Update seabios and upstream qemu versions used on xen-unstable

Update seabios and upstream qemu versions used on xen-unstable

[Fabio Fantoni]

I think it would be good to update seabios and upstream qemu versions 
used with xen-unstable.
Recently I tested on my testing system with xen-unstable seabios updated 
to debian package version 1.7.4-4 without finding regressions and 
upstream qemu 2.0-rc0 from git where I found 2 regressions already 
reported and solved on upstream git.
I updated qemu to 2.0-rc2 where I did some tests without finding other 
regressions or bugs for now.
I also noticed on the wiki page of the next hackathon the topic "QEMU 
and SeaBIOS regression testing". As I reported some time ago it would be 
of great help.

Thanks for any reply.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Sander Eikelenboom]

Monday, April 14, 2014, 1:11:36 PM, you wrote:

> I think it would be good to update seabios and upstream qemu versions 
> used with xen-unstable.
> Recently I tested on my testing system with xen-unstable seabios updated 
> to debian package version 1.7.4-4 without finding regressions and 
> upstream qemu 2.0-rc0 from git where I found 2 regressions already 
> reported and solved on upstream git.
> I updated qemu to 2.0-rc2 where I did some tests without finding other 
> regressions or bugs for now.
> I also noticed on the wiki page of the next hackathon the topic "QEMU 
> and SeaBIOS regression testing". As I reported some time ago it would be 
> of great help.

> Thanks for any reply.

I'm also running with current qemu-upstream, also without regressions at the 
moment, so +1 :-)

--
Sander

Re: Update seabios and upstream qemu versions used on xen-unstable

[Anthony PERARD]

On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
> I think it would be good to update seabios and upstream qemu versions used
> with xen-unstable.
> Recently I tested on my testing system with xen-unstable seabios updated to
> debian package version 1.7.4-4 without finding regressions and upstream qemu
> 2.0-rc0 from git where I found 2 regressions already reported and solved on
> upstream git.
> I updated qemu to 2.0-rc2 where I did some tests without finding other
> regressions or bugs for now.
> I also noticed on the wiki page of the next hackathon the topic "QEMU and
> SeaBIOS regression testing". As I reported some time ago it would be of
> great help.

I was thinking of submitting the update for QEMU when QEMU 2.0 will be
out.

But thanks for the reminder.

-- 
Anthony PERARD

Re: Update seabios and upstream qemu versions used on xen-unstable

[Sander Eikelenboom]

Tuesday, April 15, 2014, 1:18:32 PM, you wrote:

> On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
>> I think it would be good to update seabios and upstream qemu versions used
>> with xen-unstable.
>> Recently I tested on my testing system with xen-unstable seabios updated to
>> debian package version 1.7.4-4 without finding regressions and upstream qemu
>> 2.0-rc0 from git where I found 2 regressions already reported and solved on
>> upstream git.
>> I updated qemu to 2.0-rc2 where I did some tests without finding other
>> regressions or bugs for now.
>> I also noticed on the wiki page of the next hackathon the topic "QEMU and
>> SeaBIOS regression testing". As I reported some time ago it would be of
>> great help.

> I was thinking of submitting the update for QEMU when QEMU 2.0 will be
> out.

Wouldn't it be sensible for xen-unstable to follow at least the qemu rc's so those 
get build and tested in the osstests ?

> But thanks for the reminder.

Re: Update seabios and upstream qemu versions used on xen-unstable

[George Dunlap]

On 04/15/2014 12:51 PM, Sander Eikelenboom wrote:
>
> Tuesday, April 15, 2014, 1:18:32 PM, you wrote:
>
>> On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
>>> I think it would be good to update seabios and upstream qemu versions used
>>> with xen-unstable.
>>> Recently I tested on my testing system with xen-unstable seabios updated to
>>> debian package version 1.7.4-4 without finding regressions and upstream qemu
>>> 2.0-rc0 from git where I found 2 regressions already reported and solved on
>>> upstream git.
>>> I updated qemu to 2.0-rc2 where I did some tests without finding other
>>> regressions or bugs for now.
>>> I also noticed on the wiki page of the next hackathon the topic "QEMU and
>>> SeaBIOS regression testing". As I reported some time ago it would be of
>>> great help.
>
>> I was thinking of submitting the update for QEMU when QEMU 2.0 will be
>> out.
>
> Wouldn't it be sensible for xen-unstable to follow at least the qemu rc's so those
> get build and tested in the osstests ?

It might make more sense to have a osstest target that used vanilla 
upstream -- in fact, there may be one already.

  -George

Re: Update seabios and upstream qemu versions used on xen-unstable

[Sander Eikelenboom]

Tuesday, April 15, 2014, 3:24:15 PM, you wrote:

> On 04/15/2014 12:51 PM, Sander Eikelenboom wrote:
>>
>> Tuesday, April 15, 2014, 1:18:32 PM, you wrote:
>>
>>> On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
>>>> I think it would be good to update seabios and upstream qemu versions used
>>>> with xen-unstable.
>>>> Recently I tested on my testing system with xen-unstable seabios updated to
>>>> debian package version 1.7.4-4 without finding regressions and upstream qemu
>>>> 2.0-rc0 from git where I found 2 regressions already reported and solved on
>>>> upstream git.
>>>> I updated qemu to 2.0-rc2 where I did some tests without finding other
>>>> regressions or bugs for now.
>>>> I also noticed on the wiki page of the next hackathon the topic "QEMU and
>>>> SeaBIOS regression testing". As I reported some time ago it would be of
>>>> great help.
>>
>>> I was thinking of submitting the update for QEMU when QEMU 2.0 will be
>>> out.
>>
>> Wouldn't it be sensible for xen-unstable to follow at least the qemu rc's so those
>> get build and tested in the osstests ?

> It might make more sense to have a osstest target that used vanilla 
> upstream -- in fact, there may be one already.

I don't know if by "vanilla" you mean an vanilla upstream stable .. or a vanilla 
upstream devel ?

Because Xen has a (soft) dependency on qemu for HVM .. i would expect Xen as a 
project to test ... regularly devel ... or at least RC's from devel ... to make sure that what 
becomes a qemu-stable actually works with Xen (and have some time to actually 
get fixes upstream in the event that things don't work out).

It shouldn't be a blocker for the push-gates, but as an earlier warning things are go 
a miss in relation to one of the dependencies .. and things need attention 
preferably before that dependency project is going to release a stable version 
that would end up in distro's. 


The same goes for linux-linus and linux-next, although linux-next can be very 
unpolished and full of non really xen related oopses and build failures.

Perhaps an "linux-xen-next" (so that would be only xen patches targeted for 
linux-next) combined with linux-linus would be even more valuable .. 
since the xen linux tree at git.kernel.org has pretty clear branchnames this is already
pretty good manual testable (in later linux RC series, i mostly pull the xen 
tree on top).

But perhaps in the future also an easy pull-able branch for the linux net/net-next 
patches (because these are sometimes a PITA to test beforehand due to: many 
subsequent patch sessies .. and merge conflicts) would be nice and handy for 
automatic testing and that would also facilitate manual testing of the net 
patches before they get applied.

--
Sander

>   -George

Re: Update seabios and upstream qemu versions used on xen-unstable

[Fabio Fantoni]

Il 15/04/2014 15:24, George Dunlap ha scritto:
> On 04/15/2014 12:51 PM, Sander Eikelenboom wrote:
>>
>> Tuesday, April 15, 2014, 1:18:32 PM, you wrote:
>>
>>> On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
>>>> I think it would be good to update seabios and upstream qemu 
>>>> versions used
>>>> with xen-unstable.
>>>> Recently I tested on my testing system with xen-unstable seabios 
>>>> updated to
>>>> debian package version 1.7.4-4 without finding regressions and 
>>>> upstream qemu
>>>> 2.0-rc0 from git where I found 2 regressions already reported and 
>>>> solved on
>>>> upstream git.
>>>> I updated qemu to 2.0-rc2 where I did some tests without finding other
>>>> regressions or bugs for now.
>>>> I also noticed on the wiki page of the next hackathon the topic 
>>>> "QEMU and
>>>> SeaBIOS regression testing". As I reported some time ago it would 
>>>> be of
>>>> great help.
>>
>>> I was thinking of submitting the update for QEMU when QEMU 2.0 will be
>>> out.
>>
>> Wouldn't it be sensible for xen-unstable to follow at least the qemu 
>> rc's so those
>> get build and tested in the osstests ?
>
> It might make more sense to have a osstest target that used vanilla 
> upstream -- in fact, there may be one already.
>
>  -George

Yes is the vanilla upstream test that is missing, at least testing each 
stable version from rc0 can be a start but test daily or weekly git 
master would be perfect.
Now only upstream qemu of xen git is tested but is old of some upstream 
versions ago and the merging with newer versions for now were always 
after a long time out of the stable upstream version :(

Re: Update seabios and upstream qemu versions used on xen-unstable

[Konrad Rzeszutek Wilk]

On Tue, Apr 15, 2014 at 04:33:22PM +0200, Fabio Fantoni wrote:
> Il 15/04/2014 15:24, George Dunlap ha scritto:
> >On 04/15/2014 12:51 PM, Sander Eikelenboom wrote:
> >>
> >>Tuesday, April 15, 2014, 1:18:32 PM, you wrote:
> >>
> >>>On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
> >>>>I think it would be good to update seabios and upstream qemu
> >>>>versions used
> >>>>with xen-unstable.
> >>>>Recently I tested on my testing system with xen-unstable
> >>>>seabios updated to
> >>>>debian package version 1.7.4-4 without finding regressions
> >>>>and upstream qemu
> >>>>2.0-rc0 from git where I found 2 regressions already
> >>>>reported and solved on
> >>>>upstream git.
> >>>>I updated qemu to 2.0-rc2 where I did some tests without finding other
> >>>>regressions or bugs for now.
> >>>>I also noticed on the wiki page of the next hackathon the
> >>>>topic "QEMU and
> >>>>SeaBIOS regression testing". As I reported some time ago it
> >>>>would be of
> >>>>great help.
> >>
> >>>I was thinking of submitting the update for QEMU when QEMU 2.0 will be
> >>>out.
> >>
> >>Wouldn't it be sensible for xen-unstable to follow at least the
> >>qemu rc's so those
> >>get build and tested in the osstests ?
> >
> >It might make more sense to have a osstest target that used
> >vanilla upstream -- in fact, there may be one already.
> >
> > -George
> 
> Yes is the vanilla upstream test that is missing, at least testing
> each stable version from rc0 can be a start but test daily or weekly
> git master would be perfect.
> Now only upstream qemu of xen git is tested but is old of some
> upstream versions ago and the merging with newer versions for now
> were always after a long time out of the stable upstream version :(

Fabio,

Is that something you would be comfortable taking a look at? I fear
that a lot of us are busy with some of the hypervisor development and
this is kind of on the end of the plate - which means we would probably
get to it but not right away :-(

Re: Update seabios and upstream qemu versions used on xen-unstable

[Fabio Fantoni]

[-- Attachment #1.1: Type: text/plain, Size: 3516 bytes --]

2014-04-15 19:59 GMT+02:00 Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>:

> On Tue, Apr 15, 2014 at 04:33:22PM +0200, Fabio Fantoni wrote:
> > Il 15/04/2014 15:24, George Dunlap ha scritto:
> > >On 04/15/2014 12:51 PM, Sander Eikelenboom wrote:
> > >>
> > >>Tuesday, April 15, 2014, 1:18:32 PM, you wrote:
> > >>
> > >>>On Mon, Apr 14, 2014 at 01:11:36PM +0200, Fabio Fantoni wrote:
> > >>>>I think it would be good to update seabios and upstream qemu
> > >>>>versions used
> > >>>>with xen-unstable.
> > >>>>Recently I tested on my testing system with xen-unstable
> > >>>>seabios updated to
> > >>>>debian package version 1.7.4-4 without finding regressions
> > >>>>and upstream qemu
> > >>>>2.0-rc0 from git where I found 2 regressions already
> > >>>>reported and solved on
> > >>>>upstream git.
> > >>>>I updated qemu to 2.0-rc2 where I did some tests without finding
> other
> > >>>>regressions or bugs for now.
> > >>>>I also noticed on the wiki page of the next hackathon the
> > >>>>topic "QEMU and
> > >>>>SeaBIOS regression testing". As I reported some time ago it
> > >>>>would be of
> > >>>>great help.
> > >>
> > >>>I was thinking of submitting the update for QEMU when QEMU 2.0 will be
> > >>>out.
> > >>
> > >>Wouldn't it be sensible for xen-unstable to follow at least the
> > >>qemu rc's so those
> > >>get build and tested in the osstests ?
> > >
> > >It might make more sense to have a osstest target that used
> > >vanilla upstream -- in fact, there may be one already.
> > >
> > > -George
> >
> > Yes is the vanilla upstream test that is missing, at least testing
> > each stable version from rc0 can be a start but test daily or weekly
> > git master would be perfect.
> > Now only upstream qemu of xen git is tested but is old of some
> > upstream versions ago and the merging with newer versions for now
> > were always after a long time out of the stable upstream version :(
>
> Fabio,
>
> Is that something you would be comfortable taking a look at? I fear
> that a lot of us are busy with some of the hypervisor development and
> this is kind of on the end of the plate - which means we would probably
> get to it but not right away :-(
>

>From qemu 1.3 I kept testing the upstream qemu as far as I could, to find and
report any regressions with xen.
Unfortunately I did not have enough time and in some cases the knowledge to
fix and/or improve some things to try to ensure a good integration, use of
the main new qemu features, and exemptions from criticism regressions with
xen before of the new qemu stable versions release.
An important first step, perhaps the most important would be to keep the
tested with qemu upstream xen during the development of each version (
http://git.qemu.org/?p=qemu.git;a=shortlog;h=refs/heads/master) in
order to immediately
identify any regression with xen and resolve as soon as possible or at
least report
it on qemu-devel.
I gave a quick look at the git osstest to eventually try to use it on a
test server and run necessary automated tests for the nights but I did not
understand enough and I have almost no knowledge of perl, and perhaps even
more my very bad english...

My first doubts about osstest for example:
must to have 2 test systems (osstest controller and test xen host) on a network
that is without others dhcp and dns servers outside of the controller
but connected
to the internet or am I wrong?
There is a method to easy and fast have working dns/dhcp/tftp enough to
osstest?

Thanks for any reply and sorry for my bad english.

[-- Attachment #1.2: Type: text/html, Size: 7879 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Update seabios and upstream qemu versions used on xen-unstable

[Wei Liu]

On Tue, Apr 15, 2014 at 09:41:20PM +0200, Fabio Fantoni wrote:
[...]
> >
> 
> >From qemu 1.3 I kept testing the upstream qemu as far as I could, to find and
> report any regressions with xen.
> Unfortunately I did not have enough time and in some cases the knowledge to
> fix and/or improve some things to try to ensure a good integration, use of
> the main new qemu features, and exemptions from criticism regressions with
> xen before of the new qemu stable versions release.
> An important first step, perhaps the most important would be to keep the
> tested with qemu upstream xen during the development of each version (
> http://git.qemu.org/?p=qemu.git;a=shortlog;h=refs/heads/master) in
> order to immediately
> identify any regression with xen and resolve as soon as possible or at
> least report
> it on qemu-devel.
> I gave a quick look at the git osstest to eventually try to use it on a
> test server and run necessary automated tests for the nights but I did not
> understand enough and I have almost no knowledge of perl, and perhaps even
> more my very bad english...
> 

I presume you've checked out osstest related posts on blog.xen.org.
Given your requirement you might want to use it in "production" mode
which automatically runs testcases and do bisection. I don't think it's
very easy to setup, you might want to speak to Ian Campbell or Ian
Jackson.

> My first doubts about osstest for example:
> must to have 2 test systems (osstest controller and test xen host) on a network
> that is without others dhcp and dns servers outside of the controller
> but connected
> to the internet or am I wrong?

Not sure I understand this.

You can run all IT infrastructure and controller on the same host. The
minimal setup is two machines: one for runing testcase and the other for
everything else.

> There is a method to easy and fast have working dns/dhcp/tftp enough to
> osstest?
> 

DNS/DHCP/TFTP are generanl IT infrastructure, not part of OSSTest. But I
admit it's never easy to maintain them...

Wei.

> Thanks for any reply and sorry for my bad english.

> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

Re: Update seabios and upstream qemu versions used on xen-unstable

[Ian Campbell]

On Wed, 2014-04-16 at 10:33 +0100, Wei Liu wrote:
> On Tue, Apr 15, 2014 at 09:41:20PM +0200, Fabio Fantoni wrote:
> [...]
> > >
> > 
> > >From qemu 1.3 I kept testing the upstream qemu as far as I could, to find and
> > report any regressions with xen.
> > Unfortunately I did not have enough time and in some cases the knowledge to
> > fix and/or improve some things to try to ensure a good integration, use of
> > the main new qemu features, and exemptions from criticism regressions with
> > xen before of the new qemu stable versions release.
> > An important first step, perhaps the most important would be to keep the
> > tested with qemu upstream xen during the development of each version (
> > http://git.qemu.org/?p=qemu.git;a=shortlog;h=refs/heads/master) in
> > order to immediately
> > identify any regression with xen and resolve as soon as possible or at
> > least report
> > it on qemu-devel.
> > I gave a quick look at the git osstest to eventually try to use it on a
> > test server and run necessary automated tests for the nights but I did not
> > understand enough and I have almost no knowledge of perl, and perhaps even
> > more my very bad english...
> > 
> 
> I presume you've checked out osstest related posts on blog.xen.org.
> Given your requirement you might want to use it in "production" mode
> which automatically runs testcases and do bisection. I don't think it's
> very easy to setup, you might want to speak to Ian Campbell or Ian
> Jackson.

Adding a new branch flight is a bit harder to do without access to the
infra. You can see my libvirt patch for an example of the sort of thing
which is needed.

I can't speak for IanJ but I'm pretty busy with various things,
including the libvirt flight for osstest. I can probably look into doing
seabios at some point though.

> > There is a method to easy and fast have working dns/dhcp/tftp enough to
> > osstest?
> > 
> 
> DNS/DHCP/TFTP are generanl IT infrastructure, not part of OSSTest. But I
> admit it's never easy to maintain them...

At home i use dnsmasq which is pretty simple and provides at least the
first two (DNS+DHCP) I think it can do TFTP too but I tend to us the
regular tftpd on a difference host for various tedious reasons...

Ian.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Fabio Fantoni]

Il 16/04/2014 12:04, Ian Campbell ha scritto:
> On Wed, 2014-04-16 at 10:33 +0100, Wei Liu wrote:
>> On Tue, Apr 15, 2014 at 09:41:20PM +0200, Fabio Fantoni wrote:
>> [...]
>>> >From qemu 1.3 I kept testing the upstream qemu as far as I could, to find and
>>> report any regressions with xen.
>>> Unfortunately I did not have enough time and in some cases the knowledge to
>>> fix and/or improve some things to try to ensure a good integration, use of
>>> the main new qemu features, and exemptions from criticism regressions with
>>> xen before of the new qemu stable versions release.
>>> An important first step, perhaps the most important would be to keep the
>>> tested with qemu upstream xen during the development of each version (
>>> http://git.qemu.org/?p=qemu.git;a=shortlog;h=refs/heads/master) in
>>> order to immediately
>>> identify any regression with xen and resolve as soon as possible or at
>>> least report
>>> it on qemu-devel.
>>> I gave a quick look at the git osstest to eventually try to use it on a
>>> test server and run necessary automated tests for the nights but I did not
>>> understand enough and I have almost no knowledge of perl, and perhaps even
>>> more my very bad english...
>>>
>> I presume you've checked out osstest related posts on blog.xen.org.
>> Given your requirement you might want to use it in "production" mode
>> which automatically runs testcases and do bisection. I don't think it's
>> very easy to setup, you might want to speak to Ian Campbell or Ian
>> Jackson.
> Adding a new branch flight is a bit harder to do without access to the
> infra. You can see my libvirt patch for an example of the sort of thing
> which is needed.
>
> I can't speak for IanJ but I'm pretty busy with various things,
> including the libvirt flight for osstest. I can probably look into doing
> seabios at some point though.
>
>>> There is a method to easy and fast have working dns/dhcp/tftp enough to
>>> osstest?
>>>
>> DNS/DHCP/TFTP are generanl IT infrastructure, not part of OSSTest. But I
>> admit it's never easy to maintain them...
> At home i use dnsmasq which is pretty simple and provides at least the
> first two (DNS+DHCP) I think it can do TFTP too but I tend to us the
> regular tftpd on a difference host for various tedious reasons...
>
> Ian.
>

Thanks for your reply.
Trying to test myself if what I want try when I'll have some time.
For controller domU o phisical pc with debian 7 and dnsmasq and tftp 
from official package can be working and good?
About server for dom0 testing I can't at home beacause I have the 
computer in the room where I sleep, I can on office but probably there 
is a problem of other dhcp and dns server of samba 4 pdc on the only and 
same network.
For osstest should be not have other dns and dhcp server (or probably 
only dhcp) on the same network, right?

Thanks for any reply and sorry for my bad english.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Ian Campbell]

On Wed, 2014-04-16 at 13:22 +0200, Fabio Fantoni wrote:

> Thanks for your reply.
> Trying to test myself if what I want try when I'll have some time.
> For controller domU o phisical pc with debian 7 and dnsmasq and tftp 
> from official package can be working and good?

You might need to teach Osstest/DhcpWatch/leases.pm about the dnsmasq
leases file format if it differs from ISC DHCP, or just use the latter.

> About server for dom0 testing I can't at home beacause I have the 
> computer in the room where I sleep, I can on office but probably there 
> is a problem of other dhcp and dns server of samba 4 pdc on the only and 
> same network.
> For osstest should be not have other dns and dhcp server (or probably 
> only dhcp) on the same network, right?

osstest is quite happy to use existing DNS and DHCP, it doesn't require
its own.

All that is needed is some way for it to see the DHCP leases, for
example our local DHCP servers will cat their complete leases file over
a known port which osstest then uses.

Ian.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Fabio Fantoni]

Il 16/04/2014 14:14, Ian Campbell ha scritto:
> On Wed, 2014-04-16 at 13:22 +0200, Fabio Fantoni wrote:
>
>> Thanks for your reply.
>> Trying to test myself if what I want try when I'll have some time.
>> For controller domU o phisical pc with debian 7 and dnsmasq and tftp
>> from official package can be working and good?
> You might need to teach Osstest/DhcpWatch/leases.pm about the dnsmasq
> leases file format if it differs from ISC DHCP, or just use the latter.
>
>> About server for dom0 testing I can't at home beacause I have the
>> computer in the room where I sleep, I can on office but probably there
>> is a problem of other dhcp and dns server of samba 4 pdc on the only and
>> same network.
>> For osstest should be not have other dns and dhcp server (or probably
>> only dhcp) on the same network, right?
> osstest is quite happy to use existing DNS and DHCP, it doesn't require
> its own.
>
> All that is needed is some way for it to see the DHCP leases, for
> example our local DHCP servers will cat their complete leases file over
> a known port which osstest then uses.
>
> Ian.
>

Thanks for reply.
This is probably a stupid question/idea: osstest require dhcp and dns 
only for pxe boot on dom0 test systems or also for other things?
If is only for dom0 server, where in my case is only one and always the 
same, add dhcp ip reservation to dom0 mac and manual add dns record of 
dom0 will make me able to fast use osstest without adding other dhcp/dns 
or waste time to configure them for osstest?
Sorry if this is a stupid question/idea.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Ian Campbell]

On Wed, 2014-04-16 at 16:04 +0200, Fabio Fantoni wrote:
> Il 16/04/2014 14:14, Ian Campbell ha scritto:
> > On Wed, 2014-04-16 at 13:22 +0200, Fabio Fantoni wrote:
> >
> >> Thanks for your reply.
> >> Trying to test myself if what I want try when I'll have some time.
> >> For controller domU o phisical pc with debian 7 and dnsmasq and tftp
> >> from official package can be working and good?
> > You might need to teach Osstest/DhcpWatch/leases.pm about the dnsmasq
> > leases file format if it differs from ISC DHCP, or just use the latter.
> >
> >> About server for dom0 testing I can't at home beacause I have the
> >> computer in the room where I sleep, I can on office but probably there
> >> is a problem of other dhcp and dns server of samba 4 pdc on the only and
> >> same network.
> >> For osstest should be not have other dns and dhcp server (or probably
> >> only dhcp) on the same network, right?
> > osstest is quite happy to use existing DNS and DHCP, it doesn't require
> > its own.
> >
> > All that is needed is some way for it to see the DHCP leases, for
> > example our local DHCP servers will cat their complete leases file over
> > a known port which osstest then uses.
> >
> > Ian.
> >
> 
> Thanks for reply.
> This is probably a stupid question/idea: osstest require dhcp and dns 
> only for pxe boot on dom0 test systems or also for other things?
> If is only for dom0 server, where in my case is only one and always the 
> same, add dhcp ip reservation to dom0 mac and manual add dns record of 
> dom0 will make me able to fast use osstest without adding other dhcp/dns 
> or waste time to configure them for osstest?

It uses DHCP for guests too. 

PXE isn't used for PV guests I suppose, but might be for HVM guests, not
sure.

> Sorry if this is a stupid question/idea.

It's not...

Re: Update seabios and upstream qemu versions used on xen-unstable

[Ian Jackson]

Ian Campbell writes ("Re: [Xen-devel] Update seabios and upstream qemu versions used on xen-unstable"):
> It uses DHCP for guests too. 

Indeed.  While it is perhaps possible to configure the host details
statically, this isn't possible for guests (of which there may be many
created ad hoc by individual test cases).

> PXE isn't used for PV guests I suppose, but might be for HVM guests, not
> sure.

We don't currently have any tests of PXE booting guests.  Perhaps we
should :-).

Thanks,
Ian.

Re: Update seabios and upstream qemu versions used on xen-unstable

[Ian Campbell]

On Thu, 2014-04-17 at 17:14 +0100, Ian Jackson wrote:
> > PXE isn't used for PV guests I suppose, but might be for HVM guests, not
> > sure.
> 
> We don't currently have any tests of PXE booting guests.  Perhaps we
> should :-).

s/Perhaps w/W/

;-)