Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

1619

March 3rd, 2009 03:00

symmask -dynamic_lun: how to determine which device is assigned which LUN?

Hi everybody,

I'm trying out Dynamic LUN Addressing using the "-dynamic_lun" option to the symmask command. It works just fine, but I do have some questions about it.

To start, I'd like to know if there's a way to display which Host LUN ID a device was assigned. The output of "symdev -sid xxx show yyy" does have a Host LUN column, but that shows "N/A" even though the host itself reports a different LUN ID for that device:

    Front Director Paths (2):
        {
        ----------------------------------------------------------------------
                                 POWERPATH  DIRECTOR   PORT             LUN
                                 --------- ----------  ---- -------- ---------
        PdevName                 Type      Type Num    Sts  VBUS TID SYMM Host
        ----------------------------------------------------------------------
        Not Visible              N/A       FA   08A:0  RW   000  00  0EB  N/A
        Not Visible              N/A       FA   09A:0  RW   000  00  0EB  N/A
        }


Also, "symcfg -sid xxx -dir 8a -p 0 -address list" only shows the Symm LUN ID, not the Host LUN ID. Is there a way to display the current Host LUN assignment of a device?

Further, I'd like to know exactly at which point the Host LUN assignment is determined, because I'm a bit worried about it changing under certain circumstances. For my tests, I've assigned devices 071-078 to a server over two paths. Those were assigned Host LUN ids of 0-7, as you would expect. Then I removed device 072, which caused the host to lose access to Host LUN 1 (as expected). I then removed device 78, and immediately assigned it back. This caused the Host LUN ID to change from 7 to 1. Again, this was expected because removing device 72 left Host LUN ID 1 free for use.

Are there other circumstances under which a Host LUN ID assignment would change? Suppose that I remove device 78 again; this would leave a gap in the Host LUN ID numbering. What would happen for example, when the host is switched off, and back on again? In that case, it would logout and login to the Symm. Would this cause a renumbering of the Host LUN IDs? Or when the FA board is replaced?

All in all, this feature looks very useful. If my worries about Host LUN ID numbering turn out to be a non-issue, it would be even better. :)

Regards,

Jurjen Oskam

11 Posts

March 3rd, 2009 10:00

Since the host address is set per the WWN of the initiator, you can't get to the host LUN ID of the device through symdev. You need to use something similar to:

symmaskdb -sid SYMID list devs -wwn WWNOFHBA

Which should dump all the masked devices for that WWN, which will include the SYMM and the HOST LUN IDs.

111 Posts

March 5th, 2009 02:00

Hi Jurjen,

If you want the symdev command to show you the Host LUN ID field, you need to issue the symdev command directly from the host the dev is masked to.
*Try doing symcfg discover before*
The symdev output should also change the Pdevname from "Not Visible" to the exact pv in the host OS.

As for the second question, I think it depends on which OS you use. But basically, you can use persistant binding. Again it differs between OS and HBAs.

Maxim

April 9th, 2009 14:00

Hi Jurjen,

as for the second question (the first is answered sufficiently):
The Host LUN ID (or HLU as opposed to ALU for array LUN ID) is set by the Symmetrix, and was introduced to get around host LUN addressing issues. There are OSs that can only deal with LUN ID <= 255 (e.g. Windows) or worse. Thus the Symmetrix adds another mapping from HLU to ALU when addressing LUNs in Host I/Os. These things do not change, but the Symmetrix will be told to use the lowest possible HLU if you use -dynamic_lun. Thus your scenario is fully expected. Using -lun you can specify the HLU yourself.

These HLUs do not change between reboots at all, since they are dictated at the Symmetrix level: if an initiator does a device discovery, the Symmetrix will specify the HLUs if they are available (for that WWPN), or else the ALU.

Similarly, replacing an FA board never causes ALUs to vanish, so you should expect the ALUs to survive FA board replacements. Normal, as ALUs and HLUs are effectively configuration attributes of a device masking entry, and that device masking entry is on the FA board, but also in the internal system configuration of the Symmetrix.

Hope this explains it a bit,
Guus
No Events found!

Top