I''ve just stumbled into the wonderful world that is lvm. At the same time I have just created a new partition for a new domain on my xen server and am therefore having to reboot to re-read the partition table. Do xen domains work under lvm vd''s? I''d assume yes and I know this has been asked before but I''d like to know the up-to-date answer before I jump in. thanks James
> I''ve just stumbled into the wonderful world that is lvm. > At the same time I have just created a new partition for a new domain on my xen server and am therefore having to reboot to re-read the partition table. Do xen domains work under lvm vd''s? I''d assume yes and I know this has been asked before but I''d like to know the up-to-date answer before I jump in.It certainly should work, though I can''t say I''ve tested it. The only annoyance I can think of is if you want to export an LVM volume as a whole-disk to another domain (e.g. to appear as /dev/sda rather than individual partitions /dev/sda1 etc.). Within dom0, I''m not sure how you''d be able to get a partition table onto the volume and then populate the internal partitions with file systems and files. I guess you''d really want dom0 to export the LVM volume back to itself as e.g. /dev/sdb so you could then mount the individual partitions, but this is currently supported (perhaps device mapper in 2.6 will give us this for free?) Using lvm should work fine if you take volume per exported partition approach. It''ll all be transparent to the other domain anyhow, and has the advantage that you can resize individual partitions. It should be possible to use the lvm snapshot feature to do CoW disks. However, the current behaviour is a bit confusing as new writes go to the main volume and copies of the replaced data goes to the log volume. Converting the lvm snapshot module into a CoW modeule with the more conventional semantics should be pretty easy (original volume stays read-only, writes go to log volume). We haven''t looked too hard at this yet as I think life will be easier under device mapper in 2.6. Ian ------------------------------------------------------- This SF.Net email sponsored by Black Hat Briefings & Training. Attend Black Hat Briefings & Training, Las Vegas July 24-29 - digital self defense, top technical experts, no vendor pitches, unmatched networking opportunities. Visit www.blackhat.com _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, Jul 14, 2004 at 07:58:23AM +0100, Ian Pratt wrote:> The only annoyance I can think of is if you want to export an LVM > volume as a whole-disk to another domain (e.g. to appear as > /dev/sda rather than individual partitions /dev/sda1 etc.). > Within dom0, I''m not sure how you''d be able to get a partition > table onto the volume and then populate the internal partitions > with file systems and files. I guess you''d really want dom0 to > export the LVM volume back to itself as e.g. /dev/sdb so you > could then mount the individual partitions, but this is currently > supported (perhaps device mapper in 2.6 will give us this for free?)I don''t think you''d want to export an LVM logical volume as a whole-disk to another doain since then you won''t have the ability to resize the partitions in that whole-disk. I expect that would have been the reason why one would use LVM in the first place. You can use device mapper to create devices for the partitions on the logical volume, I don''t think there''s a tool to set this up but it''s quite straightforward since fdisk works on logical volumes and gives you all the information you need (start/end). losetup should have an option to read a partition table off the underlying device and then present the partitions as loopa, loopa1, loopa2, ...> Using lvm should work fine if you take volume per exported > partition approach. It''ll all be transparent to the other domain > anyhow, and has the advantage that you can resize individual > partitions.exactly. christian ------------------------------------------------------- This SF.Net email sponsored by Black Hat Briefings & Training. Attend Black Hat Briefings & Training, Las Vegas July 24-29 - digital self defense, top technical experts, no vendor pitches, unmatched networking opportunities. Visit www.blackhat.com _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Jul 14, 2004, at 4:47 AM, Christian Limpach wrote:> I don''t think you''d want to export an LVM logical volume as a > whole-disk to another doain since then you won''t have the ability > to resize the partitions in that whole-disk. I expect that would > have been the reason why one would use LVM in the first place.It''s exactly why I would like to use LVM. And XFS resizes very nicely with LVMs. It also gives you the ability to create up to 256 LVs on a single Volume Group, as opposed to the I think 16 partition limit if you DOS-like partition a physical disk. And you get things like snapshotting and migration of LVs between volume groups/physical disks that are really nice. And it can ride on top of the software RAID layer for those of us out here with cheap IDE subsystems, so you make a single RAID1 or RAID5 and LVM-partition that into as many volumes as you need, instead of having to make a zillion RAID devices. so yeah, LVM is nice for tossing lots of VMs around on a single machine. :) And I was going to post some more questions about it today because I''m not able to get the domain to recognize the exported LVs. (Is there a xen-users list or something so I don''t have to bog up the -dev list with noob user questions? Or are those just as well posted here anyway?) I''ve currently created LVs with LVM2 for swap and root for the new domain and changed the xmdefaults file to point to these partitions: disk = [ ''phy:/dev/vg_pool/lv_vm0%d_swap,hda1,w'' % (vmid), ''phy:/dev/vg_pool/lv_vm0%d_root,hda2,w'' % (vmid) ] root = /dev/hda2 lv_vm01_root already has a filesystem created on it with gentoo already installed, ready for the new domain to boot right up. But the new domain comes up and doesn''t find the root volume. The status messages seem to indicate that the LVs are getting exported correctly, but either I''ve built my xenU kernel wrong and/or I just don''t know what I''m doing, since it doesn''t seem to see any block devices at all. (Probably I should be exporting them as something other than hda1/2 ?) I haven''t gone back and simply fdisk''ed the drive to try to export "normal" partitions to the new domain; I''ve kind of picked a "hard" way to do it right off the bat I guess. I''m hoping I just have the syntax wrong. (Is there another type than "phy" ?) I dug through the python for "xm" but didn''t see anything that screamed out at any "right" way to do it. (FYI and FWIW - this is a spare IDE dual-celeron machine I built from stuff scrounged out of old servers, not anything fancy with iSCSI or anything like that. I work for a hosting/consulting company who currently have somewhere on the order of 1000 UML-based virtual-hosts deployed all over the place. I''m mostly learning Xen for myself because I need to deploy a couple new servers of my own to replace my aging mail/web server and want to start off using UML or Xen to deploy everything as VMs and prefer the Xen approach, but I''d also like to try to push at least a few Xen servers into work to compare against the UML machines. UML has a lot of mindshare right now though from the amount of work we''ve put in to support/maintain it, but I personally think the Xen approach is a lot cleaner/easier to maintain than UML, and live migration would be a huge bonus.) # xm create -n -c vmid=1 dhcp=on Using config file /etc/xen/xmdefaults (vm (name ''This is VM 1'') (memory ''64'') (cpu ''1'') (image (linux (kernel /boot/bzImage-2.4.26-xenU) (ip :::::eth0:on) (root ''/dev/hda2 ro'') ) ) (device (vbd (uname phy:/dev/vg_pool/lv_vm01_swap) (dev hda1) (mode w))) (device (vbd (uname phy:/dev/vg_pool/lv_vm01_root) (dev hda2) (mode w))) (device (vif (mac aa:0:0:1c:c8:7))) # xm create -c vmid=1 dhcp=on Started domain 28, console on port 9628 ************ REMOTE CONSOLE: CTRL-] TO QUIT ******** Linux version 2.4.26-xeno-xenU (root@vargas) (gcc version 3.3.3 20040412 (Gentoo Linux 3.3.3-r6, ssp-3.3.2-2, pie-8.7.6)) #9 Tue Jul 13 22:54:03 EDT 2004 On node 0 totalpages: 16384 zone(0): 4096 pages. zone(1): 12288 pages. zone(2): 0 pages. Kernel command line: ip=:::::eth0:on root=/dev/hda2 ro Initializing CPU#0 Xen reported: 451.031 MHz processor. Calibrating delay loop... 4508.87 BogoMIPS Memory: 62828k/65536k available (1368k kernel code, 2708k reserved, 165k data, 44k init, 0k highmem) Dentry cache hash table entries: 8192 (order: 4, 65536 bytes) Inode cache hash table entries: 4096 (order: 3, 32768 bytes) Mount cache hash table entries: 512 (order: 0, 4096 bytes) Buffer cache hash table entries: 4096 (order: 2, 16384 bytes) Page-cache hash table entries: 16384 (order: 4, 65536 bytes) CPU: L1 I cache: 16K, L1 D cache: 16K CPU: L2 cache: 128K CPU: Intel Celeron (Mendocino) stepping 05 POSIX conformance testing by UNIFIX Linux NET4.0 for Linux 2.4 Based upon Swansea University Computer Society NET3.039 Initializing RT netlink socket Starting kswapd Journalled Block Device driver loaded devfs: v1.12c (20020818) Richard Gooch (rgooch@atnf.csiro.au) devfs: boot_options: 0x1 SGI XFS with no debug enabled Event-channel device installed. Xen virtual console successfully installed as tty Starting Xen Balloon driver pty: 2048 Unix98 ptys configured Initialising Xen virtual block device RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize Initializing Cryptographic API Initialising Xen virtual ethernet frontend driver<6>NET4: Linux TCP/IP 1.0 for NET4.0 IP Protocols: ICMP, UDP, TCP, IGMP IP: routing cache hash table of 512 buckets, 4Kbytes TCP: Hash tables configured (established 4096 bind 8192) Linux IP multicast router 0.06 plus PIM-SM NET4: Unix domain sockets 1.0/SMP for Linux NET4.0. VFS: Cannot open root device "hda2" or 03:02 Please append a correct "root=" boot option Kernel panic: VFS: Unable to mount root fs on 03:02 <0>Rebooting in 1 seconds.. ************ REMOTE CONSOLE EXITED ***************** -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- "We all enter this world in the | Support Electronic Freedom same way: naked; screaming; soaked | http://www.eff.org/ in blood. But if you live your | http://www.anti-dmca.org/ life right, that kind of thing |--------------------------- doesn''t have to stop there." -- Dana Gould ------------------------------------------------------- This SF.Net email sponsored by Black Hat Briefings & Training. Attend Black Hat Briefings & Training, Las Vegas July 24-29 - digital self defense, top technical experts, no vendor pitches, unmatched networking opportunities. Visit www.blackhat.com _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> And I was going to post some more questions about it today because I''m > not able to get the domain to recognize the exported LVs. (Is there a > xen-users list or something so I don''t have to bog up the -dev list > with noob user questions? Or are those just as well posted here > anyway?)Here''s fine. I think it might be time to split off a xen-users when 2.0 comes out.> I''ve currently created LVs with LVM2 for swap and root for the new > domain and changed the xmdefaults file to point to these partitions: > > disk = [ ''phy:/dev/vg_pool/lv_vm0%d_swap,hda1,w'' % (vmid), > ''phy:/dev/vg_pool/lv_vm0%d_root,hda2,w'' % (vmid) ]That''s exactly what I''d do, and would expect it to work, though haven''t tested it. What do you get if you do ''sfdisk -s /dev/vg_pool/lv_vm01_root'' ?> I haven''t gone back and simply fdisk''ed the drive to try to export > "normal" partitions to the new domain; I''ve kind of picked a "hard" way > to do it right off the bat I guess. I''m hoping I just have the syntax > wrong. (Is there another type than "phy" ?) I dug through the python > for "xm" but didn''t see anything that screamed out at any "right" way > to do it.The ''phy:'' syntax is a legacy of 1.2 xend should print out a line something like the following, providing the debug hasn''t been quietened down too much: blkif_dev_create> 35 2049 w {''device'': 774, ''type'': ''Disk'', ''start_sector'': 0L, ''nr_sectors'': 37881206L} Please can you post me this output and I''ll take a look. I don''t really want to have to install lvm myself to debug this... Thanks, Ian ------------------------------------------------------- This SF.Net email sponsored by Black Hat Briefings & Training. Attend Black Hat Briefings & Training, Las Vegas July 24-29 - digital self defense, top technical experts, no vendor pitches, unmatched networking opportunities. Visit www.blackhat.com _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, Jul 14, 2004 at 06:13:20PM +0100, Ian Pratt wrote:> > And I was going to post some more questions about it today because I''m > > not able to get the domain to recognize the exported LVs. > > > I''ve currently created LVs with LVM2 for swap and root for the new > > domain and changed the xmdefaults file to point to these partitions: > > > > disk = [ ''phy:/dev/vg_pool/lv_vm0%d_swap,hda1,w'' % (vmid), > > ''phy:/dev/vg_pool/lv_vm0%d_root,hda2,w'' % (vmid) ] > > That''s exactly what I''d do, and would expect it to work, though > haven''t tested it.Although i use LVM, i use it through the new-style EVMS, not directly. However, be that as may be, i don''t have any problems using EVMS volumes with Xen, and since, afaia, both systems use device-mapper to do their work, i also can''t imagine that using LVM directly would be a problem. -- Jody Belka knew (at) pimb (dot) org ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Jul 14, 2004, at 1:13 PM, Ian Pratt wrote:>> disk = [ ''phy:/dev/vg_pool/lv_vm0%d_swap,hda1,w'' % (vmid), >> ''phy:/dev/vg_pool/lv_vm0%d_root,hda2,w'' % (vmid) ] > > That''s exactly what I''d do, and would expect it to work, though > haven''t tested it.At least I''m not totally off-track...> What do you get if you do ''sfdisk -s /dev/vg_pool/lv_vm01_root'' ?# sfdisk -s /dev/vg_pool/lv_vm01_root 2097152 it''s a 2G LV.> xend should print out a line something like the following, > providing the debug hasn''t been quietened down too much: > > blkif_dev_create> 35 2049 w {''device'': 774, ''type'': ''Disk'', > ''start_sector'': 0L, ''nr_sectors'': 37881206L} > > > Please can you post me this output and I''ll take a look. I don''t > really want to have to install lvm myself to debug this...no problem: blkif_dev_create> 35 769 w {''device'': 65024, ''type'': ''Disk'', ''start_sector'': 0L, ''nr_sectors'': 204800L} blkif_dev_create> 35 770 w {''device'': 65025, ''type'': ''Disk'', ''start_sector'': 0L, ''nr_sectors'': 4194304L} If there''s anything else you''d like me to do to test this is fine; this is a purely dev box I can do anything to. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- "We all enter this world in the | Support Electronic Freedom same way: naked; screaming; soaked | http://www.eff.org/ in blood. But if you live your | http://www.anti-dmca.org/ life right, that kind of thing |--------------------------- doesn''t have to stop there." -- Dana Gould ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, Jul 14, 2004 at 06:01:51PM +0000, Jody Belka wrote:> On Wed, Jul 14, 2004 at 06:13:20PM +0100, Ian Pratt wrote: > Although i use LVM, i use it through the new-style EVMS, not directly. > However, be that as may be, i don''t have any problems using EVMS volumes > with Xen, and since, afaia, both systems use device-mapper to do their > work, i also can''t imagine that using LVM directly would be a problem.Oh are you running 2.4.26 with the device-mapper and vfs locking patches? Or some other legacy EVMS thing that doesn''t use device-mapper? bill ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> At least I''m not totally off-track...Your config looks sane and Xend appears to be doing the right thing. I notice you''re running devfs in the guest domain. There have been problems with this in the past - could you try disabling it? Cheers, Mark ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, Jul 14, 2004 at 07:17:54PM +0100, Wm wrote:> Oh are you running 2.4.26 with the device-mapper and vfs locking > patches?Yep, 2.4.26 with the v4 ioctl device-mapper, the vfs lock patch, and all the other evms 2.3.4 patches relevant for 2.4.26. my current tree is laid out like this: hda4 (Disk segment: DosSegMgr) -> hda4bbr (Disk segment: BBRseg) -> lvm/disk1 (Storage container: LvmRegMgr) -> lvm/disk1/* (Storage regions: LvmRegMgr) -> Logical volumes and another tree starting with hdc1. I''m then attaching individual volumes to my xen instances. My typical domN volume layout is 1 for the root, 1 for swap, and at least 1 data volume. One thing i''m looking forward to in the future is being able to connect/disconnect volumes to/from running domains. -- Jody Belka knew (at) pimb (dot) org ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> One thing i''m looking forward to in the future is being able > to connect/disconnect volumes to/from running domains.Yep, I miss this functionality from Xen 1.2 too. Three things need to happen, none very hard: * cleanup of the blkdev backend code. (This is mostly deleting clever code that we no longer need since we don''t have to implement all the extent stuff ourselves as we only export whole devices). We could also add some probe functions so that xend can ask what devices a backend has available to export. * add support for dynamically adding and changing a VM''s config in xend. We need this for a bunch of things, and I belive Mike is thinking about it. * have xend poke the frontend driver to tell it that it has a new device (add a new message). Ian ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> > At least I''m not totally off-track... > > Your config looks sane and Xend appears to be doing the right thing. I notice > you''re running devfs in the guest domain. There have been problems with this > in the past - could you try disabling it?Well spotted. devfs is almost certainly the problem. Anyone understand devfs enough to know where we need to calls to the blkif/frontend driver so that the imported device gets registered? Ian ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Jul 14, 2004, at 2:29 PM, Mark Williamson wrote:>> At least I''m not totally off-track... > > Your config looks sane and Xend appears to be doing the right thing. > I notice > you''re running devfs in the guest domain. There have been problems > with this > in the past - could you try disabling it?That was it. No devfs and it will boot. (It''s there initially because "stock" gentoo without some recent updates whines if there''s no devfs. Fortunately they''ve gotten away from that.) I''d have gotten there eventually. :) Thanks! -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- "We all enter this world in the | Support Electronic Freedom same way: naked; screaming; soaked | http://www.eff.org/ in blood. But if you live your | http://www.anti-dmca.org/ life right, that kind of thing |--------------------------- doesn''t have to stop there." -- Dana Gould ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, Jul 14, 2004 at 07:49:03PM +0100, Ian Pratt wrote:> Anyone understand devfs enough to know where we need to calls to > the blkif/frontend driver so that the imported device gets > registered?I''d suggest looking at the Linux Device Drivers book, 2nd edition, to be found at http://www.xml.com/ldd/chapter/book/ devfs information starting at page 85 -- Jody Belka knew (at) pimb (dot) org ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
did I read the last message correctly w.r.t. block devices? Can I create a file and use it as / for a guest OS? ron ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> Can I create a file and use it as / for a guest OS?Yes. You should be able to export anything dom0 can see as a block device to a guest OS. You can use losetup to make a file look like a block device in dom0, then export that block device as / for a guest. Since you can export *anything* dom0 sees as a block device, you could also export to domains: * iSCSI devices imported from the network by an initiator in dom0 * NBD devices imported from the network by dom0 * LVM devices * MD devices (software raid) * Ramdisk (although this is arguably rather pointless) * Any other random weird block device you have in dom0! Cheers, Mark ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> On Wed, Jul 14, 2004 at 07:49:03PM +0100, Ian Pratt wrote: > > Anyone understand devfs enough to know where we need to calls to > > the blkif/frontend driver so that the imported device gets > > registered? > > I''d suggest looking at the Linux Device Drivers book, 2nd edition, > to be found at http://www.xml.com/ldd/chapter/book/ > devfs information starting at page 85To make ourselves look like proper IDE or SCSI devfs nodes would be a hassle I think, as devfs dictates a different naming schgeme to the usual Linux style. Since it''s dying it''s probably not worth it -- especially since the major distros are no longer dependent on it. Possibly we should disable it in the kernel config to avoid problems. If the udev/sysfs effort withers and devfs is revived then it might be worth considering supporting it. -- Keir ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Jul 14, 2004, at 2:49 PM, Ian Pratt wrote:>>> At least I''m not totally off-track... >> >> Your config looks sane and Xend appears to be doing the right thing. >> I notice >> you''re running devfs in the guest domain. There have been problems >> with this >> in the past - could you try disabling it? > > Well spotted. devfs is almost certainly the problem. > > Anyone understand devfs enough to know where we need to calls to > the blkif/frontend driver so that the imported device gets > registered?I wouldn''t even worry about it. devfs is deprecated in 2.6 and barely even works there. gentoo is the only distro I know that still to some extent "depends" on devfs being present by default, and that''s deprecated once you get the machine up and running with the latest updates. "sysfs/hotplug/udev" is where it''s at in the 2.6 world. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- "I think that''s what they mean by | "nickels a day can feed a child." | http://www.eff.org/ I thought, "How can food be so | http://www.anti-dmca.org/ cheap over there?" It''s not, they |-------------------------- just eat the nickels." -- Peter Nguyen ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
Just as a followup - i''m the one who started this thread :) i got it working almost straight away. dom0 is set up thus: xen1:~# pvscan pvscan -- reading all physical volumes (this may take a while...) pvscan -- ACTIVE PV "/dev/sda6" of VG "vg00" [35.25 GB / 29.25 GB free] pvscan -- total: 1 [35.28 GB] / in use: 1 [35.28 GB] / in no VG: 0 [0] xen1:~# vgscan vgscan -- reading all physical volumes (this may take a while...) vgscan -- found active volume group "vg00" xen1:~# lvscan lvscan -- ACTIVE "/dev/vg00/mail2" [2 GB] lvscan -- ACTIVE "/dev/vg00/sbssgw1" [2 GB] lvscan -- ACTIVE "/dev/vg00/gaia" [2 GB] lvscan -- 3 logical volumes with 6 GB total in 1 volume group lvscan -- 3 active logical volumes my config file looks like this: disk = [ ''phy:vg00/mail2,sda1,w'' ] root = "/dev/sda1 ro" both dom0 and domU are all running debian ''sarge''. No devfs (I could never get devfs working and the general consensus was that it''s a horrible thing that deserves to die and will be replaced by something better in 2.6. I like devfs but not enough to fight for it). No kernel patches. debian lvm10 package installed in dom0. hth James