File Storage NAS:Best practices for migrating Extreme NAS file systems

Step 2.1: Prepare and mount the migration ECS instance

1. Log on to the ECS Management Console. Create a temporary pay-as-you-go ECS instance in the same region, zone, and VPC as the NAS instance.

2. Log on to the ECS instance. Create local directories to mount the old and new file systems:

   sudo mkdir -p /mnt/nas_old
   sudo mkdir -p /mnt/nas_new

3. Mount the old and new instances. Go to the details page of the target file system, choose Mount Targets, and copy the full mount target address from the mount target list.

   # Mount the old instance (replace <old_mount_target_address> with your actual address)
   sudo mount -t nfs -o vers=3,noacl,nolock,proto=tcp,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport <old_mount_target_address>:/share /mnt/nas_old
   
   # Mount the new instance (replace <new_mount_target_address> with your actual address)
   sudo mount -t nfs -o vers=3,noacl,nolock,proto=tcp,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport <new_mount_target_address>:/share /mnt/nas_new

4. Run the df -h command to confirm that both file systems are mounted.

Step 2.2: Install tools and synchronize data

1. Install the rsync and tmux tools:

   sudo yum install -y rsync tmux

   • rsync is the core data synchronization tool.

   • tmux is a terminal multiplexer. We strongly recommend running long rsync commands in tmux to prevent the migration task from being interrupted by an unexpected Secure Shell (SSH) disconnection.

2. Start a tmux session:

   tmux

3. Perform a full data synchronization. This process may take a long time, so please be patient.

   # Note: The source path /mnt/nas_old/ must end with a forward slash (/).
   sudo rsync -avP /mnt/nas_old/ /mnt/nas_new/

   If you have many small files or a large amount of data, see Parallel rsync commands to significantly improve synchronization efficiency.

4. Perform the final incremental synchronization.

   1. During off-peak hours, stop all applications that read from or write to the old NAS instance.

   2. On the migration ECS instance, run the incremental synchronization command.

      Warning

      The --delete option deletes data from the destination file system that has been deleted from the source file system. Use this option with caution to avoid accidentally deleting data from the destination file system.

      # The --delete parameter ensures that the destination directory is identical to the source directory.
      sudo rsync -avP --delete /mnt/nas_old/ /mnt/nas_new/

5. Perform a data consistency check.

   Before the service switchover, you must confirm that the data on the old and new instances is identical.

   echo "Checking data consistency. This process will not modify any data..."
   sudo rsync -rvn --delete /mnt/nas_old/ /mnt/nas_new/

   Expected result: If the data is identical, the command output does not list any file names between the header and footer statistics. If any file names appear, it indicates that there are differences. You must roll back or re-synchronize before the switchover.

Step 3.1: Manually switch the mount target

1. Stop your business applications, such as Nginx and Tomcat.

2. Unmount the old mount target. This example uses the local mount directory /var/www/html/uploads on the application server:

   sudo umount /var/www/html/uploads

3. Mount the new file system to the same local directory:

   sudo mount -t nfs -o vers=3,noacl,nolock,proto=tcp,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport <new_mount_target_address>:/share /var/www/html/uploads

4. Run df -h to confirm that the new file system is mounted.

5. Restart the application and perform a quick functional test, such as uploading and downloading files.

Step 3.2: Configure fstab for automatic mounting

After the manual switchover is successful, configure fstab for persistent automatic mounting.

1. Run the following command to open the rpcbind.socket configuration file:

   sudo vi /etc/systemd/system/sockets.target.wants/rpcbind.socket

2. Comment out the IPv6-related rpcbind parameter as shown in the following figure. Otherwise, the NFS rpcbind service will fail to start automatically.

   

3. If you are configuring automatic mounting on a CentOS 6.x system, perform these additional steps:

   1. Run the chkconfig netfs on command to ensure that the netfs service starts on boot.

   2. Edit the /etc/netconfig configuration file and comment out the inet6-related content as shown in the following figure.

      

   3. Save and exit the file.

4. Back up the current fstab configuration.

   sudo cp /etc/fstab /etc/fstab.bak.$(date +%Y%m%d)

5. Edit the /etc/fstab file. Comment out the old mount target information and add a new line:

   <file-system-id.region.extreme.nas.aliyuncs.com>:/share </mnt> nfs vers=3,nolock,noacl,proto=tcp,noresvport,_netdev 0 0

   • <file-system-id.region.nas.aliyuncs.com>: Replace this with the actual NAS mount target address. (Example: 00a*****-****.cn-hangzhou.extreme.nas.aliyuncs.com.)

   • </mnt>: Replace this with the local mount path. (Example: /mnt).

6. Verify that the system can mount based on the new fstab configuration.

   sudo mount -a

7. Run the df -h command to check if the mount was successful.

Step 3.3: Start and verify the service

1. Restart your application.

2. Perform comprehensive functional regression testing to ensure that data reads and writes work correctly.

3. Monitor the service for a period of time, such as one hour, to confirm that there are no issues.

Step 4.1: Delete the old mount target

1. In the Apsara File Storage NAS console, find the old file system instance and go to its Mount Targets page.

2. Find the target mount target. In the Actions column, click > Delete and confirm the action.

Step 4.2: Delete the old file system instance

1. After you confirm that all mount targets are deleted, return to the File System List page.

2. Find the old target instance. In the Actions column, click > Delete.

Step 4.3: Release the migration ECS instance

1. Log on to the ECS Management Console and find the temporary ECS instance used for migration.

2. In the Actions column, choose > Instance Status > Stop.

3. Wait for the instance status to become Stopped. Then, in the Actions column, choose > Instance Status > Release and follow the instructions to release the instance.

Notes for Kubernetes environments

If your service mounts NAS through Kubernetes (K8s), the switchover steps are as follows:

1. Modify the existing dynamic or static volume YAML configuration file. Replace the old mount target with the new one.

2. Generate a new pod with the modified configuration file. Confirm that it successfully mounts the new file system and can read and write data correctly.

3. Evict all pods that use the legacy file system.

Emergency rollback plan

If you find any issues during verification in Stage 3, perform a rollback immediately:

1. Stop the application.

2. On the application server, unmount the new mount target.

3. Restore the old /etc/fstab configuration.

4. Remount the old file system (sudo mount -a) and confirm with df -h.

5. Restart the application and confirm that it has recovered.