This site contains the documentation that is relevant to older WSO2 product versions and offerings.
For the latest WSO2 documentation, visit https://wso2.com/documentation/.

Deploying WSO2 API-M in a Distributed Setup

Owned by Former user (Deleted)
Last updated: Apr 07, 2021 by Former user (Deleted)
19 min read

Follow the instructions below to deploy WSO2 API Manager (WSO2 API-M) in a distributed environment, as depicted in the following deployment diagram:


Step 1 - Install and configure WSO2 API-M

 Click here for information on installing and configuring WSO2 API-M.

The following steps describe how to download, install, and configure WSO2 API Manager, with five instances.

  1. Download the WSO2 API Manager in each of the five servers in the cluster for distributed deployment.

  2. Unzip the WSO2 API Manager zipped archive, and rename each of those directories respectively as Key Manager, Gateway, Publisher, Store, and Traffic Manager.
    These five directories are located in a server of their own and are used for each component of WSO2 API-M. Each of these unzipped directories are referred to as <API-M_HOME> or <PRODUCT_HOME> in this document.

  3. In each of the five servers, replace the default certificates (where CN=localhost) with new certificates generated with proper common name (CN) values.
    You need to do this in order to avoid getting an error with regard to the fact that the hostname in the certificate did not match.

    Note that you should use the same primary keystore for all the API Manager instances here in order to decrypt the registry resources. For more information, see Configuring Primary Keystores in the Administration Guide. When creating the keystore, always use a longer validity period so that it will avoid the need of migration on the registry data when shifting to a new keystore.

Step 2 - Install and configure the databases

You can create the required databases for the API-M deployment in a separate server and point to the databases from the respective nodes. For information on configuring the databases, see Installing and Configuring the Databases.

Step 3 - Configure your deployment with production hardening

Ensure that you have taken into account the respective security hardening factors (e.g., changing and encrypting the default passwords, configuring JVM security, etc.) before deploying WSO2 API-M. For more information, see the Production Deployment Guidelines in the Administration Guide.

Step 4 - Create and import SSL certificates

Create a SSL certificate for each of the WSO2 API-M nodes (e.g., Publisher, Store, Key Manager, Gateway, and Traffic Manager) and import them to the keyStore and the trustStore. For more information, see Creating SSL Certificates in the Administration Guide. 

When maintaining high availability (HA) in the WSO2 API-M distributed set up, you need to create and import a SSL certificate for each of the WSO2 API-M HA nodes.

Step 5 - Configure API-M Analytics

If you wish to view reports, statistics, and graphs related to the APIs deployed in the Store, you need to configure API-M Analytics. Follow the standard setup to configure API-M Analytics in a production setup, and follow the quick setup to configure API-M Analytics in a development setup.

Step 6 - Configure the connections among the components and start the servers

You will now configure the inter-component relationships of the distributed setup by modifying their <API-M_HOME>/repository/conf/api-manager.xml files. It is recommended to start the components in the following order: Key Manager, Publisher, Store, Traffic Manager, and Gateway.

In a clustered environment, you use session affinity (sticky sessions) to ensure that requests from the same client always get routed to the same server.

It is  mandatory  to set up Session Affinity in the load balancers that front the  Publisher  and  Store  clusters, and it is  optional  in the load balancer (if any) that fronts a  Key Manager  cluster or Gateway Cluster. However, you need to enable Session Affinity if you are working with multiple Gateway Managers in a Gateway High Availability (HA) deployment.

However, authentication via session ID fails when session affinity is disabled in the load balancer.

First time authentication happens via Basic Auth and the Gateway gets a cookie. This cookie is used in every consequent request along with the Basic Auth credentials. The admin service validates the cookie and if the validation fails it re-authenticates it using Basic Auth and issues a new cookie.

  • Unlike in WSO2 API-M 2.2.0, the common configurations required when configuring the connections among the components are automatically handled via the profile optimization process. Therefore, you do not need to do any common configurations manually.

  • Metrics is disabled in WSO2 API Manager by default. However, if you need to use metrics with WSO2 API Manager, make sure to enable it in the <API-M_HOME>/repository/conf/metrics.xml file in each of the API-M nodes as follows:

            <JDBC>
            <Enabled>true</Enabled>
 			...
        </JDBC>

    For more information, see Enabling Metrics and Storage Types.

Step 6.1 - Configure and start the Key Manager

 Click here for instructions to configure and start the Key Manager.

This section involves setting up the Key Manager node and enabling it to work with the other components in a distributed deployment.

Skip this step if you are using WSO2 Identity Server as the Key Manager and follow the instructions mentioned in Configuring WSO2 Identity Server as a Key Manager to configure and start the Key Manager.

  1. Open the <API-M_HOME>/repository/conf/api-manager.xml file in the Key Manager node and change the <ServerURL> element that appears under the <APIGateway> section, so that it points to the API Manager Gateway.  
    You need to add these configurations so that when a user is deleted or when the role of a user is updated in the Key Manager, it will update the Gateway cache by clearing the cache entries of a particular user. 
    • If you are working with a single Gateway in distributed set up, you need to replace [GATEWAY_SERVER_HOST] with the host of the Gateway node.
    • If you are working with Gateways in a High Availability (HA) setup that uses a shared file system (e.g., NFS), you need to replace [GATEWAY_SERVER_HOST] with the host of the Gateway load balancer node.
    • If you are working with Gateways in a High Availability (HA) setup that uses rsync, you need to replace [GATEWAY_SERVER_HOST] with the host of the Gateway Manager node.

    • You need to replace [port] with the management transport port. For more information, see Default Product Ports.

      <ServerURL>https://$[GATEWAY_SERVER_HOST]:[port]/services/</ServerURL>

  2. Configure the API key validator in the Key Manager.
    The Thrift protocol is normally enabled by default. However, if you have disabled the Thrift protocol, enable it as follows in the <API-M_HOME>/repository/conf/api-manager.xml file. 

    If you wish to encrypt the Auth Keys (access tokens, client secrets, and authorization codes), see Encrypting OAuth Keys.

  3. Disable the Policy Deployer under the Throttling configurations.

    <ThrottlingConfigurations>
		...
        <PolicyDeployer>
            <Enabled>false</Enabled>
			...
        </PolicyDeployer>
		...
    </ThrottlingConfigurations>

  4. Optionally, configure High Availability (HA) for the Key Manager.

    These steps are ONLY applicable if you need to configure HA for the Key Manager.

    1. Make a copy of the active instance configured above and use this copy as the second Key Manager active instance.

    2. Configure a load balancer to front the two Key Manager nodes.
      For information on configuring the load balancer, see Configuring the Proxy Server and the Load Balancer.

  5. Configure the Key Manager to communicate with the Traffic Manager.

    This is available only as a  WUM update and is effective from 30th September 2020 (2020-09-30). For more information on updating WSO2 API Manager, see Updating WSO2 API Manager.

  6. Add the following in the <API-M_HOME>/repository/conf/api-manager.xml file for each of the Key Manager nodes in the deployment.

    <TokenRevocationNotifier>
       <Enabled>true</Enabled>
</TokenRevocationNotifier>

  7. Start the WSO2 API-M Key Manager node(s).
    Make sure to run the product optimizer either before starting the server or while starting the server, so that the resource utilization can be optimized on each of the nodes. For more information on product profile optimization, see Product Profiles.

It is not recommended to share the Solr directory between the Store and Publisher servers. You need to have separate Solr directories for each of the latter mentioned servers so that they will perform Solr indexing separately.

If you get an error similar to the following in both or one of the nodes, check whether you have shared the Solr directory.

org.apache.solr.common.SolrException: 
SolrCore 'registry-indexing' is not available due to init failure: Index locked for write for core registry-indexing

Step 6.2 - Configure and start the API Publisher

 Click here for instructions to configure and start the Publisher.

This section involves setting up the API Publisher node and enabling it to work with the other components in the distributed deployment.

  1. Open the <API-M_HOME>/repository/conf/api-manager.xml file in the API Publisher node and make the following changes.

    1. Configure the Publisher with the Key Manager.
      You need to update the following configuration ONLY when you do not wish to share the user stores with the WSO2 API-M instance.

      This step is not applicable if you are enabling Single Sign-on (SSO).

    2. Configure the Publisher with the Traffic Manager.
      This configuration enables the publishing of throttling policies, custom templates, and block conditions to the Gateway node.

    3. Configure the Publisher with the Gateway.
      You need to add these configurations, because when creating an API, it calls the Gateway endpoint to create the actual Synapse file. 


      • If you are using a single Gateway node, configure the Publisher with the Gateway as follows:

        <APIGateway>
   <Environments>
       <Environment type="hybrid" api-console="true">
            <Name>Production and Sandbox</Name>
            <Description>This is a hybrid gateway that handles both production and sandbox token traffic.</Description>           
			<ServerURL>https://[API-Gateway-Host-or-IP]:9443/services/</ServerURL>
            <Username>${admin.username}</Username>
            <Password>${admin.password}</Password>          
			<GatewayEndpoint>http://[API-Gateway-Host]:8280,https://[API-Gateway-Host]:8243</GatewayEndpoint>
       </Environment>
   </Environments>
</APIGateway>

      • If you are using multiple Gateway nodes, configure the Publisher with the Gateway nodes as follows:

      To change the admin password, see Changing the super admin password. Furthermore, make sure to adhere to the note given under step 2 in the latter mentioned section if your password has special characters.

    4. Configure the Store URL to appear in the Publisher UI.
      For this purpose you need to set the <DisplayURL> to true and provide the URL of the Store.

  2. Disable the Thrift Server to optimize performance.
    You need to configure this in the Publisher <API-M_HOME>/repository/conf/api-manager.xml file.

    <APIKeyValidator> 
...  
	<EnableThriftServer>false</EnableThriftServer>
</APIKeyValidator>

  3. Optionally, configure High Availability (HA) for the Publisher.

    These instructions are ONLY applicable if you need to configure HA for the Publisher.

    1. Make a copy of the active Publisher instance configured above and use this copy as the second active Publisher instance.

    2. Configure a load balancer to front the two Publisher nodes.
      For information on configuring the load balancer, see Configuring the Proxy Server and the Load Balancer.

  4. Start the WSO2 API-M Publisher node(s).
    Make sure to run the product optimizer either before starting the server or while starting the server, so that the resource utilization can be optimized on each of the nodes. For more information on product profile optimization, see Product Profiles.

Step 6.3 - Configure and start the API Store

 Click here for instructions to configure and start the API Store.

This section involves setting up the API Store node and enabling it to work with the other components in the distributed deployment.

  1. Open the  <API-M_HOME>/repository/conf/api-manager.xml  file in the API Store node and make the following changes.

    1. Configure the API Store with the Key Manager.

    2. Make the following throttling related changes that correspond to the Traffic Manager.

      <ThrottlingConfigurations>
        <EnableAdvanceThrottling>true</EnableAdvanceThrottling>
        <DataPublisher>
            <Enabled>false</Enabled>
	    ...
        </DataPublisher>
        <PolicyDeployer>
            <Enabled>false</Enabled>
			...
        </PolicyDeployer> 
	    ...
        <BlockCondition>
            <Enabled>false</Enabled>
	    ...
        </BlockCondition>
        <JMSConnectionDetails>
            <Enabled>false</Enabled>
	     ...
        </JMSConnectionDetails>
     ...
</ThrottlingConfigurations>

    3. Configure the Store with the Gateway. 

      • If you are using a single Gateway node, configure the Store with the Gateway as follows:

        <APIGateway>
	<Environments>
		<Environment type="hybrid">
		...
			<ServerURL>https://[API-Gateway-host-or-IP]:9443/services/</ServerURL>
			<Username>${admin.username}</Username>
			<Password>${admin.password}</Password> 
			<GatewayEndpoint>http://[API-Gateway-host]:8280,https://[API-Gateway-host]:8243</GatewayEndpoint>
		</Environment>
	</Environments> 
...
</APIGateway>

      • If you are using multiple Gateway nodes, configure the Store with the Gateway nodes as follows:

    4. Configure the Token Revoke endpoint to point to Gateway.

  2. Optionally, configure High Availability (HA) for the Store.

    This is ONLY applicable if you need to configure HA for the Store.

    Make a copy of the active instance configured above and use this copy as the second API Store active instance.

  3. Start the API Store node(s).
    Make sure to run the product optimizer either before starting the server or while starting the server, so that the resource utilization can be optimized on each of the nodes. For more information on product profile optimization, see Product Profiles.

Step 6.4 - Configure and start the Traffic Manager

 Click here for instructions to configure and start the Traffic Manager.

This section involves setting up the Traffic Manager node(s) and enabling it to work with the other components in a distributed deployment.

Note that this is the recommended method to sync events among the Traffic Manager nodes. This does not require the use of Hazelcast, because it increases the network calls between nodes. Therefore, we recommend using this method by following the instructions below. 

  1. Delete the <API-M_HOME>/repository/conf/registry.xml file and rename the <API-M_HOME>/repository/conf/registry_TM.xml file as the registry.xml  file.
    To disable registry indexing when setting up the Traffic Manager, see Registry indexing configurations.

  2. Delete the <API-M_HOME>/repository/conf/axis2/axis2.xml file and rename the <API-M_HOME>/repository/conf/axis2/axis2_TM.xml file as the axis2.xml file.

    Note that all the jaggery apps and webapps in the <API-M_HOME>/repository/deployment/server/ directory are removed when you optimize the profile. For more details, see Product Profiles.

  3. Optionally, mount the <API-M_HOME>/repository/deployment/server directory of all the Traffic Manager nodes to the shared file system.

    This step is ONLY applicable if you are configuring the Traffic Manager with HA and shared file system as the content synchronization mechanism.

    You need to do this to share all the Throttling policies between traffic management nodes.

  4. Disable the Thrift Server to optimize performance.
    You need to configure this in the Traffic Manager <API-M_HOME>/repository/conf/api-manager.xml file.

    <APIKeyValidator> 
...  
	<EnableThriftServer>false</EnableThriftServer>
</APIKeyValidator>

  5. Make sure that the Policy Deployer is enabled under the Throttling configurations.

    <ThrottlingConfigurations>
		...
        <PolicyDeployer>
            <Enabled>true</Enabled>
        </PolicyDeployer>
		...
    </ThrottlingConfigurations>

  6. Optionally, configure High Availability (HA) for the Traffic Manager.

    This is ONLY applicable if you need to configure HA for the Traffic Manager.

    Make a copy of the active instance configured above and use this copy as the second active Traffic Manager instance.

  7. Start the WSO2 API-M Traffic Manager node(s).
    Make sure to run the product optimizer either before starting the server or while starting the server, so that the resource utilization can be optimized on each of the nodes. For more information on product profile optimization, see Product Profiles.

    Always start the Traffic Manager using the -Dprofile=traffic-manager profile to avoid FATAL errors such as the following.

    FATAL - ServiceBusInitializer Couldn't initialize the ESB...
org.apache.synapse.SynapseException: The synapse.xml location ././
        ./repository/deployment/server/synapse-configs
    /default doesn't exist

Troubleshooting

If you have a firewall between the Traffic Manager and the Gateway, you need to configure the heartbeat value to keep the JMS connection alive. To configure this, open the  <APIM_HOME>/repository/conf/advanced/qpid-config.xml  file and set the heartbeat to a non-zero value as shown below.

<heartbeat>    
	<delay>60</delay>
    <timeoutFactor>2.0</timeoutFactor>
</heartbeat>

Step 6.5 - Configure and start the Gateway

 Click here for instructions to configure and start the Gateway.

This section involves setting up the Gateway node and enabling it to work with the other components in the distributed deployment.

Steps 1 to 5 in the following section are common irrespective of your API-M deployment, such as deploying a single Gateway node or deploying multiple Gateway nodes for High Availability (HA). However, if you are using two Gateway nodes for high availability (HA), first follow the instructions that is available in the Distributed Deployment of the Gateway document, and then carry out the following steps to configure the connections from Gateway(s) to other components.

  1. Open the <API-M_HOME>/repository/conf/api-manager.xml file in the Gateway node.

  2. Modify the api-manager.xml file as follows. This configures the connection to the Key Manager component.

  3. If you need to enable JSON Web Token (JWT),  you have to enable it in all Gateway and Key Manager components.
    For more information on configuring JWT, see Generating JSON Web Token.

  4. Configure the Gateway to communicate with the Traffic Manager.
    You need to do this to enable Throttling for the Traffic Manager node(s).

    These configurations vary based on whether you have a single Traffic Manager node or multiple Traffic Manager nodes.

    By default, WSO2 API Manager is shipped with a keystore (wso2carbon.jks) and a trust store (client-truststore.jks). For more information on how to create a new key store and a trust store with a private key and a self-signed certificate, see Configuring Keystore and Truststore and also see the Administration guide for recommendations on setting up keystores in WSO2 products.

  5. Start the WSO2 API-M Gateway node(s).
    Make sure to run the product optimizer either before starting the server or while starting the server, so that the resource utilization can be optimized on each of the nodes. For more information on product profile optimization, see Product Profiles.