Has any one had any success with cow implementation in xen ? I found this somewhere http://www.atconsultancy.nl/cowloop/, has anyone tried it out with xen ? -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> Has any one had any success with cow implementation in xen ? I found this > somewhere http://www.atconsultancy.nl/cowloop/, has anyone tried it out with > xen ?Have you seen this page? http://wiki.xensource.com/xenwiki/COWHowTo Thanks, Michael _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Yes i saw that however cant seem to find a howto or something which i can use to implement it for testing. On 1/6/06, Michael Heyse <mhk@designassembly.de> wrote:> > Anand wrote: > > Has any one had any success with cow implementation in xen ? I found > this > > somewhere http://www.atconsultancy.nl/cowloop/, has anyone tried it out > with > > xen ? > > Have you seen this page? > > http://wiki.xensource.com/xenwiki/COWHowTo > > Thanks, > Michael >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> > Anand wrote: > > > http://www.atconsultancy.nl/cowloop/> Michael Heyse wrote: > > Have you seen this page? > > > > http://wiki.xensource.com/xenwiki/COWHowToAnand wrote:> Yes i saw that however cant seem to find a howto or something which i can > use to implement it for testing.Are you going to do performance tests? Finding out which one of UnionFS, cowloop and minifo: http://www.denx.de/wiki/bin/view/Know/MiniFOHome is the best performer would be very relevant when running multiple instances in virtual machines! Are you going to CC your results to the list? :-) I read somewhere that unionfs has a 10% performance overhead when using 2 filesystems and compiling a kernel. Can''t remember the hardware or other details :-/. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/9/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:>Are you going to do performance tests?No i wasn''t going to do performance tests.>Finding out which one of UnionFS, cowloop and minifo: >http://www.denx.de/wiki/bin/view/Know/MiniFOHome >is the best performer would be very relevant when running multipleinstances in virtual machines! I was looking for a cow like solution for my requirements in which i will have many domains on a single dom0. Since most of the domains will have same root fs, sharing a single instance made sense. I tried to compile cowloop, however it gave errors as i posted earlier. -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Molle Bestefich wrote:> Are you going to do performance tests? > > Finding out which one of UnionFS, cowloop and minifo: > http://www.denx.de/wiki/bin/view/Know/MiniFOHome > > is the best performer would be very relevant when running multiple > instances in virtual machines!Finding out which one is actually *stable* would be interesting enough for me. UnionFS is not enough of a lookalike to be reliably used as a root filesystem: /proc/<pid>/exe links point to the underlying filesystems (breaking several distributions'' tools for checking daemons'' status); there''s funkiness with mmap; and it''s quite crash-prone. Granted, I''m trying to run a full SuSE image with Oracle on top, so I''m perhaps demanding more than some users might -- but in my case, UnionFS is simply not stable enough... yet. There''s progress being made, though, and I''m watching the list. mini_fo looks interesting; I''ll give that a try. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Charles Duffy wrote:> > is the best performer would be very relevant when running multiple > > instances in virtual machines! > > Finding out which one is actually *stable* would be interesting enough > for me.Ok, good point :-).> UnionFS is not enough of a lookalike to be reliably used as a root > filesystem: /proc/<pid>/exe links point to the underlying filesystems > (breaking several distributions'' tools for checking daemons'' status);There''s that ''mount --move blah; pivot $unionfs $unionfs'' trick which should get around that. I''ve had absolutely no luck getting that to work though, I''m stuck getting -EBUSY from pivot_root :-/.> there''s funkiness with mmap;I''m curious, which?> and it''s quite crash-prone. Granted, I''m > trying to run a full SuSE image with Oracle on top, so I''m perhaps > demanding more than some users might -- but in my case, UnionFS is > simply not stable enough... yet. There''s progress being made, though, > and I''m watching the list.Gah. I''ve used it a lot with SLAX where it seemed stable, and I''ve just got Xen up and running with it. So I''m going to try it out anyway. Can''t get any worse than LVM snapshots, so I''m hopeful. I''ll probably be back here in a month, regretting :-).> mini_fo looks interesting; I''ll give that a try.Cool. I''ll take a look-see too when I get the time. I''m wondering, have you any verdict over cowloop since you do not mention that? _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > I''m wondering, have you any verdict over cowloop since you do not mention > that? >Where you able to compile cowloop ? When i tried it gave me compile errors under xen kernel. -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> Where you able to compile cowloop ?Haven''t tried. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > Anand wrote: > > Where you able to compile cowloop ? > > Haven''t tried. >Ok i got cowloop compiled. It doesn''t compile when booted with xen kernel. However under the default kernel it compiles. -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
After i compiled cowloop under the non xen kernel, the module was copied to /lib/modules/2.6.9-22.0.1.EL/misc/cowloop.ko. I copied the file to /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc/cowloop.ko. However doing a modprobe cowloop doesn''t work. Anyone else tried with cowloop ? On 1/10/06, Anand <xen.mails@gmail.com> wrote:> > On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote: > > > > Anand wrote: > > > Where you able to compile cowloop ? > > > > Haven''t tried. > > > > Ok i got cowloop compiled. It doesn''t compile when booted with xen kernel. > However under the default kernel it compiles. > > -- > > regards, > > Anand-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > Gah. > I''ve used it a lot with SLAX where it seemed stable, and I''ve just got > Xen up and running with it. So I''m going to try it out anyway. Can''t > get any worse than LVM snapshots, so I''m hopeful. I''ll probably be > back here in a month, regretting :-). >Can you share your experience of lvm snapshots ? I was thinking about using them. What problems did you face ? Thanks for your time. -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
cowloop is compiled under xen-kernel now. It was a stupid mistake. There needs to be 2 directories created in the /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/ directory before compiling cowloop /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc make install now compiles cowloop ./gplaccept if [ ! -d /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc ]; then mkdir -p /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc; fi cp cowloop.ko /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc /sbin/depmod -a cp cowdev /usr/sbin cp cowsync /usr/sbin cp cowrepair /usr/sbin cp cowmerge /usr/sbin cp cowlist /usr/sbin cp man/cowloop.4 /usr/share/man/man4 cp man/cowdev.1 /usr/share/man/man1 cp man/cowsync.1 /usr/share/man/man1 cp man/cowrepair.1 /usr/share/man/man1 cp man/cowmerge.1 /usr/share/man/man1 cp man/cowlist.1 /usr/share/man/man1 ./makecows 241 Now running modprobe cowloop gives the following error: FATAL: Error inserting cowloop (/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc/cowloop.ko): Invalid module format looks like its back to square one :( any clues ? On 1/10/06, Anand <xen.mails@gmail.com> wrote:> > After i compiled cowloop under the non xen kernel, the module was copied > to /lib/modules/2.6.9-22.0.1.EL/misc/cowloop.ko. I copied the file to > /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/misc/cowloop.ko. > > However doing a modprobe cowloop doesn''t work. > > Anyone else tried with cowloop ? > > On 1/10/06, Anand <xen.mails@gmail.com> wrote: > > > > On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com > wrote: > > > > > > Anand wrote: > > > > Where you able to compile cowloop ? > > > > > > Haven''t tried. > > > > > > > Ok i got cowloop compiled. It doesn''t compile when booted with xen > > kernel. However under the default kernel it compiles. > > > > -- > > > > regards, > > > > Anand > > > > > -- > > regards, > > Anand-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Missed this in the earlier mail. While running modprobe, the following is logged in /var/log/messages Jan 10 16:20:07 xen kernel: cowloop: disagrees about version of symbol struct_module -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Tuesday 10 January 2006 11:53, Anand wrote:> Missed this in the earlier mail. While running modprobe, the following is > logged in /var/log/messages > > Jan 10 16:20:07 xen kernel: cowloop: disagrees about version of symbol > struct_moduledid you do a ''make ARCH=xen'' ? Michael -- ---------------------------------------------------------------------------------------- Michael Mey Thinking Objects Software GmbH | mailto: michael.mey@to.com Lilienthalstrasse 2/1 | phone: +49 711 88770-147 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 ---------------------------------------------------------------------------------------- _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
make ARCH=xen returns the following error: make -C /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build SUBDIRS=/home/anand/root/xen-development/cowloop-2.15 -I. -Wall modules make[1]: Entering directory `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' make[1]: *** No rule to make target `modules''. Stop. make[1]: Leaving directory `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' make: *** [cowloop.ko] Error 2 Thanks for the response. On 1/10/06, Michael Mey <michael.mey@to.com> wrote:> > On Tuesday 10 January 2006 11:53, Anand wrote: > > Missed this in the earlier mail. While running modprobe, the following > is > > logged in /var/log/messages > > > > Jan 10 16:20:07 xen kernel: cowloop: disagrees about version of symbol > > struct_module > > did you do a ''make ARCH=xen'' ? > > Michael > > -- > > ---------------------------------------------------------------------------------------- > Michael Mey > Thinking Objects Software GmbH | mailto: michael.mey@to.com > Lilienthalstrasse 2/1 | phone: +49 711 88770-147 > 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 > > ---------------------------------------------------------------------------------------- > > >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> Can you share your experience of lvm snapshots ? > I was thinking about using them. What problems did you face ?Already have - search the archives ;-). (Or click here: http://article.gmane.org/gmane.linux.raid/10063/) _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Maybe it''s ''make install ARCH=xen'' This was only an example. Check the cowloop documentation for which make-targets you have to run and what option to use for specifying a different architecture and kernel-source (should point to the xen-src/linux-2.6.12-xen0 directory) Michael On Tuesday 10 January 2006 12:08, Anand wrote:> make ARCH=xen returns the following error: > > make -C /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build > SUBDIRS=/home/anand/root/xen-development/cowloop-2.15 -I. -Wall modules > make[1]: Entering directory `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' > make[1]: *** No rule to make target `modules''. Stop. > make[1]: Leaving directory `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' > make: *** [cowloop.ko] Error 2 > > Thanks for the response. > > On 1/10/06, Michael Mey <michael.mey@to.com> wrote: > > On Tuesday 10 January 2006 11:53, Anand wrote: > > > Missed this in the earlier mail. While running modprobe, the following > > > > is > > > > > logged in /var/log/messages > > > > > > Jan 10 16:20:07 xen kernel: cowloop: disagrees about version of symbol > > > struct_module > > > > did you do a ''make ARCH=xen'' ? > > > > Michael > > > > -- > > > > ------------------------------------------------------------------------- > >--------------- Michael Mey > > Thinking Objects Software GmbH | mailto: michael.mey@to.com > > Lilienthalstrasse 2/1 | phone: +49 711 88770-147 > > 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 > > > > ------------------------------------------------------------------------- > >--------------- > > -- > > regards, > > Anand-- ---------------------------------------------------------------------------------------- Michael Mey Thinking Objects Software GmbH | mailto: michael.mey@to.com Lilienthalstrasse 2/1 | phone: +49 711 88770-147 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 ---------------------------------------------------------------------------------------- _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
cowloop docs only talk about using make install. There is no ARCH in that. I read through the makefile also. Anyone else worked with cowloop and xen ? On 1/10/06, Michael Mey <michael.mey@to.com> wrote:> > Maybe it''s ''make install ARCH=xen'' > This was only an example. > Check the cowloop documentation for which make-targets you have to run and > what option to use for specifying a different architecture and > kernel-source > (should point to the xen-src/linux-2.6.12-xen0 directory) > > Michael > > On Tuesday 10 January 2006 12:08, Anand wrote: > > make ARCH=xen returns the following error: > > > > make -C /lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build > > SUBDIRS=/home/anand/root/xen-development/cowloop-2.15 -I. -Wall modules > > make[1]: Entering directory > `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' > > make[1]: *** No rule to make target `modules''. Stop. > > make[1]: Leaving directory > `/lib/modules/2.6.12.6-xen3_7.1_rhel4.1/build'' > > make: *** [cowloop.ko] Error 2 > > > > Thanks for the response. > > > > On 1/10/06, Michael Mey <michael.mey@to.com> wrote: > > > On Tuesday 10 January 2006 11:53, Anand wrote: > > > > Missed this in the earlier mail. While running modprobe, the > following > > > > > > is > > > > > > > logged in /var/log/messages > > > > > > > > Jan 10 16:20:07 xen kernel: cowloop: disagrees about version of > symbol > > > > struct_module > > > > > > did you do a ''make ARCH=xen'' ? > > > > > > Michael > > > > > > -- > > > > > > > ------------------------------------------------------------------------- > > >--------------- Michael Mey > > > Thinking Objects Software GmbH | mailto: michael.mey@to.com > > > Lilienthalstrasse 2/1 | phone: +49 711 88770-147 > > > 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 > > > > > > > ------------------------------------------------------------------------- > > >--------------- > > > > -- > > > > regards, > > > > Anand > > -- > > ---------------------------------------------------------------------------------------- > Michael Mey > Thinking Objects Software GmbH | mailto: michael.mey@to.com > Lilienthalstrasse 2/1 | phone: +49 711 88770-147 > 70825 Stuttgart-Korntal, Germany | fax: +49 711 88770-449 > > ---------------------------------------------------------------------------------------- > > >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Alright i guess i should just keep my distance with lvm now. What are you using then ? On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > Anand wrote: > > Can you share your experience of lvm snapshots ? > > I was thinking about using them. What problems did you face ? > > Already have - search the archives ;-). > (Or click here: http://article.gmane.org/gmane.linux.raid/10063/) >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> Alright i guess i should just keep my distance with lvm now.Suit yourself, hehe.> What are you using then ?Uhm, tar+gzip and a differ for now :-)... I''ll admit it''s far from optimal :-). _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > Anand wrote: > > Alright i guess i should just keep my distance with lvm now. > > Suit yourself, hehe.Hehe do i have a choice ? I am looking for a stable solution wherein there will be many domains. So if lvm2 with snapshots is creating problems as you say so, i don''t have much choice. I tried only couple of times with lvm2 snapshots and they didn''t cause any problems. This is a centos 4.1 box.> What are you using then ? > > Uhm, tar+gzip and a differ for now :-)... > > I''ll admit it''s far from optimal :-). >huh ? i have to say it just went over my head. Care to fly a little low so i can understand something about it ? ;) -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Hi.> get any worse than LVM snapshots, so I''m hopeful.Can someone please give me hints on performance/stability/scalability issues of backing domUs with LVM snapshot files? -- Udvozlet -------------------------------------------------------------------- Sipos Ferenc email: frank@sudo.hu phone: 06/20/491-3428 -------------------------------------------------------------------- GnuPG Signature: http://keys.sudo.hu/pubkey-frank.gpg Key fingerprint = D604 140C 28BF 58C1 24D8 D20B 2A60 86DA 226D 7162 ____________________________________________________________________ _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> > > What are you using then ? > > > > Uhm, tar+gzip and a differ for now :-)... > > > > I''ll admit it''s far from optimal :-). > > huh ? i have to say it just went over my head. > Care to fly a little low so i can understand something about it ? ;)I make a tar archive of each domU filesystem: # tar /backup/domain1_`date +%s`.tar /{bin,sbin,etc,home,lib,opt,usr,var} Then gzip it. I''m experimenting with various binary differs so I can diff the backup locally on the server against the previous backup .tar.gz and only send the couple of MB that actually changed to my backup server. The backups I''ve made in the past worked by sending the entire .tar.gz, which is a couple GB each, which was sort of inefficient. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Thanks for the reply. And for the filesystem are you using flat images or volumes ? Isn''t tar and gzip combination going to be a cpu hog ? On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > Anand wrote: > > > > What are you using then ? > > > > > > Uhm, tar+gzip and a differ for now :-)... > > > > > > I''ll admit it''s far from optimal :-). > > > > huh ? i have to say it just went over my head. > > Care to fly a little low so i can understand something about it ? ;) > > I make a tar archive of each domU filesystem: > # tar /backup/domain1_`date +%s`.tar /{bin,sbin,etc,home,lib,opt,usr,var} > > Then gzip it. > > I''m experimenting with various binary differs so I can diff the backup > locally on the server against the previous backup .tar.gz and only > send the couple of MB that actually changed to my backup server. The > backups I''ve made in the past worked by sending the entire .tar.gz, > which is a couple GB each, which was sort of inefficient. >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
________________________________
From: xen-users-bounces@lists.xensource.com
[mailto:xen-users-bounces@lists.xensource.com] On Behalf Of Anand
Sent: 10 January 2006 11:59
To: Molle Bestefich
Cc: xen-users@lists.xensource.com
Subject: Re: [Xen-users] Re: cow implementation
Thanks for the reply. And for the filesystem are you using flat
images or volumes ?
Isn''t tar and gzip combination going to be a cpu hog ?
Gzip will use up a fair amount of cpu-time, but the time it take to
transfer the image that is 30-70% smaller is significantly more than the
CPU-time it''s taken, so unless you have something else productive to do
with your cpu whilst you''re backing up/transferring the file(s),
you''re
probably better off compressing the file(s).
--
Mats
On 1/10/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:
Anand wrote:
> > > What are you using then ?
> >
> > Uhm, tar+gzip and a differ for now :-)...
> >
> > I''ll admit it''s far from optimal :-).
>
> huh ? i have to say it just went over my head.
> Care to fly a little low so i can understand something
about it ? ;)
I make a tar archive of each domU filesystem:
# tar /backup/domain1_`date +%s`.tar
/{bin,sbin,etc,home,lib,opt,usr,var}
Then gzip it.
I''m experimenting with various binary differs so I can
diff the backup
locally on the server against the previous backup
.tar.gz and only
send the couple of MB that actually changed to my backup
server. The
backups I''ve made in the past worked by sending the
entire .tar.gz,
which is a couple GB each, which was sort of
inefficient.
--
regards,
Anand
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users
On 1/10/06, Petersson, Mats <mats.petersson@amd.com> wrote:> > > Gzip will use up a fair amount of cpu-time, but the time it take to > transfer the image that is 30-70% smaller is significantly more than > the CPU-time it''s taken, so unless you have something else productive to do > with your cpu whilst you''re backing up/transferring the file(s), you''re > probably better off compressing the file(s). >Well my backup server is going to be just sitting on the next rack probably so the network transfer won''t affect much. As compared to cpu time, i would rather prefer to use the cpu time for processing inside the domains than for running tar+gzip. Hmm... given this, one solution would be to run some kind of background process which monitors the cpu usage, when its low, run the backup service, if there is a demand of the cpu by any process/ domain, just pause the backup and resume whenever cpu becomes free. However if the host keeps on being too much busy always the backup process can be hanged up in memory for ages (have seen this myself). -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
lvm snapshots (at least in combination with xen) will sooner or later eat your data. at least this is my experience. It would have a good performance and should run very well in theory, but my tests showed me thats its not usable right now. If I create a lvm snapshot of a domainU and mount it in domain0, then I can access the data of the snapshot. But if I try to backup the data (for example with rsync or tar) then most of the time the corresponding domainU seems to hang. After the backup is completed the domainU seems to work again. But for about 50% of the time after the backup was completed the domainU crashed in the next minutes after deleting the snapshot. It is not always crashing as well as not alawys the domainU seems to hang, but most of the time. I read also some reports from other people having this kind of problems too. --Ralph Am Dienstag, 10. Januar 2006 12:47 schrieb Sipos Ferenc:> Hi. > > > get any worse than LVM snapshots, so I''m hopeful. > > Can someone please give me hints on performance/stability/scalability > issues of backing domUs with LVM snapshot files?_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I made another post after that one which said that i compiled it under the xen kernel as well. However the module refuses to work. On 1/11/06, tbrown@baremetal.com <tbrown@baremetal.com> wrote:> > On Tue, 10 Jan 2006, Anand wrote: > > > After i compiled cowloop under the non xen kernel, the module was copied > to > > /lib/modules/2.6.9-22.0.1.EL/misc/cowloop.ko. I copied the file to > > that means you built it for the wrong kernel... >-- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> Thanks for the reply.np.> And for the filesystem are you using flat images or volumes ?A base volume shared between domU''s and an image on top with per-domU files. Going to drop the images and switch to volumes I think, since I don''t like the maximum of 8 virtual machines that this feature imposes. Too bad, cause I did like the sparseness that the images allowed for ;-). _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/11/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> Anand wrote: > > Thanks for the reply. > > np. > > > And for the filesystem are you using flat images or volumes ? > > A base volume shared between domU''s and an image on top with per-domU files. > Going to drop the images and switch to volumes I think, since I don''t > like the maximum of 8 virtual machines that this feature imposes. Too > bad, cause I did like the sparseness that the images allowed for ;-). > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >>maximum of 8 virtual machines that this feature imposesMolle, If you are refering to the default limit on the number of loop devices (8) you can change this setting. Please see: http://wiki.xensource.com/xenwiki/XenFaq?highlight=%28loop%29#head-b4167b873a6b3a6657ecf9339c44e4a0692f8ada David _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/12/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > > And for the filesystem are you using flat images or volumes ? > > A base volume shared between domU''s and an image on top with per-domU > files. > Going to drop the images and switch to volumes I think, since I don''t > like the maximum of 8 virtual machines that this feature imposes. Too > bad, cause I did like the sparseness that the images allowed for ;-). >Can you explain this please ? base volume shared between domU''s ? correct me if i am wrong but doesn''t that imply a cow like system ? -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
David H wrote:> If you are refering to the default limit on the number of loop > devices (8) you can change this setting.Thanks for the pointer! :-) Anand wrote:> base volume shared between domU''s ? > correct me if i am wrong but doesn''t that imply a cow like system ?It does.> Can you explain this please ?In my current (experimental) setup I''m using unionfs. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/16/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > > base volume shared between domU''s ? > > correct me if i am wrong but doesn''t that imply a cow like system ? > > It does. > > > Can you explain this please ? > > In my current (experimental) setup I''m using unionfs. >Could i ask for a howto / link where i can know more about how to try the same setup ? -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Anand wrote:> > In my current (experimental) setup I''m using unionfs. > > Could i ask for a howto / link where i can know more about how to try the > same setup ?I used the classic trial''n''error approach, haven''t seen any howto on the subject. The approach is rather poorly documented AFAICT, in fact my best source of information so far has been this mailing list! Try setting it up and if you have any questions, I''ll be happy to answer them? _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 1/17/06, Molle Bestefich <molle.bestefich@gmail.com> wrote:> > I used the classic trial''n''error approach, haven''t seen any howto on > the subject. > > The approach is rather poorly documented AFAICT, in fact my best > source of information so far has been this mailing list! > > Try setting it up and if you have any questions, I''ll be happy to answer > them? >Thanks. I will get to it once i finish the present tests i am doing on xen. -- regards, Anand _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Greetings! I would really like to see someone with experience on this start a wiki. I think a lot of people could benifit from it. I also wonder what kind of performance and overhead a COW setup would have? Thanks in advance for future and past contributions to the subject! Regards, hikenboot --------------------------------- Yahoo! Photos Got holiday prints? See all the ways to get quality prints in your hands ASAP. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Steven Anderson wrote:> I would really like to see someone with experience on this start a wiki. I > think a lot of people could benifit from it.That would be nice. I''d be happy to contribute my experiences when I get more time. (Which will probably be a couple of months from now, but anyway.) Personally, I''d like to see some more information on filesystems that zero blocks when they''re freed and COW implementations that nukes blocks that gets zeroed from the write target :-).> I also wonder what kind of performance and overhead a COW setup would have?I''ve seen suggestions that unionfs adds about 15% overhead in a test setup. It''s of course dependant on your setup.. The more filesystems you add on top of each other, the slower it will get, I imagine. OTOH, dentrys are usually cached, so it might not be that big in the common case. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Possibly Parallel Threads
- Error: Device 769 (vbd) could not be connected. Backend device not found.
- Tutorial : Debian, Xen and CLUSTER / GFS Support
- Sharing a partition between dom0 and domU
- RE: live migration with xen 2.0.7 with fibre channel onDebian - help needed
- Managing multiple Dom0''s