Xen devel - Apr 2010 - Xen 3.0.3 tap:aio disk image fails on RAID5/XFS filesystem

Hi all,

Upon bashing my head against a wall trying to get Xen working on my new
hardware I came across an interesting discovery. Every install using
virt-install was failing and I could only use the file: access method for
my DomUs. Every time I tries to use tap:aio the guest would error on boot
saying it couldn''t find the root partition.

My system is as follows:
OS: CentOS 5.4
Xen: 3.0.3-94.el5_4.3
/ = 2 x 80Gb HDDs, dmraid1, ext3
/mnt/raid = 3 x 1Tb HDDs, dmraid5, xfs

On whim, I moved the DomU images from /mnt/raid/vm-images to /vm-images
and instantly tap:aio worked again.

This brings me to my question:

Why would tap:aio fail when the images are on an XFS/RAID5 filesystem but
work correctly when on a RAID1/ext3 filesystem?

To me, this seems like a bug.

-- 
Steven Haigh
 
Email: netwiz@crc.id.au
Web: http://www.crc.id.au
Phone: (03) 9001 6090 - 0412 935 897
Fax: (03) 8338 0299

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Tue, 27 Apr 2010 14:01:17 +1000, Steven Haigh <netwiz@crc.id.au>
wrote:
> Hi all,
> 
> Upon bashing my head against a wall trying to get Xen working on my new
> hardware I came across an interesting discovery. Every install using
> virt-install was failing and I could only use the file: access method
for
> my DomUs. Every time I tries to use tap:aio the guest would error on
boot
> saying it couldn''t find the root partition.
> 
> My system is as follows:
> OS: CentOS 5.4
> Xen: 3.0.3-94.el5_4.3
> / = 2 x 80Gb HDDs, dmraid1, ext3
> /mnt/raid = 3 x 1Tb HDDs, dmraid5, xfs
> 
> On whim, I moved the DomU images from /mnt/raid/vm-images to /vm-images
> and instantly tap:aio worked again.
> 
> This brings me to my question:
> 
> Why would tap:aio fail when the images are on an XFS/RAID5 filesystem
but
> work correctly when on a RAID1/ext3 filesystem?
> 
> To me, this seems like a bug.
Oh, and for the record, this is my config file:

name = "mail.crc.id.au"
uuid = "929c5a29-10c2-b388-ff01-42110c4ea66e"
maxmem = 512
memory = 512
vcpus = 2
bootloader = "/usr/bin/pygrub"
on_poweroff = "destroy"
on_reboot = "restart"
on_crash = "restart"
#disk = [ "file:/vm-images/mail.crc.id.au.img,xvda,w" ]
disk = [ "tap:aio:/vm-images/mail.crc.id.au.img,xvda,w" ]
vif = [ "mac=00:16:3E:00:00:13, bridge=virbr0" ]

-- 
Steven Haigh
 
Email: netwiz@crc.id.au
Web: http://www.crc.id.au
Phone: (03) 9001 6090 - 0412 935 897
Fax: (03) 8338 0299

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Tue, Apr 27, 2010 at 05:19:54PM +1000, Steven Haigh
wrote:
> 
> On Tue, 27 Apr 2010 14:01:17 +1000, Steven Haigh <netwiz@crc.id.au>
wrote:
> > Hi all,
> > 
> > Upon bashing my head against a wall trying to get Xen working on my
new
> > hardware I came across an interesting discovery. Every install using
> > virt-install was failing and I could only use the file: access method
> for
> > my DomUs. Every time I tries to use tap:aio the guest would error on
> boot
> > saying it couldn''t find the root partition.
> > 
> > My system is as follows:
> > OS: CentOS 5.4
> > Xen: 3.0.3-94.el5_4.3
> > / = 2 x 80Gb HDDs, dmraid1, ext3
> > /mnt/raid = 3 x 1Tb HDDs, dmraid5, xfs
> > 
> > On whim, I moved the DomU images from /mnt/raid/vm-images to
/vm-images
> > and instantly tap:aio worked again.
> > 
> > This brings me to my question:
> > 
> > Why would tap:aio fail when the images are on an XFS/RAID5 filesystem
> but
> > work correctly when on a RAID1/ext3 filesystem?
> > 
> > To me, this seems like a bug.
> 
> Oh, and for the record, this is my config file:
> 
> name = "mail.crc.id.au"
> uuid = "929c5a29-10c2-b388-ff01-42110c4ea66e"
> maxmem = 512
> memory = 512
> vcpus = 2
> bootloader = "/usr/bin/pygrub"
> on_poweroff = "destroy"
> on_reboot = "restart"
> on_crash = "restart"
> #disk = [ "file:/vm-images/mail.crc.id.au.img,xvda,w" ]
> disk = [ "tap:aio:/vm-images/mail.crc.id.au.img,xvda,w" ]
> vif = [ "mac=00:16:3E:00:00:13, bridge=virbr0" ]
> 
Since you''re using EL5 distribution you might want to search Redhat
bugzilla
to see if there are already existing bugreports about this issue.

-- Pasi


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Tue, 27 Apr 2010 12:00:54 +0300, Pasi Kärkkäinen <pasik@iki.fi>
wrote:
> On Tue, Apr 27, 2010 at 05:19:54PM +1000, Steven Haigh wrote:
>> 
>> On Tue, 27 Apr 2010 14:01:17 +1000, Steven Haigh
<netwiz@crc.id.au>
>> wrote:
>> > Hi all,
>> > 
>> > Upon bashing my head against a wall trying to get Xen working on
my
new
>> > hardware I came across an interesting discovery. Every install
using
>> > virt-install was failing and I could only use the file: access
method
>> for
>> > my DomUs. Every time I tries to use tap:aio the guest would error
on
>> boot
>> > saying it couldn''t find the root partition.
>> > 
>> > My system is as follows:
>> > OS: CentOS 5.4
>> > Xen: 3.0.3-94.el5_4.3
>> > / = 2 x 80Gb HDDs, dmraid1, ext3
>> > /mnt/raid = 3 x 1Tb HDDs, dmraid5, xfs
>> > 
>> > On whim, I moved the DomU images from /mnt/raid/vm-images to
/vm-images
>> > and instantly tap:aio worked again.
>> > 
>> > This brings me to my question:
>> > 
>> > Why would tap:aio fail when the images are on an XFS/RAID5
filesystem
>> but
>> > work correctly when on a RAID1/ext3 filesystem?
>> > 
>> > To me, this seems like a bug.
>> 
>> Oh, and for the record, this is my config file:
>> 
>> name = "mail.crc.id.au"
>> uuid = "929c5a29-10c2-b388-ff01-42110c4ea66e"
>> maxmem = 512
>> memory = 512
>> vcpus = 2
>> bootloader = "/usr/bin/pygrub"
>> on_poweroff = "destroy"
>> on_reboot = "restart"
>> on_crash = "restart"
>> #disk = [ "file:/vm-images/mail.crc.id.au.img,xvda,w" ]
>> disk = [ "tap:aio:/vm-images/mail.crc.id.au.img,xvda,w" ]
>> vif = [ "mac=00:16:3E:00:00:13, bridge=virbr0" ]
>> 
> 
> Since you''re using EL5 distribution you might want to search
Redhat
> bugzilla
> to see if there are already existing bugreports about this issue.
> 
I did have a hunt around - but I couldn''t locate anything relevant -
hence
my post here to see if it is anything that is already known - and maybe
documented somewhere...

-- 
Steven Haigh
 
Email: netwiz@crc.id.au
Web: http://www.crc.id.au
Phone: (03) 9001 6090 - 0412 935 897
Fax: (03) 8338 0299

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Tue, 27 Apr 2010 19:48:01 +1000, Steven Haigh <netwiz@crc.id.au>
wrote:
> On Tue, 27 Apr 2010 12:00:54 +0300, Pasi Kärkkäinen <pasik@iki.fi>
wrote:
>> On Tue, Apr 27, 2010 at 05:19:54PM +1000, Steven Haigh wrote:
>>> 
>>> On Tue, 27 Apr 2010 14:01:17 +1000, Steven Haigh
<netwiz@crc.id.au>
>>> wrote:
>>> > Hi all,
>>> > 
>>> > Upon bashing my head against a wall trying to get Xen working
on my
> new
>>> > hardware I came across an interesting discovery. Every install
using
>>> > virt-install was failing and I could only use the file: access
method
>>> for
>>> > my DomUs. Every time I tries to use tap:aio the guest would
error on
>>> boot
>>> > saying it couldn''t find the root partition.
>>> > 
>>> > My system is as follows:
>>> > OS: CentOS 5.4
>>> > Xen: 3.0.3-94.el5_4.3
>>> > / = 2 x 80Gb HDDs, dmraid1, ext3
>>> > /mnt/raid = 3 x 1Tb HDDs, dmraid5, xfs
>>> > 
>>> > On whim, I moved the DomU images from /mnt/raid/vm-images to
> /vm-images
>>> > and instantly tap:aio worked again.
>>> > 
>>> > This brings me to my question:
>>> > 
>>> > Why would tap:aio fail when the images are on an XFS/RAID5
filesystem
>>> but
>>> > work correctly when on a RAID1/ext3 filesystem?
>>> > 
>>> > To me, this seems like a bug.
>>> 
>>> Oh, and for the record, this is my config file:
>>> 
>>> name = "mail.crc.id.au"
>>> uuid = "929c5a29-10c2-b388-ff01-42110c4ea66e"
>>> maxmem = 512
>>> memory = 512
>>> vcpus = 2
>>> bootloader = "/usr/bin/pygrub"
>>> on_poweroff = "destroy"
>>> on_reboot = "restart"
>>> on_crash = "restart"
>>> #disk = [ "file:/vm-images/mail.crc.id.au.img,xvda,w" ]
>>> disk = [ "tap:aio:/vm-images/mail.crc.id.au.img,xvda,w" ]
>>> vif = [ "mac=00:16:3E:00:00:13, bridge=virbr0" ]
>>> 
>> 
>> Since you''re using EL5 distribution you might want to search
Redhat
>> bugzilla
>> to see if there are already existing bugreports about this issue.
>> 
> 
> I did have a hunt around - but I couldn''t locate anything relevant
-
hence
> my post here to see if it is anything that is already known - and maybe
> documented somewhere...
For what it''s worth, a bit more digging turned up this:
https://bugzilla.redhat.com/show_bug.cgi?id=217098

As I''m not quite cluey enough to know exactly what was discovered in
this
bug report, I can''t say for sure if it is related - however I have
filed
this: 
https://bugzilla.redhat.com/show_bug.cgi?id=586387

I believe I have tagged this correctly as a kmod-xfs bug - as I don''t
believe it is a problem with Xen as such - however any hints /
contributions to the bug report would be appreciated.

-- 
Steven Haigh
 
Email: netwiz@crc.id.au
Web: http://www.crc.id.au
Phone: (03) 9001 6090 - 0412 935 897
Fax: (03) 8338 0299

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel