Configurations and Operations After an HBA Replacement (in HP-UX)

Prerequisites

The old HBA of the host has been replaced.

Procedure

1. Back up the NMP multipath and disk information.
   1. Run scsimgr get_attr -a leg_mpath_enable to view and back up the multipath status.
   2. Run vgdisplay -v to view and back up volume group (VG) information.
   3. Run pvdisplay to view physical volume (PV) information.
   4. Run ioscan -fkNC disk and ioscan -fnkC fc to back up disk and HBA information.

2. Check the storage system status. If there are alarms, clear them and then proceed to the next step.
3. On the storage system, export the storage system configuration data.
   1. Log in to the command-line interface (CLI) of the storage system using PuTTY.

      The default user name is admin. To obtain its default password, see OceanStor Dorado Account List.

   2. Run export running_data to export and save the current configuration file.

4. Optional: If online replacement is used, you need to delete link information from the NMP multipath software.
   • If the HBA is replaced due to a failure, do the following:
     1. Run scsimgr lun_map -D /dev/rdisk/diskX to view disk path information, where diskX represents a disk on the host.
     2. Run ioscan -fnkC fc, ioscan -kfNC tgtpath, ioscan -P health -H hw_path, and rmsf -H hw_path to delete links, where hw_path indicates the path of the HBA to be replaced.
     3. Run sar -L 1 30 to monitor link I/Os.
     4. Run scsimgr lun_map -D /dev/rdisk/diskX to view disk path information, where diskX represents a disk on the host.

        The information about the deleted links should not appear in the command output.

   • If the HBA is replaced proactively, do the following:
     1. Run scsimgr lun_map -D /dev/rdisk/diskX to view disk path information, where diskX represents a disk on the host.
     2. Run ioscan -fnkC fc, scsimgr -f disable -H hw_path and ioscan -P health -H hw_path to disable links, where hw_path indicates the path of the HBA to be replaced.
     3. Run rmsf -H hw_path to delete all disabled links.
     4. Run sar -L 1 30 to monitor link I/Os.
     5. Run scsimgr lun_map -D /dev/rdisk/diskX to view disk path information. The information about the deleted links should not appear in the command output.

5. Optional: If the HBA is replaced online, you need to delete disk devices connected to the links.
   1. Run ioscan -fnkC disk and ioscan -fnkC disk | grep -i HUAWEI | wc -l to view system disk status.
   2. Run ioscan -fnkC fc and ioscan -fnk -C disk -H hw_path | grep -i HUAWEI | awk '{ print $3}' | xargs -n1 rmsf -C disk -H to delete disk devices connected to the links, where hw_path indicates the path of the HBA to be replaced.
   3. Run ioscan -fnkC disk and ioscan -fnkC disk | grep -i HUAWEI | wc -l again to check the system disk status, which should not contain the status of deleted disk devices.

6. Check whether the new HBA is compatible with the storage system.
   1. Run fcmsutil /dev/fcdY to check whether the WWN of the new HBA is as planned, where fcdY represents the new HBA.
   2. Run fcmsutil /dev/fcdY vpd to check whether the HBA is compatible with the storage system.

7. Check the system status and check whether there are disks or tapes connected to the HBA.
   1. Run tail -200 /var/adm/syslog/syslog.log to check the storage system status. If an error exists on the HBA or any disk, remove it and proceed to the next step.
   2. Run ioscan -fnkC fc and ioscan -fnkH HW Path to check whether there are disks or tapes connected to the new HBA, where HW Path indicates the path of the HBA to be replaced.

      There should be no disk or tape under fcdY.

8. On the switch, check whether the zoning of the new HBA is complete.

Prerequisites

• The old HBA has been replaced and the new HBA has been correctly installed.
• The WWNs of the old and new HBAs have been obtained.

Procedure

1. On the storage system, replace the old WWN.
   1. Run show initiator initiator_type=FC to view the connection status between the HBA and the storage device and obtain the ID of the host to which the old HBA belongs.

       
           WWN                 Running Status        Free       Alias      Host ID         Multipath Type 
           ----------------- ------------------- -----      -------    ----------      ------------------ 
           100000000000*       Online                Yes        --         --              Default

   2. Run remove host initiator initiator_type=FC to remove the WWN of the old HBA. Specify the WWN of the old HBA using the wwn parameter.
   3. Run create initiator fc wwn=? to create an FC initiator. wwn is the WWN of the new HBA.
   4. Run add host initiator host_id=? initiator_type=FC wwn=? to add the new WWN to the host. Specify the host and WWN of the new HBA using the host_id and wwn parameters.
   5. Run show initiator initiator_type=FC. If in the command output, the WWN status of the new HBA is not Free, then the replacement is successful.

2. Scan for disks on the host.
   1. Run ioscan -fnNkC disk and vgdisplay -v to show the disks and volume groups (VGs) on the host.
   2. Run ioscan -fnC disk to scan for disks.
   3. Rerun ioscan -fnkC disk and compare the result with that in 2.a. The number and the types of newly generated devices should be the same as planned.
   4. Run insf -eC disk and confirm that the LUN device files are generated.
   5. Rerun ioscan -fnkC disk and compare the result with that in 2.a. The number and the types of newly generated devices should be the same as planned.
   6. If the NMP software is installed on the host, do the following:
      1. Run scsimgr get_attr -a leg_mpath_enable and confirm that the NMP is enabled.
      2. Run ioscan -funC disk to view the number of disk paths when NMP is not enabled.
      3. Rerun ioscan -funNC disk and confirm that the number of newly generated aggregated disks is the same as planned.
      4. Run ioscan -m dsf to view the mappings between persistent disks and legacy disks.
      5. Run scsimgr lun_map -D /dev/rdisk/diskX to view disk path information with NMP enabled, where diskX represents a disk on the host.

3. Run sar -L 1 30 to monitor path I/Os.

   If I/Os are evenly distributed in all paths and there are no errors after 30 seconds, press Ctrl+C to exit.