return HTTP 504 Gateway Timeout.

The value ranges from 1 to 300 (in seconds). The default value is 60.

Verifying Configuration

  1. Log in to the CCE console and click the cluster name to access the cluster console.
  2. In the navigation pane, choose Services and Ingresses. Locate the row that contains the created Service and click the load balancer name to go to the ELB console.
  3. Switch to the Listeners tab, click the listener name of the created port, and check whether the options are enabled.

Parent Topic: Traffic and Performance Optimization Configuration