Re: Linux vs. NetBSD disk enumeration

[christos%astron.com@localhost (Christos Zoulas)]

In article <4F3465C1.4000401%tastylime.net@localhost>,
Jeff Rizzo  <riz%tastylime.net@localhost> wrote:
>On 2/9/12 10:45 AM, Manuel Bouyer wrote:
>> Yes, it's because drives are numbered in the order they're found.
>> We have the same issue with e.g. hot-swap bays; if you insert a new drive
>> in the middle of a bay the drives can be renumbered on next boot.
>
>Yes, I'm aware of that.
>
>> /dev/sda1 will be translater to the linux's block device number. This 
>> is what xen uses as 'id'. In my config files I don't use linux's 
>> devices names but 0x0, 0x1, 0x2, .... Up to now I've never seen a 
>> NetBSD kernel probe xbd is random order: the xen tools create the 
>> entries in the xenstore in the domains's config file order, and the 
>> domU kernel reads the xenstore sequentially. 
>
>Does this block device number remain consistent every time?  Is there 
>any way we can use it in the kernel as a locator for xbd*?
>
>OK, I'm not 100% sure what's going on, but I'm sure it's because Amazon 
>have their own method of putting the devices in xenstore.
>
>Here's the dmesg from an instance I just tried;  two of the xbd devices 
>are ones I normally use (and which reliably and consistently show up as 
>xbd0/xbd1 when I only use them), and two are ones I added for this 
>test.  Unfortunately, it doesn't print any information about them but 
>their xenbus id, so I'm not even certain which one gets mapped where.
>
>NetBSD 5.1.2 (XEN3PAE_DOMU) #0: Thu Feb  2 17:18:36 UTC 2012
>         
>builds%b6.netbsd.org@localhost:/home/builds/ab/netbsd-5-1-2-RELEASE/i386/201202021012Z-obj/home/builds/ab/netbsd-5-1-2-RELEASE/src/sys/arch/i386/compile/XEN3PAE_DOMU
>total memory = 1700 MB
>avail memory = 1662 MB
>mainbus0 (root)
>hypervisor0 at mainbus0: Xen version 3.4
>vcpu0 at hypervisor0: Intel 686-class, 2266MHz, id 0x106a5
>xenbus0 at hypervisor0: Xen Virtual Bus Interface
>xencons0 at hypervisor0: Xen Virtual Console Driver
>xencons0: using event channel 2
>npx0 at hypervisor0: using exception 16
>xbd0 at xenbus0 id 2049: Xen Virtual Block Device Interface
>xbd1 at xenbus0 id 2080: Xen Virtual Block Device Interface
>xbd2 at xenbus0 id 2050: Xen Virtual Block Device Interface
>xbd3 at xenbus0 id 2051: Xen Virtual Block Device Interface
>xennet0 at xenbus0 id 0: Xen Virtual Network Interface
>xennet0: MAC address 12:31:38:1b:a9:b6
>raidattach: Asked for 8 units
>Kernelized RAIDframe activated
>Searching for RAID components...
>boot device: xbd1
>root on xbd1a dumps on xbd1b
>mountroot: trying lfs...
>mountroot: trying ffs...
>mountroot: trying ext2fs...
>Ext2 fs: unsupported inode size
>mountroot: trying nfs...
>mountroot: trying smbfs...
>mountroot: trying ntfs...
>mountroot: trying msdos...
>mountroot: trying cd9660...
>mountroot: trying coda...
>no file system for xbd1 (dev 0x8e08)
>cannot mount root, error = 79
>root device (default xbd1a):
>

Can wedges help here? And recognize which one is which by their UUIDs?

christos