Skip to content

Latest commit

 

History

History
157 lines (120 loc) · 6.91 KB

load-balancer-tcp-idle-timeout.md

File metadata and controls

157 lines (120 loc) · 6.91 KB
title description services documentationcenter author manager ms.assetid ms.service ms.devlang ms.topic ms.tgt_pltfrm ms.workload ms.date ms.author
Configure Load Balancer TCP idle timeout | Microsoft Docs
Configure Load Balancer TCP idle timeout
load-balancer
na
kumudd
timlt
4625c6a8-5725-47ce-81db-4fa3bd055891
load-balancer
na
article
na
infrastructure-services
09/25/2017
kumud

Configure TCP idle timeout settings for Azure Load Balancer

[!INCLUDE load-balancer-basic-sku-include.md]

In its default configuration, Azure Load Balancer has an idle timeout setting of 4 minutes. If a period of inactivity is longer than the timeout value, there's no guarantee that the TCP or HTTP session is maintained between the client and your cloud service.

When the connection is closed, your client application may receive the following error message: "The underlying connection was closed: A connection that was expected to be kept alive was closed by the server."

A common practice is to use a TCP keep-alive. This practice keeps the connection active for a longer period. For more information, see these .NET examples. With keep-alive enabled, packets are sent during periods of inactivity on the connection. These keep-alive packets ensure that the idle timeout value is never reached and the connection is maintained for a long period.

This setting works for inbound connections only. To avoid losing the connection, you must configure the TCP keep-alive with an interval less than the idle timeout setting or increase the idle timeout value. To support such scenarios, we've added support for a configurable idle timeout. You can now set it for a duration of 4 to 30 minutes.

TCP keep-alive works well for scenarios where battery life is not a constraint. It is not recommended for mobile applications. Using a TCP keep-alive in a mobile application can drain the device battery faster.

TCP timeout

The following sections describe how to change idle timeout settings in virtual machines and cloud services.

Configure the TCP timeout for your instance-level public IP to 15 minutes

Set-AzurePublicIP -PublicIPName webip -VM MyVM -IdleTimeoutInMinutes 15

IdleTimeoutInMinutes is optional. If it is not set, the default timeout is 4 minutes. The acceptable timeout range is 4 to 30 minutes.

Set the idle timeout when creating an Azure endpoint on a virtual machine

To change the timeout setting for an endpoint, use the following:

Get-AzureVM -ServiceName "mySvc" -Name "MyVM1" | Add-AzureEndpoint -Name "HttpIn" -Protocol "tcp" -PublicPort 80 -LocalPort 8080 -IdleTimeoutInMinutes 15| Update-AzureVM

To retrieve your idle timeout configuration, use the following command:

PS C:\> Get-AzureVM -ServiceName "MyService" -Name "MyVM" | Get-AzureEndpoint
VERBOSE: 6:43:50 PM - Completed Operation: Get Deployment
LBSetName : MyLoadBalancedSet
LocalPort : 80
Name : HTTP
Port : 80
Protocol : tcp
Vip : 65.52.xxx.xxx
ProbePath :
ProbePort : 80
ProbeProtocol : tcp
ProbeIntervalInSeconds : 15
ProbeTimeoutInSeconds : 31
EnableDirectServerReturn : False
Acl : {}
InternalLoadBalancerName :
IdleTimeoutInMinutes : 15

Set the TCP timeout on a load-balanced endpoint set

If endpoints are part of a load-balanced endpoint set, the TCP timeout must be set on the load-balanced endpoint set. For example:

Set-AzureLoadBalancedEndpoint -ServiceName "MyService" -LBSetName "LBSet1" -Protocol tcp -LocalPort 80 -ProbeProtocolTCP -ProbePort 8080 -IdleTimeoutInMinutes 15

Change timeout settings for cloud services

You can use the Azure SDK to update your cloud service. You make endpoint settings for cloud services in the .csdef file. Updating the TCP timeout for deployment of a cloud service requires a deployment upgrade. An exception is if the TCP timeout is specified only for a public IP. Public IP settings are in the .cscfg file, and you can update them through deployment update and upgrade.

The .csdef changes for endpoint settings are:

<WorkerRole name="worker-role-name" vmsize="worker-role-size" enableNativeCodeExecution="[true|false]">
    <Endpoints>
    <InputEndpoint name="input-endpoint-name" protocol="[http|https|tcp|udp]" localPort="local-port-number" port="port-number" certificate="certificate-name" loadBalancerProbe="load-balancer-probe-name" idleTimeoutInMinutes="tcp-timeout" />
    </Endpoints>
</WorkerRole>

The .cscfg changes for the timeout setting on public IPs are:

<NetworkConfiguration>
    <VirtualNetworkSite name="VNet"/>
    <AddressAssignments>
    <InstanceAddress roleName="VMRolePersisted">
    <PublicIPs>
        <PublicIP name="public-ip-name" idleTimeoutInMinutes="timeout-in-minutes"/>
    </PublicIPs>
    </InstanceAddress>
    </AddressAssignments>
</NetworkConfiguration>

REST API example

You can configure the TCP idle timeout by using the service management API. Make sure that the x-ms-version header is set to version 2014-06-01 or later. Update the configuration of the specified load-balanced input endpoints on all virtual machines in a deployment.

Request

POST https://management.core.windows.net/<subscription-id>/services/hostedservices/<cloudservice-name>/deployments/<deployment-name>

Response

<LoadBalancedEndpointList xmlns="http://schemas.microsoft.com/windowsazure" xmlns:i="http://www.w3.org/2001/XMLSchema-instance">
    <InputEndpoint>
    <LoadBalancedEndpointSetName>endpoint-set-name</LoadBalancedEndpointSetName>
    <LocalPort>local-port-number</LocalPort>
    <Port>external-port-number</Port>
    <LoadBalancerProbe>
        <Path>path-of-probe</Path>
        <Port>port-assigned-to-probe</Port>
        <Protocol>probe-protocol</Protocol>
        <IntervalInSeconds>interval-of-probe</IntervalInSeconds>
        <TimeoutInSeconds>timeout-for-probe</TimeoutInSeconds>
    </LoadBalancerProbe>
    <LoadBalancerName>name-of-internal-loadbalancer</LoadBalancerName>
    <Protocol>endpoint-protocol</Protocol>
    <IdleTimeoutInMinutes>15</IdleTimeoutInMinutes>
    <EnableDirectServerReturn>enable-direct-server-return</EnableDirectServerReturn>
    <EndpointACL>
        <Rules>
        <Rule>
            <Order>priority-of-the-rule</Order>
            <Action>permit-rule</Action>
            <RemoteSubnet>subnet-of-the-rule</RemoteSubnet>
            <Description>description-of-the-rule</Description>
        </Rule>
        </Rules>
    </EndpointACL>
    </InputEndpoint>
</LoadBalancedEndpointList>

Next steps

Internal load balancer overview

Get started configuring an Internet-facing load balancer

Configure a load balancer distribution mode