This is the 8th meeting of the ServerTeam, starting at 15:00 UTC and finishing at 17:00 UTC


Items we will be discussing:

  • Review ACTION points from previous meeting.
  • JeOS support/documentation for gutsy, plans for hardy
  • sudo bug discussion - dendrobates
  • Hardy community projects - dendrobates
  • Review each section of the ServerTeam/Roadmap.

  • Agree on next meeting date and time.


Review ACTION points from previous meeting

zul wrote a wiki page detailing his plans for Xen in the next release: https://wiki.ubuntu.com/XenInHardy. The blueprint can be tracked in Launchpad at https://blueprints.launchpad.net/ubuntu/+spec/xen-hardy.

soren sent an email to debian-devel about the mta meta-package proposal. The thread turned into a discussion about which mta (postfix or exim) should be the default.

ACTION: soren will try to send another email about it.

sudo bug discussion

dendrobates raised the issue that some server installations wouldn't create a user that has sudo powers. Some reviews ran into that bug. The forums also have some threads about it. dendrobates and nijaba tried to reproduce the problem, but were not successful.

There are a couple of bugs in LP: https://bugs.launchpad.net/ubuntu/+source/user-setup/+bug/158952 https://bugs.launchpad.net/ubuntu/+source/user-setup/+bug/162638

dendrobates asked if people could do various test installs to check if this bug can be reproduced. If so, please comment on one of the above bug.

Hardy community projects

dendrobates asked for help on the Ruby On Rails specification: https://blueprints.launchpad.net/ubuntu/+spec/rubyonrails. The scpecification is being written in the wiki: https://wiki.ubuntu.com/RubyOnRailsStack.

lionel volunteered to take a look at it since he has some experience with RoR. He will contact david to start working on the draft. mathiaz stated that he could be the approver, but the spec needs to be written and approved by Thursday to get it included in Hardy.

Anyone interested in improving the Ruby on Rails stack for Ubuntu should comment on this spec.

ivoks, sommer and dantalizing offered some help for the implementation.

ACTION: lionel will contact david to write the ruby-on-rails spec.

JeOS support/documentation for gutsy, plans for hardy

nealmcb started to talk about JeOS (Just Enough Operating System). A long discussion followed. It seems that people are confused about what JeOS can be used for.

The first issue identified was lack of documentation. nijaba stated that he will write a tutorial about using JeOS to build a virtual appliance providing a web application.

ACTION: nijaba will write a tutorial about JeOS.

An important source of confusion is the difference between the script used to produce a customized virtual image (or iso) and the customization made to Ubuntu to run as a guest.

nealmcb suggested to create a launchpad team and a mailing list for users and developers of JeOS.

ACTION: mathiaz will answer nealmcb post on ubuntu-server.

Review each section of the ServerTeam/Roadmap

ivoks gave an update about integrating postfix and dovecot: to goal is to replace saslauthd with dovecot. This means adding 1 or 2 lines to postfix configuration when dovecot is installed. There was some discussion about how this should be done. mathiaz suggested to file a bug against dovecot to track work being done on this.

sommer mentioned that the documentation has been updated to use Dovecot SASL.

ACTION: ivoks will file a bug about this and will start working on the packaging bits.

mathiaz mentioned the state of fastcgi. He wanted to know what were the advantage of mod-fcgid over mod-fastcgi. The next step is to write a MIR for mod-fcgid to get it included into main.

ACTION: ivoks will write a MIR for mod-fcgid.

Other business

dholbach mentioned the sponsoring queue for server packages: http://people.ubuntu.com/~dholbach/sponsoring/.

soren pointed out the list of pending sru: http://people.ubuntu.com/~ubuntu-archive/pending-sru.html. There's a few server packages on there that need testing before they can make their way to -updates.

mathiaz also noted that the focus of the development team is currently on merges.

Next Meeting

mathiaz asked if another meeting was needed in one week (instead of every other week). The answer was positive.

The time has also been pushed back by one hour (due to DST).

The next meeting will be hold on Tuesday, November the 27th, at 16:00 UTC in #ubuntu-meeting.


  • before its mention

Started logging meeting in #ubuntu-meeting
[15:00:50] <pschulz01> mathiaz: +1
[15:01:52] <mathiaz> So I'd like to appoligize for the last meeting not being run.
[15:02:20] <mathiaz> some of the team members were not available to attend it.
[15:02:34] <mathiaz> it's been a long month since the last one.
[15:02:59] <mathiaz> I think it was before UDS.
[15:03:18] <ivoks> then we have lots of things to talk about
[15:03:37] * dholbach hugs the server folks :)
[15:03:41] <mathiaz> I hope so
[15:03:50] * soren hugs dholbach
[15:04:04] * ivoks turns to the dark side and hugs dholbach too :)
[15:04:11] <dantalizing> get a channel
[15:04:12] <mathiaz> the agenda for today is here: https://wiki.ubuntu.com/ServerTeam/Meeting
[15:04:22] <soren> dantalizing: :)
[15:04:29] <mathiaz> [TOPIC] Review ACTION points from previous meeting.
[15:04:53] <mathiaz> last meeting notes can be found here: https://wiki.ubuntu.com/MeetingLogs/Server/20071023
[15:05:07] <dholbach> more hugs for those of you who do reviews marked as 'canonical-server' on http://people.ubuntu.com/~dholbach/sponsoring/
[15:05:12] * dholbach is quiet now
[15:05:23] <mathiaz> zul: did you start a wiki page about your plan for xen ?
[15:05:39] <zul> mathiaz: yes i havent updated it since Ive started it
[15:05:47] <soren> zul: What's the name of it?
[15:05:52] <zul> its a launchpad spec
[15:06:00] <zul> gimme a sec..
[15:06:29] <zul> https://wiki.ubuntu.com/XenInHardy
[15:07:03] * pschulz01 greets sommer
[15:07:07] <mathiaz> zul: did you register a spec in LP ?
[15:07:09] <zul> thats the gist of it
[15:07:12] <zul> mathiaz: yep
[15:07:13] <sommer> hey, all
[15:07:30] <ivoks> zul: those xen releases patches for newer version of kernels?
[15:07:34] <ivoks> s/those/does
[15:07:35] <mathiaz> zul: https://blueprints.launchpad.net/ubuntu/+spec/xen-hardy gives me an error
[15:07:47] <zul> argh...ill fix it after the meeting
[15:07:52] <zul> ivoks: correct.
[15:08:12] <mathiaz> zul: ok - great !
[15:08:29] <dendrobates> mathiaz: have we started?
[15:08:35] <mathiaz> dendrobates: yes
[15:08:37] <soren> dendrobates: Yes :)
[15:08:41] <dendrobates> topic?
[15:08:49] <mathiaz> dendrobates: review last meeting points.
[15:08:56] <mathiaz> dendrobates: https://wiki.ubuntu.com/MeetingLogs/Server/20071023
[15:09:57] <mathiaz> I haven't sent an email about tagging bugs for documentation to the ubuntu-doc team.
[15:10:41] <soren> I had an action point, too.
[15:11:02] <soren> I sent the e-mail to debian-devel about the default MTA stuff.
[15:11:06] <mathiaz> soren: ah says. So did you send your email ?
[15:11:14] <mathiaz> soren: what was the response ?
[15:11:41] <soren> Despite my efforts to emphasize the process and work included, no-one responded to that bit, and everyone started talking about if exim or postfix should be the default one.
[15:12:02] <soren> ...so the discussion has gotten nowhere at all :(
[15:12:20] <mathiaz> soren: has the thread finished ?
[15:12:49] <soren> Wel... it's dead.
[15:13:41] <soren> I can't find it in the ml archive right now. :/
[15:14:19] <mathiaz> soren: may be you could send another email explaining that the default choice of the mta is a hot topic and that what you propose is a technical improvement to give choice the end user
[15:14:46] <mathiaz> soren: give choice to the end user or developer
[15:14:47] <soren> I thought about that, but I can already imagine the response. :)
[15:14:49] <soren> I'll try.
[15:16:12] <mathiaz> ACTION: soren will try to come up with a new email for debian-devel about the default MTA change.
[15:16:27] <soren> "but we can't do this before we've all agreed which mta should be our default".. and then the same discussion will start again.
[15:16:37] <soren> But I'm not bitter.
[15:16:40] <mathiaz> [ACTION] soren will try to come up with a new email for debian-devel about the default MTA change.
[15:16:42] <soren> Moving right along..
[15:16:48] <ivoks> soren: may the force be with you...
[15:17:04] <dendrobates> I thought we agreed on postfix in a previous meeting.
[15:17:12] <soren> dendrobates: *we* did. Debian didn't.
[15:18:01] <soren> dendrobates: ...and we'd really like to have Debian agree on the same technical way to do this, so that we don't go in different directions.
[15:18:16] <dendrobates> perhaps slangasek can help point us to the correct people to get on board before we make another post to the ml.
[15:18:17] <mathiaz> any other toughts about the last meeting ?
[15:18:32] <soren> dendrobates: Yeah, I could try talking to him.
[15:19:07] <soren> mathiaz: no
[15:19:21] <dendrobates> no
[15:19:59] <mathiaz> ok. Let's move on then.
[15:20:07] <mathiaz> [TOPIC] JeOS support/documentation for gutsy, plans for hardy
[15:20:14] <dendrobates> nealmcb added te next topic.
[15:20:29] <dendrobates> He does not seem to be with us.
[15:21:03] <soren> Maybe we can move to the next topic and return to this one if he shows up?
[15:21:05] <dendrobates> I would like to table this discussion until he can join us.
[15:21:16] <soren> snap
[15:21:18] <mathiaz> ok. Let's defer this discussion then.
[15:21:31] <mathiaz> [TOPIC] sudo bug discussion - dendrobates
[15:22:04] <dendrobates> There have been sporadic reports of the user added at install time not being in the admin group.
[15:22:23] <dendrobates> this causes sudo not to work for that user.
[15:22:35] <ivoks> dendrobates: any examples?
[15:22:42] <soren> dendrobates: Actually, one of the reports mentioned that it was the %admin ALL=(ALL) ALL line that was missing from sudoers..
[15:23:16] <pschulz01> dendrobates: URL?
[15:23:21] <soren> http://www.enterprisenetworkingplanet.com/netos/article.php/3712031
[15:23:23] <mathiaz> dendrobates: is there a bug in LP ?
[15:23:24] <jdstrand> didn't I hear someone say it happened only when installing the mail-server task?
[15:23:38] <dendrobates> The reports have all been a little different, and I cannot reproduce it.
[15:23:45] <jdstrand> (I cannot remember where I heard/read that)
[15:23:50] <mathiaz> I've also heard about some thread in the forums
[15:23:56] <dendrobates> jdstrand: yes, that was one report, but it was not reproducable.
[15:24:05] <ivoks> m... this article... i've read it
[15:24:21] <ivoks> we really shouldn't base assumptions on it
[15:24:24] <dendrobates> I don't put too much stock in that particulat report.
[15:24:49] <dendrobates> but it has popped up too pften to be a coincidence IMO
[15:24:54] <soren> http://ubuntuforums.org/showthread.php?t=590290
[15:25:47] <dendrobates> I would like everyone to try at least once this week to reproduce this. I have done 20+ installs with no luck.
[15:26:31] <jdstrand> dendrobates: I am curious if all of this has been in Vms?
[15:26:36] <jdstrand> VMs
[15:26:37] <mathiaz> dendrobates: may be we could try to centralize all the links that describe such a problem ?
[15:26:51] <dendrobates> mathiaz: good idea.
[15:26:55] <ivoks> jdstrand: or maybe even with wubi or someting like that...
[15:27:12] <sommer> is there a bug we can leave comments on?
[15:27:13] <dendrobates> jdstrand: I have tried to reproduce it in VM and real HW.
[15:27:14] <mathiaz> dendrobates: is there a bug in LP about this ?
[15:27:25] <akincer> I did a server install yesterday and did not run into this issue. Had a separate issue, but it isn't part of this discusion
[15:27:30] <dendrobates> mathiaz: not yet.
[15:27:47] <mathiaz> ok. So I'll file a bug about this and assign to the server team then.
[15:28:19] <mathiaz> [ACTION] mathiaz will file a bug for the sudo bug to track all the links mentioning it.
[15:28:44] <dendrobates> we can keep track of the various reports on the lp bug report.
[15:29:51] <dendrobates> https://bugs.launchpad.net/ubuntu/+source/user-setup/+bug/40684 perhaps they are using expert mode.
[15:29:53] <ubotu> Launchpad bug 40684 in user-setup "expert install user not a sudoer if root password given" [Medium,Confirmed]
[15:30:06] <ivoks> ah... expert mode
[15:30:07] <soren> Yeah, I just stumbled upon that.
[15:30:17] <soren> That is clearly intentional, though.
[15:30:48] <soren> The code very clearly checks if a root password has been set, and if so, no sudo-capable user is added.
[15:30:54] <dendrobates> quite old though. Should be looked at though. I did not try an expert install.
[15:31:10] <soren> https://bugs.edge.launchpad.net/ubuntu/+source/sudo/+bug/162638
[15:31:12] <ubotu> Launchpad bug 162638 in sudo "sudo - first user not in sudoers file" [Undecided,New]
[15:31:25] <zul> dendrobates: ill be trying a server install after the meeting so I can try to reproduce it
[15:31:36] <dendrobates> in the reported cases, no root password was set, allegedly
[15:32:26] <ivoks> i'll try expert install today
[15:32:35] <mathiaz> soren: it seems that this is a bug where we should track the links.
[15:32:35] <jdstrand> dendrobates: I haven't looked at that passwrod checking code. maybe it is as simple as a 'blank' password (eg, a space or tab or something)
[15:33:14] <dendrobates> jdstrand: it shouldn't be possible to inadvertantly do that from the installer
[15:33:23] <jdstrand> dendrobates: we should ask for /etc/shadow entry for root
[15:33:41] <jdstrand> or at least whether it has a password in it
[15:33:50] <dendrobates> anyway, I just wanted everyone to be aware.
[15:33:55] <jdstrand> (as opposed to '*' or '!'
[15:34:23] <mathiaz> I've just subscribed ubuntu-server to bug 162638
[15:34:24] <ubotu> Launchpad bug 162638 in sudo "sudo - first user not in sudoers file" [Undecided,New] https://launchpad.net/bugs/162638
[15:34:37] <dendrobates> mathiaz: thanks.
[15:34:49] <mathiaz> I think it should be used to track all the information we found about this issue.
[15:34:57] <dendrobates> agreed
[15:35:35] <mathiaz> dendrobates: anything else about this issue ?
[15:36:02] <dendrobates> no
[15:36:38] <mathiaz> let's move on to the next topic then
[15:36:45] <mathiaz> [TOPIC] Hardy community projects - dendrobates
[15:37:25] <dendrobates> we would like a volunteer to pick up a project for hardy.
[15:37:42] <dendrobates> We would like to add a ruby on rails stack.
[15:38:19] <dendrobates> This would consist of adding the necessary packages to main ,and creating a tasksel task
[15:38:36] <ivoks> anyone knows anything about ruby?
[15:38:52] <dendrobates> this is perfect for someone going for core-dev.
[15:38:58] <zul> not me
[15:39:13] <coffeedude> nope.
[15:39:14] <dendrobates> or someone that wants to learn ruby.
[15:39:17] <sommer> I've done some script type programming with Ruby... cool language
[15:39:28] <dendrobates> or someone that knows ruby
[15:39:56] <nealmcb> ahh - daylight savings - damn
[15:40:20] <dendrobates> When everyone volunteers at once, I can't hear you.
[15:40:32] <ivoks> well
[15:40:40] <ivoks> i could do everything needed on apache side
[15:40:45] <sommer> I'm not haven't really done much packaging, but can probably help
[15:40:52] <ivoks> but i really don't know anything about gems and other ruby stuff
[15:41:14] <mathiaz> there was another ruby spec registered in LP.
[15:41:17] <sommer> s/I haven't/
[15:41:36] <dantalizing> I would be willing to help but would need some direction
[15:41:41] <mathiaz> I can contact the person that registered the spec to see if he is interested in doing it.
[15:42:01] <dantalizing> a lot of direction
[15:42:14] <akincer> Pardon me, but it seems the question worth asking is "Does anyone here who knows anything about Ruby on Rails want to volunteer to head up a RoR project for Gutsy?"
[15:42:32] <mathiaz> akincer: s/gutsy/hardy/
[15:42:36] <akincer> doh
[15:42:44] <akincer> No coffee today, what do you expect?
[15:43:04] * coffeedude thought that was an implicit question anyways....
[15:43:08] <mathiaz> well - the first thing is to figure out what is the current state of ror in ubuntu.
[15:43:09] <sommer> is there a link to the spec?
[15:43:41] <nealmcb> [sorry to get my clocks messed up and miss so much of the meeting. can someone shoot me a transcript somehow of the meeting so far?]
[15:44:04] <mathiaz> sommer: https://blueprints.edge.launchpad.net/ubuntu/+spec/rubyonrails
[15:44:15] <jdstrand> nealmcb: the JeOS stuff was postponed til you got here
[15:44:19] <nealmcb> :-)
[15:44:23] <sommer> mathiaz: thx
[15:44:59] <mathiaz> the actual wiki page is https://wiki.ubuntu.com/RubyOnRailsStack
[15:45:09] <lionel> mathiaz: all is free, nothing has been chossen like mongrel versus fast-cgi or something like that. This topic was discused at UDS?
[15:45:42] <mathiaz> lionel: a little bit - but there wasn't enough knowledge around the table.
[15:45:54] <dendrobates> this is actually a different spec. But we can use it. I assigned jdstrand to create the spec, but since this exists, we can use it.
[15:46:15] <dendrobates> Does anyone know who David Portwood is?
[15:46:15] <lionel> due to some nightmare in the past for me to install the RoR stack, I'm fine with helping on this topic :)
[15:46:15] <jdstrand> yea
[15:46:43] <mathiaz> lionel: well you could help with drafting the spec then.
[15:47:05] <mathiaz> lionel: which is just about figure out what's wrong with the current way of installing ror.
[15:47:18] <lionel> I'll try to contact David first.
[15:48:06] <soren> someone's at the door.. brb.
[15:48:14] <lionel> mathiaz: more or less : nothing is packaged and all is fech with ruby packaging system :-(
[15:48:33] <mathiaz> lionel: david == dzportwood ?
[15:48:41] <lionel> but I agreee that's something that's necessary
[15:48:56] <dendrobates> We need the spec completed by thursday, which is not alot of time.
[15:49:04] <lionel> mathiaz: yes. to know if he want to help us to implement :)
[15:49:21] <lionel> dendrobates: well, there is nothing at tv tonight ;)
[15:49:34] <mathiaz> lionel: ok. So could you try to contact him ?
[15:49:43] <dendrobates> lionel: Ha, thanks.
[15:49:57] <lionel> yes, when the meeting is finished I send him a mail and I will digg in the spec
[15:50:03] <mathiaz> lionel: and since we may be short on time, could also add your thoughts to a discussion section on the wiki page ?
[15:50:19] <lionel> I have not a lot of experience with specs, I may require a bit help from you
[15:50:25] <lionel> mathiaz: sure, no problem
[15:50:49] <mathiaz> lionel: don't worry. I can be the approver of the spec.
[15:51:31] <mathiaz> [ACTION] lionel will contact david to write the ruby-on-rails spec.
[15:51:47] <dendrobates> ok, that is an action item, lionel will work on the spec, and sommer, ivoks, and dantalizing will all help with the implementation.
[15:52:10] * soren is back
[15:52:25] <dendrobates> we should go back to JEOS now that nealmcb is here.
[15:52:27] <mathiaz> dendrobates: let's get the spec drafted and approved first.
[15:52:58] <mathiaz> ok. Let's get back to JEOS then.
[15:53:00] <dendrobates> true, but I want to capture who volunteered, so we don't have to ask again.
[15:53:00] <nealmcb> dendrobates: thanks
[15:53:24] <mathiaz> [TOPIC] JeOS support/documentation for gutsy, plans for hardy
[15:53:34] <mathiaz> [TOPIC] JeOS support/documentation for gutsy, plans for hardy
[15:53:40] <nealmcb> did folks read my email on the topic?
[15:53:57] <nealmcb> any comments on that? should I go thru it? someone else want to lead this item?
[15:54:13] <soren> It would be wonderful if you could lead it.
[15:54:25] <mathiaz> nealmcb: I've read your email.
[15:54:41] <nealmcb> I think documentation is a big part - lots of folks confused about what is out there
[15:55:03] <nealmcb> ...for gutsy. pretty easy to fix with a wiki page at help.ubuntu.com - right?
[15:55:04] <soren> Heck, *I*'m confused sometimes!
[15:55:59] <mathiaz> nealmcb: yes. Your proposal to create https://help.ubuntu.com/community/JeOS makes sense.
[15:56:25] <dendrobates> nealmcb: the question becomes what do we say about Jeos
[15:56:27] <nealmcb> a big question in my mind is whether we should somehow get the ubuntu-jeos-builder out there for gutsy somehow
[15:56:40] <soren> nealmcb: I plan on renaming that, by the way.
[15:56:42] <nealmcb> since it is so much nicer than the iso
[15:56:51] <soren> ...to remove a big part of the confusion in my head.
[15:56:57] <soren> To ubuntu-vm-builder or something.
[15:57:21] <nealmcb> dendrobates: what are the options?
[15:57:29] <nealmcb> ..for what we say...
[15:57:42] <soren> ...and when I'm done looking at virt-install I'll figure out if I'll roll them into one, or upload both of them. It'll only be a few days.
[15:58:07] <dendrobates> technically, I'm not sure we have much of a story at this point.
[15:58:15] <nealmcb> soren: right
[15:58:34] <nealmcb> I was also looking at rbuilder rpath.com last night - some very cool stuff there...
[15:59:04] <mathiaz> fwiw it can be used to quickly create virtual machines with specific packages installed.
[15:59:17] <nealmcb> dendrobates: my sense is that the momentum is now, and if we don't at least get the community involved in helping make jeos work well on ubuntu, they'll go elsewhere
[15:59:31] <mathiaz> ex: I want an ldap server using hardy.
[15:59:52] <dendrobates> nealmcb: ok. agreed
[15:59:58] <nealmcb> so a mailing list makes lots of sense, and more transparency from the folks working on it
[16:00:40] <nealmcb> there is a perception that canonical somehow is trying to make this a proprietary thing, which makes no sense to me
[16:00:47] <mathiaz> nealmcb: more transparency means having a roadmap
[16:00:48] * soren has just reproduced the no-user-in-sudoers bug.
[16:01:00] <soren> er... no.
[16:01:03] <soren> never mind :)
[16:01:16] * pschulz01 smiles at soren.
[16:01:20] * nijaba guesses that he forgot to update his meeting time after DST change
[16:01:33] <nealmcb> nijaba: you and me too....
[16:01:46] <akincer> OT, but DST needs to die
[16:01:48] <nealmcb> nijaba: the topic is jeos now
[16:01:53] <mathiaz> nealmcb: do you think a team is needed ?
[16:02:09] <zul> well how far along is jeos?
[16:02:13] <nealmcb> mathiaz: you mean launchpad team?
[16:02:18] <nealmcb> probably
[16:02:37] <mathiaz> nealmcb: well - not necessarly an LP team.
[16:02:42] <nealmcb> but also mailing list for users and developers. I hate talking about this on the forums....
[16:02:57] <mathiaz> nealmcb: if you want to have users joining the project, a team in the wiki makes more sense.
[16:02:58] <nealmcb> who wants to work on it?
[16:03:01] <dendrobates> zul: it exists, but there is no interesting technology, except the scritp that builds it.
[16:03:17] <soren> dendrobates: Well.. no.
[16:03:34] <soren> dendrobates: The script that builds the JeOS iso not exceptionally non-interesting.
[16:03:49] <soren> er.. s/not/is/
[16:03:59] <pookey> hm, I'm an hour late :)
[16:04:01] <nealmcb> soren: I disagree!
[16:04:06] <dendrobates> soren: at least it is technology.
[16:04:11] <soren> The ubuntu-jeos-builder (which is sort of unrelated to JeOS, actually) is rather interesting, though.
[16:04:15] <nijaba> soren: exceptionally non-uninteresting.
[16:04:21] <nealmcb> ahh - right
[16:04:40] <dendrobates> isn't that what I said?
[16:04:40] <nealmcb> though i'd like to see how the jeos iso is built.....
[16:04:48] <mathiaz> I think that the script to build iso is interesting.
[16:04:56] * nealmcb nods
[16:05:02] <mathiaz> It can be used as a base to build virtual appliances.
[16:05:14] <nealmcb> but why an iso? why not a vm?
[16:05:24] * nijaba nods as nealmcb
[16:05:24] <akincer> And I, for one, would like a plethora of documentation on how to use that
[16:05:27] <soren> mathiaz: It's no different from any other cd building script, really.
[16:05:29] <mathiaz> to me it's just another output.
[16:05:56] <mathiaz> some user want to create an iso, others want to create a vm.
[16:06:08] <nealmcb> are any doc folks here?
[16:06:09] <nijaba> Iwas asked to write an article on how to build an appliance based on JeOS
[16:06:24] <nijaba> So I guess that will be a start for a doc
[16:06:27] * sommer waves at nealmcb
[16:06:33] <nealmcb> :-)
[16:06:46] <soren> mathiaz: If the vm builder should be part of this, we need a new name for it. A generic vm builder script does not match "Just enough OS" in any sane way, IMO.
[16:06:59] <soren> mathiaz: I see the relation of the two things, though.
[16:07:00] <pschulz01> nijaba: I would be happy to review the article.
[16:07:04] <dendrobates> We could automate appliance building ala vmware vadk.
[16:07:28] <zul> I agree a vm would be more useful than an iso
[16:07:36] <nealmcb> soren: though jeos is a catchy name now. but I see your point
[16:07:39] <mathiaz> nijaba: would you mind writing a wiki page for your articile ?
[16:07:48] <nijaba> mathiaz: not at all
[16:08:14] <nijaba> soren: I will need some help from you to get started.
[16:08:19] <mathiaz> so what about starting a document on help.u.c ?
[16:08:23] <zul> for example most ISP would use a base VM rather than an ISO to install images for clients
[16:08:26] <akincer> Not everyone would find a vm more useful than an iso. Despite that being the trend, I can think of many uses of an iso as an appliance install
[16:08:32] <nijaba> mathiaz: perfect
[16:08:42] <soren> nijaba: Sure.
[16:08:49] <mathiaz> nijaba: do you already have a skeleton in mind ?
[16:09:03] <zul> or have a switch that either builds an iso or a vm
[16:09:11] <nealmcb> having easy tips on how to dpkg-reconfigure after the vm comes up (or before?) would help
[16:09:35] <nealmcb> akincer: what use case are you thinking of ?
[16:09:39] <nijaba> mathiaz: for a_z step by step install install of a given web app (which I have not picked)
[16:09:49] <mathiaz> it seems that we also need a way to capture all the other ideas of improvements
[16:10:10] <soren> bugs.lp.net/ubuntu-jeos/+new ?
[16:10:14] <dendrobates> we are technically out of time.
[16:10:15] <akincer> nealmcb: Old machines that would do fine as a simple appliance like a DNS server
[16:10:57] <mathiaz> dendrobates: is there another meeting scheduled now ?
[16:11:07] <nealmcb> akincer: I though jeos explicitely meant vm in people's minds, and putting lots of hw support for old hardware in there would be a different focus
[16:11:11] <nijaba> akincer: the problem is that the JeOS kernel is stripped of most drivers that you will need
[16:11:25] <nealmcb> does anyone have to leave now?
[16:11:25] <dendrobates> I don't think so, I was just stating the obvious.
[16:11:47] <akincer> wouldn't the point of having a builder be to add in things like drivers? Maybe I don't understand enough about JeOS
[16:11:53] <mathiaz> soren: is there a spec for jeos in hardy ?
[16:12:10] <soren> mathiaz: Well... Depends on what you mean by "jeos".
[16:12:11] <soren> See?
[16:12:24] <soren> mathiaz: If it's for Ubuntu as a guest os, then yes.
[16:12:33] <mathiaz> I'm trying to figure out where we can keep track of the proposed improvements
[16:12:46] <soren> Bugs on launchpad?
[16:13:03] <mathiaz> soren: ok.
[16:13:12] <nealmcb> but a roadmap in a wiki spec would be more helpful for many folks
[16:13:12] <soren> https://blueprints.edge.launchpad.net/ubuntu/+spec/jeos-hardy is about Ubuntu as a guest os
[16:13:25] <mathiaz> I'll also add a section about it in the ServerTeam Roadmap.
[16:13:29] <nijaba> mathiaz: I think that JeOS is and should stay as Ubuntu as a guest os, but need to increase the number of virtualization environments supported
[16:13:35] <soren> There's currently no spec about the vm builder thing.
[16:13:47] <akincer> If I'm simply uninformed, ignore my assertion
[16:14:10] <mathiaz> akincer: if you're uninformed, we need to fix it.
[16:14:15] <soren> akincer: You rarely care much about extra drivers in a vm, do you?
[16:14:33] <nijaba> akincer: your point is good, but that would be an appliance builder vs a virtual appliance builder
[16:14:40] <nijaba> we certainly could use both
[16:14:50] <zul> soren: actually if i had a weird usb key drive then yes I would
[16:14:54] <akincer> soren: Right, but I was specifically talking about building an appliance for say an unused machine to be a DNS machine
[16:14:59] <soren> zul: How weird?
[16:15:17] <mathiaz> akincer: that's out of the scope of JeOS.
[16:15:26] <zul> soren: usb thumbscanner or something like that
[16:15:29] <soren> akincer: Ah, I thought you still meant in a virtual machine, but on old hardware.
[16:16:09] <zul> soren: or a wireless driver that the kernel doesnt support but there are drivers out there that does support it
[16:16:13] <akincer> Is there a link to the scope of JeOS?
[16:16:30] <mathiaz> akincer: not that know of. That's what we need to fix
[16:16:30] <akincer> wait, think I see it
[16:16:31] <soren> zul: Wireless driver? In a VM?
[16:16:37] <zul> soren: as an exmample
[16:16:47] <soren> zul: I fail to see the use of that.
[16:17:07] <zul> soren: people have done weirder things
[16:17:11] <mathiaz> so to get back on track, what can we do ?
[16:17:30] <nealmcb> document what we offer in gutsy
[16:17:34] <mathiaz> nijaba will write a short tutorial about using jeos to build a virtual appliances.
[16:17:39] <nealmcb> backport some vm builder to gutsy
[16:17:50] <nealmcb> build a community around it
[16:17:59] <nealmcb> and go for the gold in hardy
[16:19:10] <mathiaz> anyone interested in writing a simple wiki page that presents and explain what is JeOS ?
[16:19:20] <nealmcb> getting rid of unneeded app-level stuff seems like an important goal to me, and a hard one
[16:19:24] <mathiaz> or would this part of nijaba's tutorial ?
[16:19:34] <nijaba> mathiaz: will be part of it
[16:19:47] <nealmcb> the jeos page should list various related pages, and include our working definition
[16:19:48] <nijaba> in fact, I have already started this part internally
[16:20:02] <mathiaz> it looks like the bigest problem now is documentation. let's focus on this first.
[16:20:08] <nealmcb> I've put various hints on using the jeos builder on that forums page
[16:20:11] <nijaba> preparing the eb page for u.c/server/jeos
[16:20:28] <nealmcb> and put the ubuntu-jeos-builder in my ppa
[16:20:43] <nealmcb> but don't want to go to far without a plan....
[16:20:56] <mathiaz> nealmcb: can you copy your hints on h.u.c/community/JeOS ?
[16:21:11] <nealmcb> absolutely - though I think they would be in a sub-page
[16:21:39] <soren> Ok, let me just get this straigt:
[16:21:40] <nealmcb> and picking the vm builder script of choice is a top priority so I look forward to soren's thoughts on virt-install etc
[16:21:46] <nealmcb> does virt-install require x11?
[16:21:58] <zul> no
[16:22:06] <soren> I'm the only one who's annoyed that JeOS refers to two completely separate things? Ie. both Ubuntu as a guest os and the vm builder thingie?
[16:22:35] <mralphabet> and third a physical machine os
[16:22:36] <nijaba> so let's have JeOS and JeOS-Builder
[16:22:36] <mathiaz> soren: nope. There is some confusion.
[16:22:52] <nealmcb> a jeos builder can also build simple vms right?
[16:23:04] <nealmcb> but a vm builder might not be very savvy about tight jeos images
[16:23:08] <soren> mathiaz: Then can we please, please, please not put anything about the vm builder script onto h.u.c/whatever/JeOS/whatever ?
[16:23:15] <mathiaz> soren: JeOS is ubuntu as a virtual guest.
[16:23:25] <nealmcb> so what is wrong with using the hip jeos term?
[16:23:31] <mathiaz> soren: vm builder script is part of the virtualization spec.
[16:23:31] <akincer> If they are two entirely different things, I don't think having similar names is conducive to clarity
[16:23:47] <soren> mathiaz: Not the virtualisation spec I wrote :)
[16:23:55] <soren> mathiaz: I can be, but it isn't.
[16:23:59] <nealmcb> I'm just talking about the name of the tool. ubuntu-jeos-builder now (a bit long)
[16:24:06] <nealmcb> jeos-builder would be better perhaps
[16:24:10] <soren> vm-builder!
[16:24:21] * nijaba beeps -> trademark
[16:24:25] <nealmcb> vm's are so 1990's...
[16:24:30] <nealmcb> :-)
[16:24:39] <mathiaz> soren: ok. It seems that vm----builder would build on JeOS
[16:24:49] <soren> I just hate that every conversation I have about JeOS has to start with: "So, when you say JeOS, do you mean the entirely separate installabal ISO, or do you mean the wicked cool vm builder script, I wrote?"
[16:24:50] <zul> i would have said virt-factory
[16:25:00] <nealmcb> I'm not firm on that opinion, but I think jeos has mindshare now. it jeos trademarked?
[16:25:23] <soren> nealmcb: Good question.
[16:25:24] <dendrobates> nealmcb: I don't know.
[16:25:46] * zul must go to work
[16:25:57] <nealmcb> nijaba: you brought up trademarks - in relation to what?
[16:26:07] <nijaba> to vm-xxxx
[16:26:23] <nealmcb> is vm-xxxx trademarked?
[16:26:42] <nijaba> nope, but vm- is very close to vmware
[16:26:46] <nealmcb> ah
[16:27:20] <nijaba> so I would suggest getting away from it if we want to do stuff outside of vmware
[16:27:23] <nealmcb> regardless of what we call it - will we package "jeos-builder" for gutsy somehow?
[16:27:31] <soren> I doubt it.
[16:27:32] <nealmcb> (other than my ppa :-)
[16:27:39] <mathiaz> ok. so it seems that this is still confusing. and we still have some things to talk about.
[16:27:39] <soren> Not through any official channels anyway.
[16:27:49] <mathiaz> let's get this to the mailing list.
[16:27:57] * pschulz01 says: goodnight all. will ne reading the logs
[16:27:58] <mathiaz> I'll reply to your email nealmcb
[16:28:01] <nealmcb> but it is so cool - that is where we get traction!
[16:28:10] * nijaba +1 to mathiaz proposal
[16:28:15] <nealmcb> 90 seconds to a new vm!
[16:28:24] <mathiaz> with a summary of the previous conversation.
[16:28:35] <nealmcb> great
[16:28:45] <mathiaz> [ACTION] mathiaz will reply to nealmcb mail about jeos.
[16:29:14] <mathiaz> [TOPIC] Review each section of the ServerTeam/Roadmap.
[16:29:34] <akincer> Gotta go. Goodbye and good luck.
[16:29:38] <mathiaz> The Roadmap: https://wiki.ubuntu.com/ServerTeam/Roadmap
[16:30:13] <mathiaz> nealmcb: how is the factoids going ?
[16:30:39] * soren wonders if mathiaz lives in a hexagonal flat
[16:30:48] <nealmcb> I have't put any time into them - jeos is too much fun and work is pressing. but I'll get back to them
[16:30:55] <soren> :)
[16:30:55] <ivoks> mathiaz: what about postfix+dovecot?
[16:31:21] <ivoks> mathiaz: i've tested it and i can't say anything else than 'works as advertised'
[16:31:34] <nealmcb> and I'll work on a jeos factoid :-)
[16:31:39] * mathiaz checks that his flat hasn't changed drastically during this meeting
[16:31:51] <ivoks> a long one :)
[16:32:01] <lamont> ivoks: dovecot should "just work" with the current postfix
[16:32:09] * nijaba states that mathiaz has actually moved out of the hexagon
[16:32:24] <soren> mathiaz: I'm just curious where the ".* corner" terminology comes from.
[16:32:34] <ivoks> lamont: we were talking about SASL over dovcot in postfix
[16:32:39] <mathiaz> soren: that comes from jono
[16:32:46] <soren> mathiaz: oic
[16:33:11] * nealmcb finally gets the hexagon joke :-)
[16:33:22] <lamont> ivoks: right
[16:33:34] <lamont> and it should "just work". If not, iz bug and should be filed
[16:33:34] <ivoks> lamont: so, we would like to provide that out of the box
[16:33:47] <ivoks> lamont: it does just work :)
[16:33:49] <mathiaz> ivoks: what is the goal of this ?
[16:34:09] <ivoks> mathiaz: we talked about that at UDS
[16:34:17] <ivoks> mathiaz: replacing saslauthd with dovecot
[16:34:21] <mathiaz> ivoks: yes - it's been a while since then.
[16:34:36] <mathiaz> ivoks: right.
[16:34:52] <mathiaz> ivoks: so what needs to be done ?
[16:34:56] <sommer> just to note the docs have been updated to use Dovecot SASL.
[16:35:07] <ivoks> mathiaz: we need to add 1 or 2 lines in main.cf
[16:35:09] <sommer> for Hardy anyway
[16:35:11] <nijaba> mathiaz: also update the mail task ?
[16:35:16] <mathiaz> ivoks: you said you've tested it - is there anything that can be done improve it ?
[16:35:28] <mathiaz> sommer: great ! thanks.
[16:35:34] <sommer> np
[16:35:36] <ivoks> mathiaz: yes, we could provide it out of the box?
[16:36:11] <mathiaz> ivoks: where should this be done ? in the dovecot postinst script ?
[16:36:29] <ivoks> that's the problem i wanted to disccuss
[16:36:43] <mathiaz> ivoks: OTOH I'm not sure we can mangle the postfix configuration file
[16:36:46] <ivoks> we can't do it in postinst, since that will break debian policy :/
[16:36:53] <soren> No.
[16:36:59] <lamont> mathiaz: I'd be happy to mangle the config file
[16:37:06] <ivoks> we talked about tasksel postinst, and it's the same thing...
[16:37:07] <soren> postfix provides postconf to alter its config, so all should be good.
[16:37:17] <lamont> ivoks: in dovecot postinst?
[16:37:19] <soren> SEction 11.7, IIRC.
[16:37:24] <mathiaz> lamont: what would you suggest ?
[16:37:24] <lamont> yeah - that just means that we need an interface
[16:37:30] <ivoks> soren: that's ok, yes...
[16:37:31] <soren> Ah, 10.7.
[16:37:39] <soren> 10.7.4 in particular.
[16:37:50] <soren> lamont: Interface?
[16:38:06] <ivoks> lamont: dovecot postinst could check if postconf exsist and then use it to set up sasl
[16:38:22] <soren> Hm... It would be nice if it asked first. :)
[16:38:32] <ivoks> yeah...
[16:38:33] <soren> A simple yes/no, though.
[16:38:45] <lamont> and 10.7.4 says postconf -e will make things right
[16:38:46] <mathiaz> soren: OTOH this is one more question
[16:39:04] <lamont> note that modifying config files does require that you ask first
[16:39:04] <ivoks> that's why i was thinking about new package
[16:39:09] <ivoks> ubuntu-mail-server
[16:39:23] <ivoks> wich would depend and replace postifx and dovecot
[16:39:34] <lamont> ivoks: ew
[16:39:42] <soren> mathiaz: ...that's not much of an issue, as it turns out.
[16:39:44] <lamont> it shouldn't need to replace
[16:39:51] <mathiaz> ivoks: tasksel seems like a good candidate for that
[16:39:52] <sommer> I have another meeting... thanks all
[16:39:54] <ivoks> replace means 'that's my config too, and i can edit it'
[16:39:57] <lamont> mathiaz: it's not a question in base install
[16:40:06] <soren> mathiaz: We discussed this earlier (mysql root password stuff).
[16:40:21] <lamont> ivoks: if it uses postconf, then it's not editing its config, and it's policy compliant
[16:40:40] <mathiaz> ok. It seems that it needs to be discussed a little bit more.
[16:40:45] <ivoks> sorry, i forgot that we need to change dovecot's config also
[16:40:48] <mathiaz> ivoks: could you file a bug for that ?
[16:40:56] <mathiaz> ivoks: against dovecot.
[16:41:00] <ivoks> ok
[16:41:07] <lamont> also, please note that neither main.cf nor master.cf is a conffile, nor should they ever be.
[16:41:10] <mathiaz> ivoks: Are you willing to do the packaging work also ?
[16:41:15] <ivoks> mathiaz: yes
[16:41:36] * soren hugs ivoks
[16:41:50] * nijaba hugs him too
[16:42:04] <mathiaz> ivoks: attach your debdiff and we'll keep discussing the technical bits in the bug.
[16:42:11] <ivoks> ok
[16:42:18] <lamont> ivoks: and don't replace dovecot or postfix, please.
[16:42:36] <mathiaz> [ACTION] ivoks will file a bug to work on dovecot and postfix integration.
[16:42:42] <ivoks> lamont: ok, i'll change dovecot's config and postconf postfix
[16:43:10] <lamont> ivoks: if there's a way to export an interface from dovecot, that'd be the win tere.
[16:43:11] <lamont> there.
[16:43:32] <ivoks> i know...
[16:44:48] <mathiaz> ivoks: on a related note, I've looked at fast-cgi
[16:45:03] <mathiaz> ivoks: the plan was to move fast-cgid to main
[16:45:12] <ivoks> um... fcgid
[16:45:15] <ivoks> not fast-cgi
[16:45:27] <mathiaz> ivoks: I've described this in the webapplication spec.
[16:45:30] <mathiaz> ivoks: yes.
[16:45:37] <ivoks> ok
[16:46:05] <soren> Yeah, -mod-fastcgi iz evil.
[16:46:13] <mathiaz> ivoks: I had one issue: Why choose mod_fcgid over mod_fastcgi ?
[16:46:14] <ivoks> i'm also eager to do packaging work for that
[16:46:25] <ivoks> it's evil :)
[16:46:28] <soren> mathiaz: mod_fastcgi is non-free.
[16:46:44] <mathiaz> soren: yes - I've already mentionned that.
[16:46:55] <mathiaz> is there another reason ?
[16:47:04] <soren> mathiaz: ...but used to be the only way to get apache to do fastcgi, which in turn is the reason why the switch away from mod_php[45] hasn't happened a loooong time ago.
[16:47:19] <soren> mathiaz: That's not enough? :)
[16:47:22] <mathiaz> what about performance ?
[16:47:38] <nijaba> and security ?
[16:48:13] <ivoks> i will look into all that
[16:48:13] <mathiaz> ivoks: the next step for this is to write a MIR, which needs that sort of information.
[16:48:21] <ivoks> and write about it
[16:48:29] <mathiaz> ivoks: ok. great !
[16:48:31] <ivoks> i know
[16:49:29] <mathiaz> [ACTION]: ivoks will write a MIR to include mod-fcgid in main.
[16:49:54] <nealmcb> by the way for those that came late like me: http://kryten.incognitus.net/mootbot/meetings/ubuntu-meeting.20071120_1500.html
[16:50:06] <nealmcb> and http://kryten.incognitus.net/mootbot/meetings/ubuntu-meeting.log.20071120_1500.html
[16:50:28] <mathiaz> Is there any comments on the ServerTeam Roadmap ?
[16:51:27] <nealmcb> I love how mootbot keeps meeting notes in real-time :-)
[16:51:31] <ivoks> roadmap to heaven :)
[16:53:49] <mathiaz> ok. We've almost run out of time.
[16:53:51] <ivoks> i'll also take a look at some modules for apache
[16:54:01] <mathiaz> do we need another meeting in one week ?
[16:54:03] <ivoks> which aren't in ubuntu, but are life savers...
[16:54:17] <nealmcb> mathiaz: sounds useful to me
[16:54:36] <nijaba> +1
[16:54:48] <ivoks> VOTE
[16:54:52] <mathiaz> we can schedule one next week at 16:00 UTC
[16:55:05] <nealmcb> is there a conflict with 15:00?
[16:55:22] <mathiaz> 15:00 is a bit too early for the west coast people.
[16:55:30] <soren> Slackers.
[16:55:31] * nealmcb nods
[16:55:34] <nijaba> specially for mathiaz
[16:55:39] <mathiaz> we used to run the meeting at 15:00 UTC but with DST it has changed
[16:56:01] <mathiaz> the kernel team irc meeting is at 17:00 UTC
[16:56:03] <nealmcb> is 1600 bad in the summertime? or should we (gasp) schedule it in some DST-using timezone for calendar sanity?
[16:56:32] * nealmcb needs a calendar that can schedule in UTC....
[16:56:43] <mathiaz> ok. So next meeting will be next week at 16:00 UTC
[16:56:56] <mathiaz> same place.
[16:57:05] <soren> I'd just like to direct everyone's attention at http://people.ubuntu.com/~ubuntu-archive/pending-sru.html .. There's a few server packages on there that need testing before they can make their way to -updates.
[16:57:24] <nealmcb> and you can test some with vms I bet :-)
[16:57:26] <soren> If people could test them and report on the relevant bugs if it works for them, that would be a great help.
[16:57:36] <soren> nealmcb: Oh, yeah.
[16:57:55] <dendrobates> I have to go to another meeting, I'll check the logs for further developments.
[16:57:57] <mathiaz> and don't forget the merges.
[16:58:08] <mathiaz> dendrobates: well we've finished.
[16:58:24] <mathiaz> thanks all for your participation
[16:58:35] <nijaba> thanks !
[16:58:41] <nealmcb> thanks!
[16:58:49] <mathiaz> #endmeeting
Meeting ended.

CategoryScribesTeam CategoryTemplate

MeetingLogs/Server/20071120 (last edited 2008-08-06 16:59:43 by localhost)