RE: [Iscsitarget-devel] IET Persistent Reserve
Nicholas A. Bellinger
> I'll give it a try.
>
> Can someone give me a few pointers on how to build it properly?
http://linux-iscsi.org/svn/trunk/README
>
> My test environment is Dell servers connected to an EMC CX-500 via FC. I'm running VMWare ESX on two of the nodes and have a few LUNs mapped directly to the virtual machine I want to be the iSCSI target host.
>
This is where a problem might arise. My experience with Vmware (at
least with Workstation 6) with bringing in storage LUNs into the guest
is that VMware still emulates the control CDBs. For example, the SCSI
Inquiry EVPD pages (that tell you things like the unit serial number of
the storage hardware) returns emulated, and not proper values for the
control CDBs. This has been the case with Workstation 6 on Linux and
Win32, but I seem to recall Fusion on MacOSX passing CDBs, at least USB
storage devices, without the control emulation bit.
> My goal is to have a RHEL5 VM run the iSCSI target software and support the PERSISTENT RESERVATION commanded needed for Longhorn.
>
There are binaries for RHEL5 online @ http://linux-iscsi.org/builds/el5
> Will this work? Or do I need to run the Linux iSCSI target on real hardware with a real HBA?
>
This depends if ESX server does the same emulation of control CDBs for
real SCSI devices passed into the guests as Workstation 6 does. If it
does, then you will most likely hit an exception. If the INQUIRY
information that is returned in the guest is the same as what appears on
the host, meaning that no control CDB emulation is happening, and since
you are using FC hardware, it may just work with the current code in
target/iscsi_target_pscsi.c
In any event, I am very interested in making this generic for all of the
storage objects in the engine as certainly this item will continue to
come up again. It would be great if I could get a list of which
clustering storage currently uses this functionality, and then I can go
about getting a test environment setup on the Linux-iSCSI.org hardware
for the work, and then add proper generic support. If anyone could give
me an idea of which clustering stuff uses this functionality (aside from
the Longhorn mentioned above) please contact me offline or on the LIO
Target Dev list. If someone wanted to cut me a VM image for initiator
side code that does this, even better. Speaking of which, is there any
Linux based cluster code that uses Persistent Reserve following SPC-3..?
Also, for future discussions regarding the LIO Target specifics please
following Ming's request to post on the LIO Target dev list. I usually
try to only post items to IET-dev that are generic information for all
Linux/iSCSI and which could help other projects, devs, and users.
--nab
> Thanks,
>
> Dave
>
> ________________________________________
> From: Ming Zhang [blackma...@gmail.com]
> Sent: Saturday, November 17, 2007 11:20 AM
> To: Nicholas A. Bellinger
> Cc: David R. Bjurman - Birr; iscsitar...@lists.sourceforge.net
> Subject: RE: [Iscsitarget-devel] IET Persistent Reserve
>
> On Fri, 2007-11-16 at 20:28 -0800, Nicholas A. Bellinger wrote:
> > On Fri, 2007-11-16 at 20:28 -0500, Ming Zhang wrote:
> > > On Fri, 2007-11-16 at 15:20 -0700, David R. Bjurman - Birr wrote:
> > > > It looks like these commands are mandatory to enable disks for use in Windows 2008 failover clustering (MSFT requirement).
> > > >
> > > > Does anyone on this list know how to work around this?
> > >
> > > no way to work around. it is a new scsi command support code needs to be
> > > written.
> > >
> > > @Nab, did your iscsi target solution has this implemented?
> > >
> > >
> >
> > This is available today with target/iscsi_target_pscsi.c for genuine
> > SCSI hardware (SAS/PSCSI/FC) in the current beta v2.9 release. There
> > may be SATA hba vendor driver or two as well that supports it.
> >
> > Note that the v2.9 storage engine does not support this generically for
> > all SE -> LUN object maps the our current release, but I have been
> > wanting to have another look at this item for some time. :-)
>
>
> thx for the info.
>
> so anyone who want PR support can go ahead and check linux-iscsi-target.
> of course, questions regarding that target should go to related list,
> not this one.
>
>
>
>
>
>
> >
> >
> > --nab
> >
> > > >
> > > > Thanks,
> > > >
> > > > Dave
> > > >
> > > > ________________________________________
> > > > From: Ming Zhang [blackma...@gmail.com]
> > > > Sent: Friday, November 16, 2007 3:05 PM
> > > > To: David R. Bjurman - Birr
> > > > Cc: iscsitar...@lists.sourceforge.net
> > > > Subject: Re: [Iscsitarget-devel] IET Persistent Reserve
> > > >
> > > > On Fri, 2007-11-16 at 14:45 -0700, David R. Bjurman - Birr wrote:
> > > > > Are there are plans to include iSCSI commands 5E and 5F (PERSISTENT
> > > > > RESERVE IN / PERSISTENT RESERVE OUT) in IET?
> > > > >
> > > >
> > > > it is scsi command. not iscsi command
> > > >
> > > >
> > > > no such plan, unfortunately, at least from my side.
> > > >
> > > >
> > > > >
> > > > >
> > > > > Thanks,
> > > > >
> > > > >
> > > > >
> > > > > David
> > > > >
> > > > >
> > > > > -------------------------------------------------------------------------
> > > > > This SF.net email is sponsored by: Microsoft
> > > > > Defy all challenges. Microsoft(R) Visual Studio 2005.
> > > > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> > > > > _______________________________________________ Iscsitarget-devel mailing list Iscsitar...@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/iscsitarget-devel
> > > > --
> > > > Ming Zhang
> > > >
> > > >
> > > > @#$%^ purging memory... (*!%
> > > > http://blackmagic02881.wordpress.com/
> > > > http://www.linkedin.com/in/blackmagic02881
> > > > --------------------------------------------
> >
> --
> Ming Zhang
>
>
> @#$%^ purging memory... (*!%
> http://blackmagic02881.wordpress.com/
> http://www.linkedin.com/in/blackmagic02881
> --------------------------------------------
>
Nicholas A. Bellinger
> On Sat, 2007-11-17 at 12:06 -0700, David R. Bjurman - Birr wrote:
> > I'll give it a try.
> >
> > Can someone give me a few pointers on how to build it properly?
> >
> >
> >
>
So as mentioned in the last mail, iSCSI Target mode from within a Linux
VM has been tested with Workstation 6 on Linux/win32 hosts and with
Fusion on a MacOSX host and appears to be stable. I was even able to
export out a USB optical drive from Linux VM on a non Linux host and
watch Transformers HD-DVD. :-)
As to what the performance overhead will be in real world cases, I have
not taken any numbers using the VMware passthrough interface just yet.
As Linux-iSCSI.org does have a VMware Workstation 6 license, I can put
together another box with some decently quick storage hardware that can
be used exclusively for this purpose. Also, if there are virtualization
vendors about who want to see their tech running on Linux-iSCSI.org
hardware for testing, validation and developer access purposes, I would
be more than happy to talk with them.
Ming Zhang
> On Sat, 2007-11-17 at 14:41 -0500, Ming Zhang wrote:
> > On Sat, 2007-11-17 at 12:06 -0700, David R. Bjurman - Birr wrote:
> > > I'll give it a try.
> > >
> > > Can someone give me a few pointers on how to build it properly?
> > >
> > > My test environment is Dell servers connected to an EMC CX-500 via FC. I'm running VMWare ESX on two of the nodes and have a few LUNs mapped directly to the virtual machine I want to be the iSCSI target host.
> > >
> > > My goal is to have a RHEL5 VM run the iSCSI target software and support the PERSISTENT RESERVATION commanded needed for Longhorn.
> > >
> > > Will this work? Or do I need to run the Linux iSCSI target on real hardware with a real HBA?
> >
> > should work. though unclear about performance.
>
> So as mentioned in the last mail, iSCSI Target mode from within a Linux
> VM has been tested with Workstation 6 on Linux/win32 hosts and with
> Fusion on a MacOSX host and appears to be stable. I was even able to
> export out a USB optical drive from Linux VM on a non Linux host and
> watch Transformers HD-DVD. :-)
thx for confirmation.
>
> As to what the performance overhead will be in real world cases, I have
> not taken any numbers using the VMware passthrough interface just yet.
> As Linux-iSCSI.org does have a VMware Workstation 6 license, I can put
> together another box with some decently quick storage hardware that can
> be used exclusively for this purpose. Also, if there are virtualization
> vendors about who want to see their tech running on Linux-iSCSI.org
> hardware for testing, validation and developer access purposes, I would
> be more than happy to talk with them.
maybe David have enough interest to test this out before running it in
his environment...