On Fri, Nov 26, 2010 at 05:41:23PM +0100, Iustin Pop wrote:
> On Fri, Nov 26, 2010 at 04:30:06PM +0000, Maciej Bliziński wrote:
> > Here's a patch that allows to install Red Hat based systems, for example
> > Oracle
> > Linux. Tested with OEL.
> >
> > The hypervisor by default offers a device named 'sda'. If the SCSI module
> > is
> > already loaded, the disk device can't be created due to naming conflict, and
> > the disk is not available. A workaround is to modify the initrd by removing
> > the scsi driver from it. This helps, but doesn't allow to install the OS.
> >
> > Red Hat's installer, anaconda, runs parted, which tries to execute a check
> > against /dev/sda and fails. This makes anaconda think that the disk is
> > faulty,
> > and not available. The best way to work around this, is to declare 'xvda'
> > as
> > the xen disk device.
> >
> > This patch allows to pass -H xen-pvm:blockdev_prefix="xvd" and successfully
> > run
> > the Red Hat installer.
>
> The patch looks good overall, but this bit worries me:
>
> > @@ -639,13 +642,20 @@ class XenHvmHypervisor(XenHypervisor):
> > vif_data.append("'%s'" % nic_str)
> >
> > config.write("vif = [%s]\n" % ",".join(vif_data))
> > - disk_data = cls._GetConfigFileDiskData(block_devices)
> > +
> > + blockdev_prefix = hvp[constants.HV_BLOCKDEV_PREFIX]
> > + # A special case for HT_DISK_IOEMU; if the blockdev prefix hasn't been
> > + # specified, use "ioemu:hd".
> > + default_blockdev = (
> > + constants.HVC_DEFAULTS[HT_XEN_HVM[constants.HV_BLOCKDEV_PREFIX]])
> > disk_type = hvp[constants.HV_DISK_TYPE]
> > - if disk_type in (None, constants.HT_DISK_IOEMU):
> > - replacement = ",ioemu:hd"
> > - else:
> > - replacement = ",hd"
> > - disk_data = [line.replace(",sd", replacement) for line in disk_data]
> > + disk_is_ioemu = disk_type in (None, constants.HT_DISK_IOEMU)
> > + prefix_is_default = blockdev_prefix == default_blockdev
> > + if (prefix_is_default and disk_is_ioemu):
> > + blockdev_prefix = "ioemu:hd"
> > +
> > + disk_data = cls._GetConfigFileDiskData(block_devices, blockdev_prefix)
>
> Basically we do allow partial customization, but not complete. There's
> no way to enforce the default (what if we specifically set it to
> default?). I wonder if it makes sense at all to allow customization for
> HVM, and not (like before) use hd for non-ioemu and ioemu:hd for ioemu.
Another alternative would be to say that disk type ioemu always gets
ioemu: prefix, in other words the blockdev_prefix is modified for ioemu
to be ioemu:$blockdev_prefix. In this case the behaviour would be
consistent.
Hrmm. Actually reading the current Xen documentation the ioemu prefix is
no longer needed, since 2006 :) Maybe we should just drop the ioemu
stuff, and leave it to be configured like PVM, but with 'hd' the default
prefix?
iustin
