Ubuntu/Debian support?
So, a quick non-scientific poll... Motivation: It has been suggested that adding ubuntu or debian distro support could broaden the reach of the oVirt community. To date, there has been some work towards this goal, but it is not "coming to a conclusion". Questions that comes to mind: 1) Is this still an issue? 2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact? 3) Would it have maintainers catering to it so it won't be left behind / delay development? 4) Why bother with host support, ovirt-node can be used? 5) Why bother with engine support, a virtual appliance or a docker image could be used? 6) if we do it, should we focus on Ubuntu or Debian distro first? 7) if we do it, should we focus on host or engine first? Thoughts? Thanks, Itamar
So, a quick non-scientific poll...
Motivation: It has been suggested that adding ubuntu or debian distro support could broaden the reach of the oVirt community. To date, there has been some work towards this goal, but it is not "coming to a conclusion".
Questions that comes to mind:
1) Is this still an issue? I think that to support other serious distros would highlight the fact
On 17/11/14 04:41, Itamar Heim wrote: that oVirt is distro agnostic. Today oVirt still is "RedHat-centric"
2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact? 3) Would it have maintainers catering to it so it won't be left behind / delay development? 4) Why bother with host support, ovirt-node can be used? 5) Why bother with engine support, a virtual appliance or a docker image could be used? 6) if we do it, should we focus on Ubuntu or Debian distro first? Debian, of course, it is upstream of Ubuntu and its release criteria compares to Centos or RHEL 7) if we do it, should we focus on host or engine first? I think that implementing host in Debian/Ubuntu would involve less work and after all, node/host is the core of virtualization infrastructure.
Thoughts?
Thanks, Itamar _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
On 17/11/14 08:41, Itamar Heim wrote:
1) Is this still an issue?
2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact? In general: Software which supports multiple operating systems/distros tends to be more stable, because some bugs just get found on specific
3) Would it have maintainers catering to it so it won't be left behind / delay development?
4) Why bother with host support, ovirt-node can be used? My impression is, that most users don't use ovirt-node, because
Yes of course, or is it implemented yet? platforms. the general ovirt technology is available "everywhere" (TM): java & python this is a question for devel@ovirt.org (CC'ed ;) ) this project has not enough dev power and lacks certain stuff
5) Why bother with engine support, a virtual appliance or a docker image could be used? docker is inherently insecure (a famous red hatter said: it's like download random code from the internet and run it as root), so docker is no option today, imho. 6) if we do it, should we focus on Ubuntu or Debian distro first? well, the general rule of thumb is: if it runs on debian it runs on ubuntu. I see way more debian than ubuntu servers, but for modern deployments this changes. my personal opinion is, that ubuntu sucks as a server os. 7) if we do it, should we focus on host or engine first? atm you are focusing host, there are some bugs open for that and it's planned for 3.6. (that's what I read at least)
General note: Imho ovirt would greatly benefit from more supported distros. Not just debian and ubuntu, but you have to start somewhere. you can attract many devs, which just don't use el/fedora. furthermore, the basic technology runs on any linux distribution: java and python so these are "just" integration issues (path to configfiles, different init systems, packaging not in rpm, but deb, etc.) This is still a huge effort, but I guess canonical and various debian developers would maybe join the effort to make ovirt work on their platform. my 2 cent -- Mit freundlichen Grüßen / Regards Sven Kieske Systemadministrator Mittwald CM Service GmbH & Co. KG Königsberger Straße 6 32339 Espelkamp T: +49-5772-293-100 F: +49-5772-293-333 https://www.mittwald.de Geschäftsführer: Robert Meyer St.Nr.: 331/5721/1033, USt-IdNr.: DE814773217, HRA 6640, AG Bad Oeynhausen Komplementärin: Robert Meyer Verwaltungs GmbH, HRB 13260, AG Bad Oeynhausen
Il 17/11/2014 14:22, Sven Kieske ha scritto:
On 17/11/14 08:41, Itamar Heim wrote:
1) Is this still an issue?
Yes of course, or is it implemented yet?
2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact? In general: Software which supports multiple operating systems/distros tends to be more stable, because some bugs just get found on specific platforms. the general ovirt technology is available "everywhere" (TM): java & python 3) Would it have maintainers catering to it so it won't be left behind / delay development? this is a question for devel@ovirt.org (CC'ed ;) ) 4) Why bother with host support, ovirt-node can be used? My impression is, that most users don't use ovirt-node, because this project has not enough dev power and lacks certain stuff 5) Why bother with engine support, a virtual appliance or a docker image could be used? docker is inherently insecure (a famous red hatter said: it's like download random code from the internet and run it as root), so docker is no option today, imho. 6) if we do it, should we focus on Ubuntu or Debian distro first? well, the general rule of thumb is: if it runs on debian it runs on ubuntu. I see way more debian than ubuntu servers, but for modern deployments this changes. my personal opinion is, that ubuntu sucks as a server os. 7) if we do it, should we focus on host or engine first? atm you are focusing host, there are some bugs open for that and it's planned for 3.6. (that's what I read at least)
General note:
Imho ovirt would greatly benefit from more supported distros. Not just debian and ubuntu, but you have to start somewhere. you can attract many devs, which just don't use el/fedora.
furthermore, the basic technology runs on any linux distribution: java and python
so these are "just" integration issues (path to configfiles, different init systems, packaging not in rpm, but deb, etc.)
This is still a huge effort, but I guess canonical and various debian developers would maybe join the effort to make ovirt work on their platform.
I would really like to see them joining the effort! :-)
my 2 cent
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
On 11/17/2014 02:41 AM, Itamar Heim wrote:
So, a quick non-scientific poll...
Motivation: It has been suggested that adding ubuntu or debian distro support could broaden the reach of the oVirt community. To date, there has been some work towards this goal, but it is not "coming to a conclusion".
Questions that comes to mind:
1) Is this still an issue? 2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact? 3) Would it have maintainers catering to it so it won't be left behind / delay development? 4) Why bother with host support, ovirt-node can be used? 5) Why bother with engine support, a virtual appliance or a docker image could be used? 6) if we do it, should we focus on Ubuntu or Debian distro first? 7) if we do it, should we focus on host or engine first?
Thoughts?
Thanks, Itamar _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
I think this is a must. Non-scientifically, Debian/Ubuntu is much more popular on desktops than Fedora. Because of that, it's quickly gaining traction in server roles as well. Developers will resist even trying out oVirt if they can't test it out on their Debian/Ubuntu desktop. I had several people ask me at LISA if oVirt runs on Ubuntu. 4 and 5 are valid points. I have light experience packaging for Debian. I offer my help :) -- Greg Sheremeta Red Hat, Inc. Sr. Software Engineer, RHEV Cell: 919-807-1086 gshereme@redhat.com
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: users@ovirt.org Sent: Monday, November 17, 2014 9:41:32 AM Subject: [ovirt-users] Ubuntu/Debian support?
So, a quick non-scientific poll...
Motivation: It has been suggested that adding ubuntu or debian distro support could broaden the reach of the oVirt community. To date, there has been some work towards this goal, but it is not "coming to a conclusion".
Questions that comes to mind:
1) Is this still an issue?
Define "issue". iow, define what our goal is. If it's "make it possible to easily run ovirt on debian", that's one thing, perhaps easily achieved by your points 4-5 below. If it's "have ovirt packaged inside the Debian archive" that's a considerably different issue. I think we should aim for the latter, the former is a start.
2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact?
imho yes. I agree with the other replies already posted.
3) Would it have maintainers catering to it so it won't be left behind / delay development?
For Debian, generally speaking, it's up to us to support. AFAIK almost all Debian developers are volunteers, and do whatever they find interesting or need for their own use. So we should either try to contact existing such developers dealing with virtualisation-related packages there and see how interested they are in ovirt, or try to do this on our own. To get things into the archive you do not necessarily need to be a developer yourself, they have a system of mentors etc. for that. For Ubuntu, if Canonical is interested, they are of course more than welcome to help! This will obviously help also Debian, although AFAIK not necessarily every package that is accepted to the Ubuntu archive is automatically or easily accepted to Debian's.
4) Why bother with host support, ovirt-node can be used?
As long as we see value in developing and supporting hosts based on el/fedora, this value also exists for Debian-based hosts. If we decide one day to support only ovirt-node, it's not as important what distro it's based on.
5) Why bother with engine support, a virtual appliance or a docker image could be used?
Because many users will find it a higher barrier to implement, for various reasons.
6) if we do it, should we focus on Ubuntu or Debian distro first?
IMO Debian, I agree with other replies already posted.
7) if we do it, should we focus on host or engine first?
Not sure I can see a strong preference for one of them. I'd like to add my own question, since none of the replies so far replied to it: 8. Do we have here existing users that are interested in this? If it worked equally well, would you (re?)install on Debian? Because if the answer is 'No', we might be discussing this on the wrong list, and better approach Debian lists/forums/etc. -- Didi
On 11/18/2014 03:54 AM, Yedidyah Bar David wrote:
Define "issue". iow, define what our goal is. If it's "make it possible to easily run ovirt on debian", that's one thing, perhaps easily achieved by your points 4-5 below. If it's "have ovirt packaged inside the Debian archive" that's a considerably different issue. I think we should aim for the latter, the former is a start.
I'm fine with "for" not "inside" - that's the current status for fedora and we are not there yet even for centos.
8. Do we have here existing users that are interested in this? If it worked equally well, would you (re?)install on Debian?
Because if the answer is 'No', we might be discussing this on the wrong list, and better approach Debian lists/forums/etc.
that's one of the reasons i qualified this as a "quick non-scientific poll"...
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Yedidyah Bar David" <didi@redhat.com> Cc: users@ovirt.org Sent: Tuesday, November 18, 2014 10:58:08 AM Subject: Re: [ovirt-users] Ubuntu/Debian support?
On 11/18/2014 03:54 AM, Yedidyah Bar David wrote:
Define "issue". iow, define what our goal is. If it's "make it possible to easily run ovirt on debian", that's one thing, perhaps easily achieved by your points 4-5 below. If it's "have ovirt packaged inside the Debian archive" that's a considerably different issue. I think we should aim for the latter, the former is a start.
I'm fine with "for" not "inside" - that's the current status for fedora and we are not there yet even for centos.
IIRC 3.1 was (is?) in the archive. The fact that recent versions aren't, is a different issue. oVirt is at least currently developing quite fast, and the release processes of distributions mean that users that use only stuff from the official archive will get older versions, which I think for oVirt wasn't very acceptable until recently, perhaps still is - I do not think anyone (here or elsewhere) would like to support 3.1, as recent versions are so much better. Even if we do not aim for inclusion in the official archive any time soon, we can still try to make our packages as nice as possible for that - including e.g. carefully checking the Debian policy documents and adhering to them, try to consult experienced Debian packagers for whatever pitfalls we should try to avoid, etc. All of this is not as important if you merely want to make using ovirt on Debian possible or easy.
8. Do we have here existing users that are interested in this? If it worked equally well, would you (re?)install on Debian?
Because if the answer is 'No', we might be discussing this on the wrong list, and better approach Debian lists/forums/etc.
that's one of the reasons i qualified this as a "quick non-scientific poll"...
Not sure this question makes it any more scientific :-) -- Didi
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: users@ovirt.org Sent: Monday, November 17, 2014 8:41:32 AM Subject: [ovirt-users] Ubuntu/Debian support?
So, a quick non-scientific poll...
Motivation: It has been suggested that adding ubuntu or debian distro support could broaden the reach of the oVirt community. To date, there has been some work towards this goal, but it is not "coming to a conclusion".
Questions that comes to mind:
1) Is this still an issue?
On my opinion yes.
2) Can we afford to dilute the focus we have as it is hard enough to stabilize the currently supported distro's? is it worth the potential impact?
Probably yes, it can help attracting new users.
3) Would it have maintainers catering to it so it won't be left behind / delay development?
Yes, on a lower priority.
4) Why bother with host support, ovirt-node can be used?
Yes, but for a new potential user it's probably easier, or at least more familiar, to take a look at the project running it on a distro he well know.
5) Why bother with engine support, a virtual appliance or a docker image could be used?
As 4.
6) if we do it, should we focus on Ubuntu or Debian distro first?
From Linux counter data ( http://linuxcounter.net/distributions.html ), Ubuntu seams to be today most popular distribution. I'm not really sure about the statistics value of that data cause the statistics base is really low and it doesn't distinct desktop usage from server one, but probably if the aim is to broader the user base it would be more interesting to start from Ubuntu. Maybe it would be also easier to involve somebody from Canonical.
7) if we do it, should we focus on host or engine first?
Both, maybe focusing also on all-in-one setup which, of course is not the most enterprise grade setup, but it could be more interesting for Ubuntu target users.
Thoughts?
Thanks, Itamar _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
participants (7)
-
Federico Alberto Sayd -
Greg Sheremeta -
Itamar Heim -
Sandro Bonazzola -
Simone Tiraboschi -
Sven Kieske -
Yedidyah Bar David