Environment:

DELL EMC SW : Powerpath/VE (any release)
VMWare SW: ESXi (any release)
DELL EMC HW : any storage
 

Issue

When maintenance is needed on a SAN or even on an array, disabling a path prior to bringing down a path might be needed to avoid elongated IOs (due to timeouts), or to just avoid IOs errors while PowerPath is constantly probing the path as per "autorestore" policy.

Resolution:

To better explain the solution, here is an example :

dev=emcpower1
Pseudo name=emcpower1
XtremIO ID=FNMXXXXXXXXXXX [DDC-TEST]
Logical device ID=51xxxxxxxxxxxxx
Standard UID=naa.514f0c5234c00000 [DDC-TEST_000]
type=Conventional;  state=alive; policy=ADaptive; queued-IOs=0
==============================================================================
--------------- Host ---------------   - Stor -  -- I/O Path --   -- Stats ---
###  HW Path               I/O Paths    Interf.  Mode     State   Q-IOs Errors
==============================================================================
   2 vmhba0                 C0:T1:L19  X1-C2-f1  active   alive      0      6
   2 vmhba0                 C0:T0:L19  X2-C1-f1  active   alive      0      1
   1 vmhba1                 C0:T1:L19  X2-C2-f2  active   alive      0      0
   1 vmhba1                 C0:T0:L19  X1-C1-f2  active   alive      0      0 


1- Determine which path that wants to be isolated. Here, we want to isolate the first path and we can see that it already recorded errors but not frequently enough to be in asb:iopf.  

2- Run the following command : rpowermt set mode=standby dev=vmhba0:C0:T1:L19. Instead of active/alive, we will have standby:alive 

The above command needs to be run against every device connected to this path. Of course, not to forget any device, a script is advised based on the output of a "rpowermt display dev=all".

In the output of "rpowermt display ports", the path will be found as alive or active, this is normal since the path is still active but not sending any IOs.