I have 2 HP DL585 servers with Emulex fibre channel cards that were originally running RHEL 5.3 and 5.2 with kernel 2.6.18-92.1.22.el5 connected to a san.
I have now patched these to RHEL 5.5, kernel 2.6.194-8.1.el5.
The san team have moved the cabling from a standard san switch to a fabric san switch and I have tried to configure multipath.
Since the cable move, the servers display 2 lines of:
shpchp:shpc_init: cannot reserve mmio region
then the boot halts with sda5 device or resource busy.
Returning the cables to the original switch and removing the multipath config allows the servers to boot.

So my questions are:
What is the shpchp error and is this causing the problem?
Why would changing the switch the cable plugs into cause the server to fail to boot with the device busy message?

Also, it's not easy to arrange downtime on these servers, so I don't know when I'll next have a chance to reboot them.

Since you are running RHEL contact Red Hat support

Thanks for the reply. I had logged it with Red Hat Support but do not have a resolution as yet.

The shpchp error is not relevant as it refers to PCI Hotplug.

On trying to move the servers across to the new fabric switches to collect some more diagnostics for Red Hat support, the servers now don't see the disks at all.