Service Stopped Check

Procedure

1. Confirm with frontline warning implementation personnel that service stopping operations have been completed based on the frontline implementation solution.

   Services to be stopped include cluster software and databases. The file system needs to umount the corresponding drive.

2. Use the upgrade evaluation tool to check whether services are stopped.
   • If services are stopped => No further action is required.
   • If any service fails to be stopped => 3.

3. Contact the service-side maintenance personnel to check whether all the services are stopped.
   • Yes => 4.
   • No => Stop services and go to step 4.

4. Use the upgrade evaluation tool to check whether services are stopped.
   • If services are stopped => No further action is required.
   • If any service fails to be stopped => 5.

5. Check which services are not stopped on the storage device.
   • Log in to the ISM.

     Perform 5.a to 5.c.

   • If you cannot log in to the ISM:
     • SUSE hosts => perform 5.d to 5.g.
     • Hosts running in other OSs => 6.
   1. In the ISM, go to the Performance Monitoring page.
      
      
   2. Select Block Storage Pool/Disk/LUN for Object Type, select LUNs in sequence. Select Basic Data Type for Statistics Type, select Throughput (IOPS), and check whether IOPS of the LUNs is 0 in the right performance statistics view.

      If the IOPS of the LUNs is not 0, the LUNs contain I/Os.

   3. Find the service that is not stopped based on the LUN name and stop the service under the guidance of the service-side maintenance personnel.
   4. Log in to each host, run the iostat -xd 2 command and observe the I/O status of the drives.
      
      
   5. Identify drives that have I/Os.
      • sda can be ignored because it is the local drive on the host.
      • sdb and sdc drives are the external drives connected to the host.
      • r/s represents the read I/Os per second that are sent to an external drive by the host.
      • w/s represents the write I/Os per second that are sent to an external drive by the host.

      If r/s or w/s is not zero, the host is continually sending I/Os to the external drive.

   6. Based on how the external drive is used, run the fuser commands to troubleshoot the I/O process.
      1. Run the fuser -m -v /dev/sdN command to query all processes that are accessing an external drive in the raw device mode.

         N represents the drive letter, for example, b, c, and d.

      2. Run the fuser -m -v /dev/sdb1 or fuser -m -v /mnt command to query all processes that are accessing the external drive in the file system mode.

         sdb1 represents the partition where the file system is created. mnt represents the mounting point of the partition.

      3. Run the fuser -m -v /dev/vg/lv or fuser -m -v /mnt command to query all processes that are accessing the external drive in the volume management mode.

         vg and lv represent the volume group name and volume name respectively. mnt represents the mounting point of the volume.

   7. Based on the query results, contact the service maintenance engineer and identify what services are accessing the external drive and which services are not stopped. Then stop the service under the guidance of the service maintenance engineer.
6. If you have any questions, please contact technical support engineers for further handling.