Problem with EMC Symmetrix

[Juan Luis Morcillo]

Hi!
People from EMC have installed this morning new disks on my Symmetrix 8530.
New disks are internally identified by ids greater than #FF, and they have explained to me that my MC/SG cluster of two nodes (A.11.12 on HP-UX 11) will see those disks as different devices from different nodes (different device files /dev/dsk/cxtxdx for each node).
That sounds strange to me, but they were right!
Any PATCH or walkaround, apart from reasignning disks accessed by more than one node to ids lower than #FF?
Thanks in advance!
Juan Luis

[Vince Inman]

What you are seeing is a result of the BIN file configuration within the Symm. The CE has configured the disks within your symmetrix to be at different addresses to the FA's (or SA's if SCSI attached).

This does not have to be the case for the Target/LUN id's. The CE has the ability to control this from the systemtrix configuration. The Controller ID's are host dependent.

To ensure that you are using the same devices for MC/SG, you will need to run an INQ or SYMINQ (Depending upon EMC s/w you have) from each host and match up VolID's to the applicable device files from the host.

As well, with MC/SG you simply define a VG on one server, then export a map file for that VG to the other MCSG host and perform an import without specifying the volumes to be imported. MC/SG will scan the bus looking for the same volumes, this may take longer than specifying specific volumes (depending upon the configuration), but it will in fact locate the devices as long as they've been presented to the host.

[Roger Baptiste · Accepted solution]

hi,

You should be able to see the Luns of the SYMM from the same paths on both nodes of a cluster. First, ensure the SYM lun ids are the same, as seen by both the nodes. This can be done by running syminq command.

Next, make sure the cards are located on similar numbered slots on both the nodes. Ioscan would help in figuring out this one. You can also physically check the back of the system. If the cards are not in the same slots, then definitely they would need to be placed in identical locations.

HTH
raj

[Juan Luis Morcillo]

Both ioscan and inq gives me correct information (correct path, correct SYM dev...) but one node has assigned c10txdy device files to disks, and the other node c11txdy to the same group!

Thanks!

Juan Luis

[Jeff Schussele]

Hi Juan,

Vince is correct - controller IDs are assigned by the host.
To be identical they have to be at the same HW path - which implies the same class (type) host AND in the same slot on both. There is nothing short of moving the adapter - IF you have the same type host - that you can do to change the controller ID.
But this is not insurmountable - you just need to edit any scripts you use between these hosts to add or modify these VG/LV definitions. This normally takes only minutes - provided you have the INQ and/or ioscan outputs at hand.

Rgds,
Jeff

[Joseph Nunan]

I went through the exact problem as a result of an incorrect bin file. I had the CE correct the file.