On Tue, Feb 21, 2017 at 5:42 PM, Andrea Ghelardi <a.ghelardi@iontrading.com> wrote:

Hello Gianluca,

I have ISCSI SAN here and not FC but process should be very similar.

Correct operation is

Storage maintenance -> detach -> remove

 

Once you have removed the storage, Ovirt will release the sanlock.

Note that your server will still “see” the device and multipath will still list it as active.

 

You can now unmap LUN using your SAN manager.

If your multipath is correctly configured, device will “fail” but nothing else (multipath and server won’t hang).

You will be able to perform basic list operations via server shell without experiencing freeze or locks.*


Yes, but /var/log/messages fills up with multipath errors.
This is not desirable
 

Ovirt will continue to scan new devices so you won’t be able to manually remove them until you unmap devices from SAN manager.

 

If you need to remove devices manually, it is advised to follow this guide.

https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Storage_Administration_Guide/removing_devices.html

 

* If you do face command freeze etc, re-enable LUN mappings and check your multipath.conf


 Yes, but the problem is that I receive error for the LUN previously configured as storage domain.
I presume sum stale devices of kind "dm-??" have remained somewhere.

I could reboot one ot the 3 hosts and verified that now the layout is correct on it.
But I can't easily reboot just now the remaining 2 hosts, so it would be better to find what forces the multipath device in use....

Gianluca