Microservices Engine:Migrate service registries with MSE Sync

ZooKeeper

When synchronizing persistent nodes, MSE Sync registers a watch on the source cluster for each Znode. If the source cluster contains many nodes, this creates a large number of watches and consumes significant memory. Make sure the source cluster's Znode count and resources can handle the additional watch listeners from MSE Sync.

Eureka to Nacos

Eureka registers service names in uppercase by default (for example, SERVICE-1). MSE Sync converts these names to lowercase when synchronizing to Nacos (for example, service-1).

This conversion can cause interoperability issues with mixed-case names. For example, if a service named Service-1 is registered directly in Nacos alongside the synchronized service-1, Nacos treats them as two separate services. Instances registered under one name cannot discover instances under the other.

To avoid this issue, standardize all service names to lowercase before starting the migration.

Consul to Nacos

MSE Sync synchronizes only healthy instance information from Consul. Metadata changes are not detected or synchronized automatically. Consul configuration data is not supported.

Choose a storage backend

MSE Sync supports two storage backends:

MSE Sync automatically selects the backend based on environment variables. If MYSQL_URL, MYSQL_USER_NAME, and MYSQL_PASSWORD are all set, MSE Sync connects to MySQL. Otherwise, it uses the embedded Derby database.

Configure JVM parameters

Add JVM parameters in the conf/java.env file using the JAVA_OPT variable:

JAVA_OPT="$JAVA_OPT -Xms2048m -Xmx2048m"

Deploy on Kubernetes

1. Save the following configuration as mse-sync-deployment.yaml. Replace the MySQL connection details and adjust replicas as needed.

   Note

   To access the MSE Sync web UI, create a Kubernetes Service that routes traffic to port 8000.

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     labels:
       app.kubernetes.io/name: mse-sync
     name: mse-sync-svc
   spec:
     replicas: 3
     selector:
       matchLabels:
         app.kubernetes.io/name: mse-sync
     template:
       metadata:
         labels:
           app.kubernetes.io/name: mse-sync
       spec:
         containers:
         - name: mse-sync
           image: msecrinstance-registry.cn-hangzhou.cr.aliyuncs.com/mse-demo/mse-sync:latest
           ports:
           - containerPort: 8000
           env:
           - name: MYSQL_URL
             value: "jdbc:mysql://192.xx.xx.xx:3306/sync"
           - name: MYSQL_USER_NAME
             value: "mse-sync"
           - name: MYSQL_PASSWORD
             value: password

2. Deploy MSE Sync:

   kubectl apply -f mse-sync-deployment.yaml

Deploy on Elastic Compute Service (ECS)

1. Download the MSE Sync binary:

   curl https://msesync.oss-cn-hangzhou.aliyuncs.com/MseSync.zip --output msesync.zip

2. Extract the package:

   unzip ./msesync.zip

3. (Optional) If using MySQL, set the database environment variables:

   export MYSQL_URL=""
   export MYSQL_USER_NAME=""
   export MYSQL_PASSWORD=""

4. Start MSE Sync:

   ./MseSync/bin/startup.sh start

Option A: Configure with a YAML file

A configuration file contains two sections: clusters (registry connection details) and tasks (synchronization jobs).

The following example synchronizes all services from the public and test namespaces in the source cluster to the destination cluster, using bidirectional sync:

clusters:
  - clusterName: src
    connectKeyList:
      - mse-xxxxx-nacos-ans.mse.aliyuncs.com:8848
    clusterType: NACOS
    namespace: public,test
  - clusterName: dst
    connectKeyList:
      - mse-xxxxxx-nacos-ans.mse.aliyuncs.com:8848
    clusterType: NACOS

tasks:
  - source: src
    destination: dst
    type: Service
    mode: Bidirectional
    serviceMatchPattern: ".*"
    autoScanAndAddService: true

Import the configuration using the CLI:

./bin/msesyncCMD.sh apply -f {path_to_configuration_file}

Alternatively, place the file at MseSync/conf/config.yaml. MSE Sync loads it automatically on startup.

Option B: Configure through the web UI

1. Open the MSE Sync web interface.

2. In the left navigation pane, click Service Sync.

3. Click Import Configuration.

4. In the Import Configuration dialog box, paste the YAML configuration and click OK.

Cluster fields

connectKeyList example:

connectKeyList:
  - localhost:8848
  - localhost1:8848

Task fields

Recovery scan interval

The recovery scan interval controls how frequently healthy nodes check for failed nodes. Configure it with the following Java Virtual Machine (JVM) parameter, set through the JAVA_OPT environment variable:

-Dmsesync.daemon.refresh.interval=60  // Unit: seconds

Example:

JAVA_OPT="$JAVA_OPT -Dmsesync.daemon.refresh.interval=10"

During the interval between a node failure and the next scheduled recovery scan, the client's instance list may be temporarily empty.

Description of additional YAML configuration fields

• Clusters field:

  Field	Description	Example value
  namespace	The Nacos namespace name. Note: You can enter multiple namespaces, separated by commas. If this field is left empty and the instance is the synchronization source, data from all namespaces in the instance is synchronized.	public test test1
  ak	The AccessKey for authenticating the cloud MSE Nacos instance.	LTAI5***********dXai6
  sk	The AccessKey secret for authenticating the cloud MSE Nacos instance.	Jdvdj*************6vs7wBEKO