This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-users] Using lvm for domUs

To: "Tijl Van den Broeck" <subspawn@xxxxxxxxx>
Subject: Re: [Xen-users] Using lvm for domUs
From: Goswin von Brederlow <brederlo@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
Date: Fri, 04 May 2007 11:45:34 +0200
Cc: xen-users@xxxxxxxxxxxxxxxxxxx
Delivery-date: Fri, 04 May 2007 02:44:46 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
In-reply-to: <5a4fbd6c0705020156l29eff77ar28ebcaa1961b0f68@xxxxxxxxxxxxxx> (Tijl Van den Broeck's message of "Wed, 2 May 2007 10:56:05 +0200")
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
References: <4170c1720704301129k4cc8eac3m32395ce16c7c20ec@xxxxxxxxxxxxxx> <46363F2B.1090008@xxxxxxxxx> <46364D5F.4020503@xxxxxxxxxxxxxx> <5a4fbd6c0705020156l29eff77ar28ebcaa1961b0f68@xxxxxxxxxxxxxx>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Gnus/5.110006 (No Gnus v0.6) XEmacs/21.4.19 (linux)
"Tijl Van den Broeck" <subspawn@xxxxxxxxx> writes:

>> That said, I'd love to be able to have the DomUs detect partition size
>> changes without reboots of the DomU
> We do use LVM inside domU's for that, the SAN provides fixed size
> LUN's, which is transparantly mapped through into the domU using
> domU-name as prefix for the groups. It does have some minor annoyances
> when performing live migration (ie. your source & target dom0 must be
> -aware- of the LVM, so don't forget the rescan for LVM groups). Other
> than that, I don't see why not recommending it, as long as your dom0
> does not use the LVM, there are no issues with it... and you enable
> hot disk space adding :-)
>>  or even nicer -- have the LVM
>> handling be passed through to Dom0 directly instead of handled as
>> emulated partitions.  That is to say, /dev/DomU_VG/logs could be
>> configured to pass through to Dom0's /dev/main/xen_DomU_logs directly
>> ... but I've never played with the LVM internals so I'm not sure how
>> likely that is to ever happen.
> You're referring to internal "sharing" of an LVM from a dom0 towards a
> domU with all its options, right? That's an interesting concept. I
> wonder how it would handle on security levels though, I think you'd
> have to go beyond the mere "passing-through" of block devices,
> possibly mixing things making it too complex to handle.

If you have enough disks/partitions you can make one VG per domU and
share that for dom0 and domU using cluster lvm (clvm) to keep them in
sync. Instead of real disks/partitions you can run lvm in dom0, create
a LV per domU and run lvm on it again. But then you have to
specifically include the first lvm LVs in lvm.conf and run vgscan
twice. I think that should work.

We use lvm inside LVs here on a HA cluster but the inside lvm is
exclusively for the domU. No sharing with dom0. The dom0 lvm is so we
can resize and create/destroy LVs on the fly, the inside lvm is
because all our servers are setup the same.


Xen-users mailing list

<Prev in Thread] Current Thread [Next in Thread>