DM-Crypt Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [dm-crypt] shared block devices
@ 2011-03-30 17:29 Michael Zabriskie
  2011-03-30 21:03 ` Milan Broz
  0 siblings, 1 reply; 2+ messages in thread
From: Michael Zabriskie @ 2011-03-30 17:29 UTC (permalink / raw)
  To: dm-crypt

By what I've been able to find online and what is in the man page for
cryptsetup:

"--non-exclusive
    This option is ignored. Non-exclusive access to the same block
device can cause data corruption thus this mode is no longer supported
by cryptsetup."

I have been able to gather that a shared block device type setup is not
supported. I.E. luks + clvm + gfs2, luks + asm, or luks + ocfs2 shared
across multiple servers. So what I am wondering is if this is on the
road map or if there is another open source technology out there that
can accomplish this?

Thanks.
-Michael

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [dm-crypt] shared block devices
  2011-03-30 17:29 [dm-crypt] shared block devices Michael Zabriskie
@ 2011-03-30 21:03 ` Milan Broz
  0 siblings, 0 replies; 2+ messages in thread
From: Milan Broz @ 2011-03-30 21:03 UTC (permalink / raw)
  To: Michael Zabriskie; +Cc: dm-crypt


On 03/30/2011 07:29 PM, Michael Zabriskie wrote:
 
> I have been able to gather that a shared block device type setup is not
> supported. I.E. luks + clvm + gfs2, luks + asm, or luks + ocfs2 shared
> across multiple servers. So what I am wondering is if this is on the
> road map or if there is another open source technology out there that
> can accomplish this?

If you have LUKS device and mapped (plaintext) device is exported
through iSCSI/DRBD or whatever to several cluster nodes, it will work
(and use some clustered fs, or maybe clvmd on top of that).
(IOW encryption run only on some master server.)

What you cannot do is to map the underlying device on several nodes
and run LUKS on every node separately (in parallel).
(In principle, with proper barrier/flush support and clustered
fs on top it,  it should work but nobody tests and support these
configurations.) 

Or you have to open exactly on one node and migrate it as service
(similar to HA LVM mode).

(That --non-exclusive option was to map one underlying device to
several mappings on _one_ system, nothing to do with clustered
system.)

Anyway, what is the use case here?

Milan

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2011-03-30 21:03 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-03-30 17:29 [dm-crypt] shared block devices Michael Zabriskie
2011-03-30 21:03 ` Milan Broz

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox