disk device files have changed

[KPS]

Hi,

We had a major problem this past week in our SAN where someone inadvertently took away some zones and actually deleted them. (This is an EMC Symmetrix by the way) We have 4 fibre paths to the SAN and during this issue 2 went away. When the devices were rezoned to both paths, we got our right numbers of devices back, but they showed up as different /dev/dsk/c#t#d#'s. We're wondering how we can get the original /dev/dsk/c#t#d#'s back as there will be bad implications with cluster, SAN Boot disk, and other things that we need to consider if we cannot? Simply moving the /etc/lvmtab out of the way and running vgscan is not an option.

Is insf -H an option here with this? Any suggestions would be greatly appreciated.

Thanks,
KPS

[Mridul Shrivastava]

- look at man ioconfig, You can edit the ioconfig file to reflect the old paths (test on a test system first)
- vgexport -s -p -m -f ... vgscan / vgimport -s -m
(the file from -f must be manually edited, or just use -s and be fine.)

[CharlesK]

2x Kludge =
(1)If you know the major/minor number then you could mknod the device files by hand...
(2)Setup a symbolic link using the name of the original device file pointing to the new device file...

What has changed to cause the hardware path to change..? A zone include/exclude should not change the Domain ID/Port Index in the hardware path... Has any re-cabling been undertaken..? Are the LUNs still presented on the same intefaces on the EMC..? Does the EMC still use the the same SCSI target and LUN ID for each LUN as before..? If all of these questions is YES the hardware path SHOULD not change...

What type of servers are they PA/Itanium, sx[12]000..?
OS Version ..?!?!

[KPS]

Charles,

System type we're dealing with is an RX8640 and we're dealing with a whole cell running in nPar mode.

I'm not aware of any cabling that took place with this problem and it's resolution. The LUNS are still presented on the same interface on the EMC FA's. This was put back with the rezoning. The SCSI Target for each LUN ID may have changed, I'd have to go dig for that with our SAN Team, but I'm not sure right now.

One thing to add is that I could not cleanup PowerPath properly after this issue. I'm starting to wonder if PowerPath, since it still shows dead paths down 2 of the paths is holding on to these c#t#d#'s and if we reboot, we'll get them back since the reboot will clean them up?

[KPS]

ioconfig will be the way we go about changing these /dev/dsk/c#t#d#'s back.

Thanks to all!

/KPS