Product and version:
Storage system: S5500T&S5600T&S5800T&S6800T V100R001、V100R002
Host system: AIX 5300-05
Host HBA board: IBM 5758
Host HBA driver version: df1000fd (the system’s own)
The operation steps are as follow:
1. The host system is normal, the disk array system is normal, the physical link is normal, take the multiple paths SAN networking (start the failover)
2. Execute the read write operation for the multiple virtual LUNs via the “dd”
3. Plug out the member disk of the RAID group, make the RAID gorup’s LUN ineffective (including the LUN whose hosted is 0)
4. The corresponding virtual disk become ineffective, the I/O of the ineffective disk has disconnected, the “dd” command has been hang, and we can’t kill the “dd” process.
5. After recovering the RAID group, execute the command “dd” to read and write the recovered LUN again, but it has hang up again, execute the “lsdev” or “lspath” , it can’t take any effect, and we can’t kill the “dd” process.
Via consulting the IBM office, they admit this is a known bug of the AIX MPIO framework, recently, only the versions of the AIX operating system from the 6100-04 on has settled this problem.
It’s a third part problem, we can settle it only by upgrading the operating system to the 6100-04 or more advanced.