Hi,
very tricky idea :) anyway I feel it's a bit too a complicated setup to 
manage...
What I still don't get, is why zfs internals have no way to get back from a 
call to a failing driver, and
block any call to that filesystem.
Why can't zfs internals detect the failing driver and place it in a 
"black-list", so to put the pool as
unavailable?
In these situations, there's no way to shutdown the machine but abruptly 
(...after checking that
there is no more disk activity...)
Gabriele.
----------------------------------------------------------------------------------
Da: Jim Klimov
A: Gabriele Bulfon
Data: 23 novembre 2012 12.53.38 CET
Oggetto: Re: [discuss] Again: illumos based ZFS storage failure
Hello, on a very side note - do you have enough capacity to try
serving the iSCSI volume from the second pool, or perhaps to
share over NFS a virtual disk image of the same size as zvol for
iSCSI, and mounting that into VMWare (as an NFS storage resource)
and mirroring with virtual Windows PDC itself? This way, the one
locking-up half of the mirror might be not able to kill the whole
PDC... Perhaps you might be able to kill the iSCSI initiator or
target "on the fly", thus forcing detectable disconnection of a
mirror half, so the retry loops (if any) from Windows get aborted.
Does reboot/poweroff of the PDC VM (or VMWare host) if possible
unlock your storage pool?
Also, your dump device is not compressed, right?
//Jim



-------------------------------------------
illumos-discuss
Archives: https://www.listbox.com/member/archive/182180/=now
RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4
Powered by Listbox: http://www.listbox.com

Reply via email to