Xen devel - Mar 2012 - Archiving XCI : Request for an Archivation Review for the XCI project

Hi everybody,

the XCI project appears to have been dormant for a while and there is no 
project lead for XCI today.  In line with Xen governance (see 
phttp://www.xen.org/projects/governance.html), I would like to first get 
community views on the health of the project and then schedule an 
archivation review.

Let me quote from the process: "The review is essentially a pitch why 
the project should be archived. The purpose of the review is not 
necessarily to archive a project, but also to provide a last opportunity 
for interested parties in the Xen.org community to save the project and 
step up. The Xen.org community manager will support efforts to save the 
project, should community members want to step up. There is the special 
case that a project has been completed: in this case the normal 
situation would be for the project lead to make the case, why this is so."

Right now, the project shows a number of issues:
a) There is little or no activity
b) There is no project lead
c) There is no mentor for the project
d) Maybe there is a case, that the project has actually been completed

The first step of addressing these issues, would be to find a mentor and 
new project lead for the XCI project. This could be a vendor expressing 
interest in leading the project, who would ultimately provide the 
project lead. The project lead could also be an individual. My 
suggestion is to first have an open discussion about the issues and see 
if somebody is willing to step up. If so, I will work with the vendor or 
individual who steps up. Failing this, we will schedule an archivation 
review.

My suggestion is, to allow 2 weeks for discussion. In other words, I 
will collate the input that I have received by April 12th. We will then 
schedule the actual review.

We have not had archivation reviews in the past, but I believe we should 
run this as a semi-public meeting. I.e. interested parties approach me 
and will be invited to the meeting.

Best Regards
Lars

On 3/29/12, Lars Kurth <lars.kurth@xen.org> wrote:
>
> Hi everybody,
>
> the XCI project appears to have been dormant for a while and there is no
> project lead for XCI today.  In line with Xen governance (see
> phttp://www.xen.org/projects/governance.html), I would like to first get
> community views on the health of the project and then schedule an
> archivation review.
>
> Let me quote from the process: "The review is essentially a pitch why
> the project should be archived. The purpose of the review is not
> necessarily to archive a project, but also to provide a last opportunity
> for interested parties in the Xen.org community to save the project and
> step up. The Xen.org community manager will support efforts to save the
> project, should community members want to step up. There is the special
> case that a project has been completed: in this case the normal
> situation would be for the project lead to make the case, why this is
so."
>
> Right now, the project shows a number of issues:
> a) There is little or no activity
> b) There is no project lead
> c) There is no mentor for the project
> d) Maybe there is a case, that the project has actually been completed
>
> The first step of addressing these issues, would be to find a mentor and
> new project lead for the XCI project. This could be a vendor expressing
> interest in leading the project, who would ultimately provide the
> project lead. The project lead could also be an individual. My
> suggestion is to first have an open discussion about the issues and see
> if somebody is willing to step up. If so, I will work with the vendor or
> individual who steps up. Failing this, we will schedule an archivation
> review.
>
> My suggestion is, to allow 2 weeks for discussion. In other words, I
> will collate the input that I have received by April 12th. We will then
> schedule the actual review.
>
> We have not had archivation reviews in the past, but I believe we should
> run this as a semi-public meeting. I.e. interested parties approach me
> and will be invited to the meeting.
>
While I am not necessarily wanting to take over the XCI project
(however, I might consider it), I am putting together my own XCI
distribution, OpenXCI. It is not based on the xen.org XCI project
(which is rather outdated), but instead will be based on a fork of the
free parts of the XCI toolstack from XenClient 2 running on a Debian
6.0 base. OpenXCI will be different from other desktop hypervisors in
that it will be just a multi-boot alternative without any support for
remote provisioning or management other than SSH (it''s intended for
people who can manage their own computers by themselves), but there is
no reason that the toolstack couldn''t be used elsewhere.

Currently, I have working graphics and input servers, and am using
them with a more or less stock Debian 6.0 system. I haven''t yet ported
the XCI toolstack because I am trying to find a more stable kernel
(pv-ops kernels seem to be rather unstable on my system, whereas the
XenClient 2 xenlinux kernel never crashed except when the drive
failed). The SourceForge page is at <http://sf.net/projects/openxci/>
(the graphics and input servers as well as the corresponding QEMU
patches are available there; I''ll post more information soon).

Andrew,

that is good news. I see a number of options going forward, which really 
depends on your preference and also what committers and maintainers in 
the community think:
1) OpenXCI as a new project hosted on Xen.org - starting as an 
incubation project as per http://www.xen.org/projects/governance.html 
and archive the old one. What you describe sounds sufficiently different.

2) OpenXCI as a separate project: Xen.org can help you promoting the 
project in various ways

If you want and have a bit more clarity on what you want to do and how 
you want to achieve things, I could arrange a call with you and walk you 
through the details.

Regards
Lars

On 30/03/2012 02:51, Andrew Warkentin wrote:
> On 3/29/12, Lars Kurth<lars.kurth@xen.org>  wrote:
>> Hi everybody,
>>
>> the XCI project appears to have been dormant for a while and there is
no
>> project lead for XCI today.  In line with Xen governance (see
>> phttp://www.xen.org/projects/governance.html), I would like to first
get
>> community views on the health of the project and then schedule an
>> archivation review.
>>
>> Let me quote from the process: "The review is essentially a pitch
why
>> the project should be archived. The purpose of the review is not
>> necessarily to archive a project, but also to provide a last
opportunity
>> for interested parties in the Xen.org community to save the project and
>> step up. The Xen.org community manager will support efforts to save the
>> project, should community members want to step up. There is the special
>> case that a project has been completed: in this case the normal
>> situation would be for the project lead to make the case, why this is
so."
>>
>> Right now, the project shows a number of issues:
>> a) There is little or no activity
>> b) There is no project lead
>> c) There is no mentor for the project
>> d) Maybe there is a case, that the project has actually been completed
>>
>> The first step of addressing these issues, would be to find a mentor
and
>> new project lead for the XCI project. This could be a vendor expressing
>> interest in leading the project, who would ultimately provide the
>> project lead. The project lead could also be an individual. My
>> suggestion is to first have an open discussion about the issues and see
>> if somebody is willing to step up. If so, I will work with the vendor
or
>> individual who steps up. Failing this, we will schedule an archivation
>> review.
>>
>> My suggestion is, to allow 2 weeks for discussion. In other words, I
>> will collate the input that I have received by April 12th. We will then
>> schedule the actual review.
>>
>> We have not had archivation reviews in the past, but I believe we
should
>> run this as a semi-public meeting. I.e. interested parties approach me
>> and will be invited to the meeting.
>>
> While I am not necessarily wanting to take over the XCI project
> (however, I might consider it), I am putting together my own XCI
> distribution, OpenXCI. It is not based on the xen.org XCI project
> (which is rather outdated), but instead will be based on a fork of the
> free parts of the XCI toolstack from XenClient 2 running on a Debian
> 6.0 base. OpenXCI will be different from other desktop hypervisors in
> that it will be just a multi-boot alternative without any support for
> remote provisioning or management other than SSH (it''s intended
for
> people who can manage their own computers by themselves), but there is
> no reason that the toolstack couldn''t be used elsewhere.
>
> Currently, I have working graphics and input servers, and am using
> them with a more or less stock Debian 6.0 system. I haven''t yet
ported
> the XCI toolstack because I am trying to find a more stable kernel
> (pv-ops kernels seem to be rather unstable on my system, whereas the
> XenClient 2 xenlinux kernel never crashed except when the drive
> failed). The SourceForge page is at<http://sf.net/projects/openxci/>
> (the graphics and input servers as well as the corresponding QEMU
> patches are available there; I''ll post more information soon).
>
> _______________________________________________
> Xci-devel mailing list
> Xci-devel@lists.xen.org
> http://lists.xen.org/cgi-bin/mailman/listinfo/xci-devel

On Thu, Mar 29, 2012 at 07:51:02PM -0600, Andrew Warkentin
wrote:
> 
> While I am not necessarily wanting to take over the XCI project
> (however, I might consider it), I am putting together my own XCI
> distribution, OpenXCI. It is not based on the xen.org XCI project
> (which is rather outdated), but instead will be based on a fork of the
> free parts of the XCI toolstack from XenClient 2 running on a Debian
> 6.0 base. OpenXCI will be different from other desktop hypervisors in
> that it will be just a multi-boot alternative without any support for
> remote provisioning or management other than SSH (it''s intended
for
> people who can manage their own computers by themselves), but there is
> no reason that the toolstack couldn''t be used elsewhere.
> 
Nice!
> Currently, I have working graphics and input servers, and am using
> them with a more or less stock Debian 6.0 system. I haven''t yet
ported
> the XCI toolstack because I am trying to find a more stable kernel
> (pv-ops kernels seem to be rather unstable on my system, whereas the
> XenClient 2 xenlinux kernel never crashed except when the drive
> failed). The SourceForge page is at <http://sf.net/projects/openxci/>
> (the graphics and input servers as well as the corresponding QEMU
> patches are available there; I''ll post more information soon).
> 
It''d be nice if you could capture the pvops dom0 kernel errors/logs,
so the problems can be fixed. Everyone should switch to pvops kernel by now,
so all the issues and rough edges can be sorted out!

You can use serial console to capture the boot messages/errors:
http://wiki.xen.org/wiki/Xen_Serial_Console

In the case of a laptop you can use SOL (if your hardware has it),
or you can get a suitable ExpressCard (PCIe) serial adapter.

-- Pasi

On 3/30/12, Lars Kurth <lars.kurth@xen.org> wrote:
> Andrew,
>
> that is good news. I see a number of options going forward, which really
> depends on your preference and also what committers and maintainers in
> the community think:
> 1) OpenXCI as a new project hosted on Xen.org - starting as an
> incubation project as per http://www.xen.org/projects/governance.html
> and archive the old one. What you describe sounds sufficiently different.
>
> 2) OpenXCI as a separate project: Xen.org can help you promoting the
> project in various ways
>
> If you want and have a bit more clarity on what you want to do and how
> you want to achieve things, I could arrange a call with you and walk you
> through the details.
>
I''m not quite sure which would be a better idea. I don''t
really have
much on the OpenXCI SourceForge project (just the Mercurial repository
for the graphics and input servers, as well as a tarball with the
corresponding qemu patches), so it might be easier to move it to
xen.org now rather than later. I do have a pretty good idea of what I
want OpenXCI to be - a configurable, high-performance multi-boot
alternative with good support for graphics passthrough, using a
similar architecture to XenClient (but using direct GPU passthrough
for one domain and either a "reverse graphics adapter" or a VNC
connection to display other domains, possibly with some kind of
support for 3D on non-passthrough domains using virtual GPUs being
added later).