HP-UX 11.31 has persistent Device Special File (DSF) Based on Storage node WWN port id & Lun ID. This behaviour is because of the SCSI stack discovers disk at a lunpath with same target id and LUN id but different LUN WWID. It sets the lunpath’s Last Open or Close (SPOC) state in “authentication failure” to prevent data corruption in case the change was not intentional, such as a backup of the original vdisk(LUN) using Business Copy. 

The SCSI stack also logs the following message to alert the system administrator. To validate disk replacement, run the ‘scsimgr replace_wwid’ command. 

This command has three options,any one of these can be used.  

•A) Validate the replacement for a specific lunpath      
•B) Validate the replacement for all paths of disk     
•C) Validate the replacement for all lunpaths beneath the target paths

---------------------------------------------------------------------------------
Note: Before proceeding with the replacement,determine the replace LUN hardware path by running the following commands after the authentication failure is logged:      

i. Determine the hardware path of the lunpath:
[formatted]
# scsimgr -p get_attr -C lunpath -I 111 -a hw_path   -->this will result the h/w path X.Y.Z.XXX 
[unformatted]
ii. Determine the hardware path of the replaced LUN   
[formatted]
# ioscan -m hwpath -H X.Y.Z.XXX
[unformatted]
---------------------------------------------------------------------------------


Validating disk replacement on a single lunpath


1) You can view the SPOC state of the lunpath in the log message by running the following command for the lunpath. Here we use lunpath 111 
[formatted]
# scsimgr get_info -C lunpath -I 111
[unformatted]
Look for "Last Open or Close state". You will see an authentication failure.

2)If so execute     
[formatted]
# scsimgr -f replace_wwid -C lunpath -I 111 
[unformatted]
Similarly you need to validate the remaining lunpath whatever appearing in the syslog,if you have more than one. 



Validate the replacement for all paths of disk


1)Look for the disks with "NO_HW" from the output of the command        
[formatted]
# ioscan -m lun

disk    118  64000/0xfa00/0x15   esdisk  NO_HW       DEVICE       offline  HP      HSV210                           0/3/0/0/0/0.0x50001fe1500c5dae.0x4011000000000000               0/3/0/0/0/0.0x50001fe1500c5da8.0x4011000000000000               0/3/0/0/0/1.0x50001fe1500c5da9.0x4011000000000000               0/3/0/0/0/1.0x50001fe1500c5dab.0x4011000000000000               0/3/0/0/0/0.0x50001fe1500c5dac.0x4011000000000000               0/3/0/0/0/1.0x50001fe1500c5daf.0x4011000000000000                        /dev/disk/disk118   /dev/rdisk/disk118  
[unformatted]
2)execute the below command. The disk device file with "NO_HW" in the example "/dev/rdisk/disk118"    
[formatted]
# scsimgr -f replace_wwid -D /dev/rdisk/disk118 dsf
[unformatted]
Since the new disk has a different WWID, the system assigns it a new instance and creates new DSFs. To avoid reconfiguring applications and upper layer modules (Volume Managers, file systems, etc.) using disk118, assign disk118 DSFs and instance to the new disk by specifying the ‘dsf’ keyword at the command line.

Repeat the steps for other disks, if you have more than one.

Validating disk replacement for all lunpaths under a target path


When a LUN is replaced or the SAN is reconfigured, several lunpaths under a given target can be put in “authentication failure” state. All lunpaths beneath each target path specified will be validated. Hence validating the replacement using 'tgtpath' needs to executed only few times. You would notice this tgtpath from the error message in syslog or dmesg.
[formatted]
#scsimgr -f replace_wwid -C tgtpath -I 3
[unformatted]
Similarly you need to validate the remaining tgtpath whatever appearing in the syslog,if you have.

Once the replacement is validated, the disk through the specified lunpath is accessible.

----------------------------------------------------------------
Note:'scsimgr replace_wwid' need to be run only if the system  was not rebooted.In other words a reboot of the system will also resolve this issue.
----------------------------------------------------------------