Restarting a Single Controller

Procedure

1. Log in to the command-line interface (CLI) of the storage system as user admin.
2. Before the reset, check whether the node to be reset meets the five resetting prerequisites.
   1. The cluster status is normal and the clustered system is not in the power-on or power-off state.
      1. Run the change user_mode current_mode user_mode=developer command to go to the developer mode.
      2. Run the debug command to enter the diagnose mode.
      3. Run the sys showflowstatus command to check the current system status.

         
         
         If FlowClusterServiceStaus is Normal, the cluster status is normal and a single node can be reset.

   2. The host connects to other controllers in the cluster at the same time through multiple paths. You can use either of the following methods to check information about multiple paths.
      • Method 1: Check whether the host connects to different nodes through two or more physical channels.
      • Method 2: Query system alarms and events to check whether an alarm (alarm ID F00150019ULL) exists indicating that no redundant paths are available between the host and the storage array. If there is such an alarm, you cannot reset the single node.

   3. Controllers in the engine are connected to all disks (all detected member disks and one of the first four disks) in the disk array.

      In DeviceManager, check the multipath status of disks.
      
      If Multipath is A,B, the disks are connected to controllers A and B.

   4. Controller nodes in the engine work properly.
      1. Run the change user_mode current_mode user_mode=developer command to go to the developer mode.
      2. Run the debug command to enter the diagnose mode.
      3. Run the sys showflowstatus command to check the current system status.

         
         
         If FlowNodeServiceStaus(id X) is Normal, node X works properly.

   5. The current capacity of controller B can handle service pressure of controller A.
      1. Run the change user_mode current_mode user_mode=developer command to go to the developer mode.
      2. Run the debug command to enter the diagnose mode.
      3. Run the ioperf iobandwidth 5 command to check whether the front-end pressure of the current controller is smaller than 80%.

         
         
         If the returned value is True, the front-end pressure of the current controller is smaller than 80%.

3. After all check items are passed, reset the node.
   1. Run the change user_mode current_mode user_mode=developer command to go to the developer mode.
   2. Run the reboot controller command to reset a single node. Use controller to specify the ID of the node that you want to reset.
4. After the node reset, confirm that the node is successfully powered on and can take over services.
   1. Run the change user_mode current_mode user_mode=developer command to go to the developer mode.
   2. Run the debug command to enter the diagnose mode.
   3. Run the sys showflowstatus command to check the current system status.

      
      
      If FlowNodeServiceStaus(id X) is Normal, node X works properly and can take over services.

5. Repeat 2, 3, and 4 to reset other controller nodes that failed to pass the check.