20080624

Agenda

Items we will be discussing:

  • Review ACTION points from previous meeting.
  • Relocation of web pages from /var/www/ to /srv/www/ in future releases LukeHasNoName

  • Intrepid spec status (still - when will stuff get approved?) - ScottKitterman

  • Iso testing - 8.04.1 and 8.10-alpha1 - MathiasGug

  • Open Discussion.
  • Agree on next meeting date and time.

Minutes

Ebox and augeas

mathiaz started a thread on the ebox-devel mailing list about using augeas in the backend code. Response has been positive so far and upstream developers will look into it.

On a related note, nxvl packaged augeas. He uploaded it to revu and is looking for comments.

text browser on the server cd

mathiaz sent a reply to the w3m thread on ubuntu-devel to announce the position of the ubuntu-server team. The thread is still active on the ubuntu-server mailing list but has drifted to the topic of server flavors.

Relocation of web pages from /var/www/ to /srv/www/

lukehasnoname put out a proposal for moving the default location of the RootDir in apache2 from /var/www to /srv/www. The main reason is to be compliant with the FHS. Objections were raised: increased delta with Debian, common knowledge about the location of the root directory, FHS doesn't require any structure in /srv. It was suggested to take this proposal to the Debian maintainers and discuss it with them.

Intrepid Spec status

ScottK asked about the state of approval for specs. dendrobates answered it was working on them now and the list should be ready in the next few days. ScottK also ask for feedback on his server flavor spec. i

Iso testing for 8.04.1

mathiaz reminded that the iso testing tracker has 8.04.1 candidates for the ubuntu-server isos. Testing is more than welcome in that area. He also added that isos for 8.10-alpha1 are being worked on, but nothing is ready for testing yet. Those interested in helping in iso-testing should register with the iso testing tracker.

MIR for Recommends

ScottK asked whether MIR for Recommends should be created. Starting from Intrepid, apt installs Recommends by default. However if packages are not available it won't fail. The question is whether packages in main can recommend packages in universe. mathiaz suggested to send an email to ubuntu-devel to raise this issue as there wasn't a clear answer given in the discussion.

Agree on next meeting date and time

Next meeting will be on Tuesday, July 1st at 15:00 UTC in #ubuntu-meeting.

Log

[16:03] <lukehasnoname> heh
[16:03] <mathiaz> meh - no Mootbot
[16:03] <ivoks> damn robots
[16:03] <kraut> lukehasnoname: thanks
[16:03] <mathiaz> so today's agenda: https://wiki.ubuntu.com/ServerTeam/Meeting
[16:04] <mathiaz> last week meeting minutes: https://wiki.ubuntu.com/MeetingLogs/Server/20080617
[16:04] <mathiaz> [TOPIC] Ebox and augeas
[16:05] <mathiaz> so we had some discussion about augeas and ebox
[16:05] <mathiaz> I sent an email to the ebox developers about this issue
[16:05] <mathiaz> https://lists.warp.es/pipermail/ebox-devel/2008-June/000376.html
[16:06] <mathiaz> there has been some discussion about this and the ebox developers seem open to the idea
[16:06] <mathiaz> they've started to look into using augeas and experiment with it
[16:06] <mathiaz> on a related note, nxvl has packaged augeas for ubuntu
[16:07]  * nealmcb cheers
[16:07] <mathiaz> http://nvalcarcel.aureal.com.pe/?p=196
[16:07] <mathiaz> he uploaded it to REVU and is waiting for feedback - http://revu.ubuntuwire.com/details.py?package=augeas
[16:08] <mathiaz> so I think that all I have to report from the ebox front
[16:09]  * lukehasnoname claps
[16:10] <mathiaz> [TOPIC] text browser on the server cd
[16:10] <mathiaz> so I sent a reply to the thread with the official decision from the server meeting
[16:10] <mathiaz> and the discussion is still going on - also it turned into a broader subject
[16:11] <mathiaz> but I don't think there anything more to say in this meeting about it.
[16:11] <lukehasnoname> well
[16:11] <mathiaz> anyone wants to add something wrt to last week meeting ?
[16:11] <nealmcb> well it seemed to me to relate well to scottk's blueprint
[16:12] <lukehasnoname> nealmcb: pretty close, ya
[16:12] <lukehasnoname> One idea is making custom ISOs from a javascript web-app
[16:12] <mathiaz> nealmcb: right - so I'd suggest to update the blueprint with ideas that emerged during the thread
[16:12]  * nijaba not sure FAI is the right approach though, if we are talking about the same blueprint
[16:13] <mathiaz> and continue the discussion on the blueprint
[16:13] <mathiaz> nijaba: we're talking about this blueprint
[16:13] <nealmcb> do we have consensus that it should be easy for folks to either install some sort of bare-bones server, or a comfortable more friendly server
[16:13] <lukehasnoname> https://lists.ubuntu.com/archives/ubuntu-server/2008-June/001812.html
[16:13] <lukehasnoname> nealmcb: Looks like it from the ML
[16:13] <mathiaz> nealmcb: from what I read from the thread, it seems so
[16:14] <ivoks> bare-bone server is waporware...
[16:14] <mathiaz> nealmcb: that may be included in the blueprint rationale
[16:14] <mathiaz> nealmcb: to explain the reason for doing so
[16:14] <nealmcb> ivoks: ?
[16:14] <kirkland> mathiaz: i think it's essential that Ubuntu provide a mechanism for a bare minimal CLI-only system
[16:15] <kirkland> mathiaz: I say Ubuntu in that it may or may not be the Server Team, but I think it would suit well for our team to provide that sort of install
[16:15] <ivoks> nealmcb: there will always be people considering it not so bare as it should be; just ignore me
[16:15] <mathiaz> kirkland: agreed - I like to the approach of installing a bare-system and then having a way to turn it into a specific type of machine
[16:16] <nealmcb> I think the question is how bare bare is...  there are lots of utilities in unix that are helpful but not essential
[16:16] <kirkland> mathiaz: right...  and if that's JEOS->real hardware, or Alternate CD, or mini.iso+CLI ...  whatever
[16:16] <ivoks> there can't be bare bone server; there can be bare bone linux system
[16:16] <nealmcb> just give 'em busybox :)
[16:16] <mathiaz> kirkland: right - but at least we start to have some general idea about where we'd like to head for
[16:16] <ivoks> so, as i agree with kirkland; this should be ubuntu project, not ubuntu server project
[16:16] <kirkland> nealmcb: if it can run apt-get and connect to the network, it's minimal enough for me
[16:17] <mathiaz> ok - let's move on
[16:17] <nealmcb> ivoks: in the sense that you can give them a platform for putting their own services on it, with a good kernel etc, you can give them a jeos with hardware support and let them build from there
[16:17] <soren> kirkland: apt-get? Sheesh! What's wrong with telnet+ar+tar+gzip?
[16:17] <lukehasnoname> soren: ha
[16:17] <mathiaz> [TOPIC] Relocation of web pages from /var/www/ to /srv/www/ in future releases
[16:17] <lukehasnoname> Ah, yes
[16:17] <mathiaz> lukehasnoname: why ?
[16:18] <soren> *blink*
[16:18] <kirkland> ???
[16:18] <soren> Why, oh why?
[16:18] <lukehasnoname> >_>
[16:18] <lukehasnoname> ok
[16:18] <zul> mathiaz: no no no no no
[16:18] <ivoks> let's all agree - no
[16:18] <kraut> may i say something to that?
[16:18] <nealmcb> soren: :)
[16:18] <sommer> seems like that would cause issues
[16:18] <lukehasnoname> Simply consider it, though I know chances are slim. But it would conform much better to logic and the FHS if we did
[16:18]  * kraut would prefer that idea and lean the directory hierachie completly to FHS
[16:18] <kraut> (FHS: http://www.pathname.com/fhs/pub/fhs-2.3.pdf)
[16:18] <lukehasnoname> same with mysql and ftp
[16:19] <lukehasnoname> I've spoken with some people who agree on principle
[16:19] <kraut> AOL
[16:19] <lukehasnoname> it's up to you guys to put it into practice... let me very quickly grab a quote
[16:19]  * nealmcb always puts server stuff in /srv - what would the impacts be?
[16:19] <kraut> many people would cry after the change, but after everybody knows the "issue", it would be much better
[16:19] <lukehasnoname>  /srv contains site-specific data which is served by this system.
[16:19] <kraut> and FHS is just the best standard i think
[16:19] <lukehasnoname>  /var contains variable data files. This includes spool directories and files, administrative and logging data, and transient and temporary files.
[16:20] <lukehasnoname> those are from the FHS
[16:20] <ivoks> ahm...
[16:20] <nealmcb> and what do other distros do?
[16:20] <lukehasnoname> If you read the page kraut linked to, just the two paragraphs on those two dirs, you'll see the logic I'm looking at.
[16:20] <sommer> nealmcb: /var/www/*
[16:20] <kraut> nealmcb: redhat and suse does the same
[16:20] <nijaba> Suse - /srv
[16:20] <ivoks> 'Therefore, no program should rely on a specific subdirectory structure of /srv existing or data necessarily being stored in /srv.'
[16:20] <kirkland> nealmcb: RH is /var/www/html
[16:21] <zul> gentoo is /srv if I remember correctly but they dont count ;)
[16:21] <kraut> ah, sorry, redhat uses /var/www/html
[16:21] <kraut> blame on me, i've just watched it out :)
[16:21] <lukehasnoname> ivoks: That is talking cross -platform, but from an Ubuntu perspective, it would be much more logical to have static web files in its own directory, in a different area from logs
[16:21] <lukehasnoname> zul:  aw
[16:21] <kraut> i'm a friend of standards and best practices and FHS is at the moment the best choice.
[16:21]  * soren points at ivoks
[16:22] <kraut> i'm out. need to work. just highlight me.
[16:22] <lukehasnoname> kraut:  thanks.
[16:22] <soren> kraut: Certainly. And the FHS itself says why it's a bad idea, as ivoks pointed out.
[16:22] <emgent> hello.
[16:22] <emgent> @schedule rome
[16:22] <ubottu> emgent: Schedule for Europe/Rome: Current meeting: Server Team | 24 Jun 20:00: LoCo Council | 25 Jun 19:00: QA Team | 26 Jun 00:00: Platform Team | 26 Jun 12:00: MOTU School Session - Apport retraces
[16:23] <lukehasnoname> brb
[16:23] <lukehasnoname> think about it
[16:23] <zul> I dont agree with this at all besides this should really be done in debian
[16:24]  * sommer agrees with zul
[16:24] <ivoks> there are some benefits of change, but are they so big to really do it?
[16:24] <lukehasnoname> zul: I can see your debian argument
[16:24]  * ivoks agrees with zul and soren 
[16:24]  * sommer also agrees with ivoks 
[16:24] <nealmcb> has debian discussed it?
[16:24] <kraut> FHS: http://www.pathname.com/fhs/pub/fhs-2.3.pdf - Point: 3.16.1. Purpose
[16:24] <mathiaz>  /srv seems to be targeted at site customization, ie the local sysadmin is reponsible for figuring out the layout - as the package maintainers, we can default to /var/ww
[16:24] <soren> Even if this was a good idea (which I don't think it is), it would have to be an *extremely* good idea to justify the horror of maintaining a delta for each and every debian package that ships stuff in /var/Www
[16:25] <ivoks> i mean... there's a lot of packages, we would have huge delta just cause of location change
[16:25] <mathiaz> there are a lot of packages that relay on /var/www and also common knowledge
[16:25] <kraut> soren: so pointing to that, it would be a good idea to use /srv/<protocol>, for example /srv/www
[16:25] <lukehasnoname> So the consensus is: If it were ever to happen, run it through deb?
[16:25] <soren> kraut: "that"?
[16:25] <mathiaz> it's not fundamitally wrong to use /var/www
[16:25] <kraut> soren: why to use /srv/www instead of /var/www
[16:26] <lukehasnoname> I'm also the guy who thinks that /usr is redundant
[16:26] <zul> its just a nightmare to maintain from a developers point of view
[16:26] <ivoks> lukehasnoname: if debian changes it, i would accept it; i'm not strongly oposed, but i am oposed to big delta for no gain
[16:26] <kraut> mathiaz: of course not, but /srv/www is more a "best practice"
[16:26] <soren> kraut: "unspecified and there is currently no consensus on how this should be done"
[16:26] <lukehasnoname> however, I will run it through Debian if I write up a formal proposal that I think has strong case points.
[16:26] <kraut> soren: that's business as usual
[16:26] <soren> And if we start shipping stuff in there, we might *very* well overwrite user's stuff that they already put in there.
[16:27] <ivoks> kraut: there a tons of tutorials with 'place the website into /var/www'
[16:27] <kraut> soren: it's more a guide for "best practisces", and /srv/<protocol> sounds reasonable
[16:27] <mathiaz> lukehasnoname: that seems like the best course of action
[16:27] <kraut> ivoks: only while years ago somebody choiced that, we need to choice it also in our future, also if there is a better soloution?
[16:27] <mathiaz> lukehasnoname: we won't change the default location if debian is against it because of the huge delta it would introduce
[16:27] <lukehasnoname> Alright... I know it's a change, and change is harsh, but bad habits are formed out of archaic standards that never get moved around.
[16:27] <Brazen> There can also be a transition period with /var/www symlinked to /srv/www (sorry I'm late, don't know if this has been mentioned)
[16:27] <lukehasnoname> mathiaz: That is the only point I completely agree with here,.
[16:28] <lukehasnoname> Debian has to be aboard
[16:28] <soren> kraut: I don't think so, actually. I order stuff in /srv by domain.
[16:28] <ivoks> kraut: as you can see, we aren't very sure it's better; FHS is very... hm... blur about /srv
[16:28] <ScottK-palm> Sorry for being late. Did we talk about specs already.
[16:28] <zul> ScottK-palm: no
[16:28] <lukehasnoname> no ScottK-palm
[16:28] <mathiaz> we have to balance these issues (delta, history and common knowledge) with the best practices (which aren't so clear according to the FHS)
[16:28] <ivoks> i use /srv for smb shares :D
=== ArneGoet1e is now known as ArneGoetje
[16:28] <ScottK-palm> Thanks.
[16:28] <kraut> ivoks: that's correct, but if i understood correctly, /srv/<protocol> or /srv/<department>/</protocol> would be the best choice
[16:29] <zul> ivoks: you are weird ;)
[16:29] <kraut> ivoks: i use /home/kraut/mobiledisk for samba, so should that be a worldwide standard? ;)
[16:29] <lukehasnoname> Hell with you guys, I'm going to gobolinux. :p That's that then. I'll discuss the possibility with Debian if I feel I have a case, otherwise I'll just manually do it for me.
[16:29] <kraut> (just kidding)
[16:29] <nealmcb> kraut: common, yes.  "best"?  ymmv...
[16:29] <ivoks> kraut: 'such as' - FHS doesn't have a strong opinion on /srv
[16:29] <ivoks> zul: we all are :D
[16:29] <ScottK-palm> On /var/www we should definitely follow Debianm
[16:29] <kirkland> i definitely do not see how this change would improve the Ubuntu experience for anyone--sysadmins or users
[16:30] <mathiaz> allright - let's move on
[16:30] <lukehasnoname> yes
[16:30] <Brazen> definately follow Debian, but strongly encourage Debian to move to /srv
[16:30] <ivoks> leave it to debian
[16:30] <nealmcb> ScottK-palm: or perhaps lead them, but be in sync....
[16:30] <mathiaz> I think we've got some opinions on why not to do it
[16:30] <kraut> Brazen: *5*
[16:30] <mathiaz> and why to do it
[16:30] <mathiaz> next topic
[16:30] <lukehasnoname> Welcome Brazen. I got slaughtered, heh.
[16:31] <mathiaz> [TOPIC] Intrepid spec status
[16:31] <Brazen> sorry lhnn, I meant to be here
[16:31] <ScottK-palm> I'm not saying don't try to convince them, but don't diverge over it.
[16:31] <mathiaz> dendrobates is the last steps of getting things approved
[16:31] <mathiaz> ScottK-palm: which specs are looking for ?
[16:32]  * ScottK-palm got one comment from dendrobates.  thanks.  how does it look on getting approved?
[16:32] <mathiaz> ScottK-palm: I think that the two specs about email are quit straightforward and could be implemented
[16:32] <ScottK-palm> It's the flavors spec I'm most interested in.
[16:33] <mathiaz> ScottK-palm: right - there has been some discussion about it
[16:33] <mathiaz> ScottK-palm: during this meeting and on the w3m thread
[16:33] <dendrobates> ScottK-palm: I have been delayed while canoncial worked some things out.  I will be starting the approval process this afternoon.
[16:33] <mathiaz> ScottK-palm: so it seems that there is more work to be done on the spec
[16:33] <ScottK-palm> Great.
[16:33] <nealmcb> ScottK-palm: fyi, (09:06:48 AM) mathiaz: on a related note, nxvl has packaged augeas for ubuntu
[16:34] <ScottK-palm> I saw.
[16:34] <nijaba> ScottK-palm: how would you consider replacing FAI by augeas in your spec (or supplementing it?)
[16:34] <ScottK-palm> Comments on the flavors spec please. I'll fix it up.
[16:35] <ScottK-palm> nijaba: one step at a time.
[16:35] <ScottK-palm> I think with FAI will be hard enough.
[16:36] <mathiaz> I haven't seen any new specs subscribed
[16:36] <ivoks> ssl thing?
[16:36] <mathiaz> so make sure that the specs are in a pending aproval state
[16:36] <ivoks> maybe i didn't do subscribing right :/
[16:36] <mathiaz> so that dendrobates can go through the list
[16:36] <ScottK-palm> Plus one on thay one.
[16:36] <mathiaz> ivoks: I saw it
[16:36] <ivoks> ok, thanks,
[16:37] <ScottK-palm> SSL v2 needs to die.
[16:37] <mathiaz> anything else on the spec front ?
[16:37] <ivoks> ScottK-palm: we'll kill it
[16:37] <ScottK-palm> Not from me.
[16:38] <ScottK-palm> Anyone?
[16:38] <mathiaz> let's move on then
[16:38] <mathiaz> [TOPIC] Iso testing - 8.04.1 and 8.10-alpha1
[16:38] <mathiaz> we're gearing for 8.04.1 and 8.10-alpha1
[16:39] <mathiaz> new isos have been created and testing the ubuntu-server isos is welcomed
[16:39] <mathiaz> slangasek: are there 8.10-alpha1 ubuntu-server isos ready for testing ?
[16:40] <mathiaz> the iso testing tracker (http://iso.qa.ubuntu.com/) has 8.04.1 isos
[16:40] <mathiaz> slangasek: but I don't see any -server isos for intrepid alpha1
[16:41] <ScottK-palm> Wasn't that delayed?
[16:41] <mathiaz> ScottK-palm: we've started the alpha1 isos now
[16:41] <mathiaz> ScottK-palm: the new kernel has been uploaded
[16:41] <ScottK-palm> OK.
[16:42] <mathiaz> ScottK-palm: so alpha1 isos are being prepared
[16:42] <mathiaz> anyway - -server isos for 8.04.1 are already ready
[16:42] <dendrobates> it will be interesting to see if installing Recommends by default causes us a problem.
[16:42] <dendrobates> It has already been problematic on my dev vm's
[16:42] <mathiaz> so iso testing is much appreciated
[16:43] <nijaba> dendrobates: that's not in 8.04.1 though...
[16:43] <ScottK-palm> We need some clear policy if Universe recommends need MIR.
[16:43] <lukehasnoname> dendrobates: I assume packages have been getting worked on to have reasonable recommends?
[16:43] <zul> nijaba: no it isnt
[16:43] <mathiaz> we're targeting a release of 8.04.1 next week
[16:43] <dendrobates> nijaba: the topic includes 8.10 alpha.
[16:44] <nijaba> dendrobates: my mistake, did not notice
[16:44] <ScottK-palm> Fortunately Debian has had reccomends by default for some time so they've done a lot of work.
[16:44] <mathiaz> ScottK-palm: right - well it depends on the package
[16:44] <dendrobates> bzr will install X  :)
[16:45] <zul> dendrobates: ergh
[16:45] <mathiaz> ScottK-palm: I guess dropping to suggest if needed
[16:45] <ivoks> devscripts installs... well... a lot :D
[16:45] <lukehasnoname> I had trouble with this recently, since the C# compiler is "recommends" for mono-develop and it didn't get installed
[16:45] <mathiaz> ScottK-palm: other wise, MIR will have to be filled for each recommends
[16:45] <soren> dendrobates: What?
[16:45] <mathiaz> ScottK-palm: since main has to be installable on its own
[16:46] <mathiaz> ScottK-palm: why would you think MIR should be avoided for Recommends ?
[16:46] <ScottK-palm> Yes, but install won't fail on missing recommends.
[16:46] <dendrobates> if it is too large of a problem, we can push the change back in server to intrepid+1, but only if we have too many problems to handle.
[16:47] <ScottK-palm> Not saying either way.  It's a lot of MIR though.
[16:47] <mathiaz> ScottK-palm: hm - good question then. You may wanna ask about it on ubuntu-devel
[16:48] <lukehasnoname> MIR?
[16:48] <lukehasnoname> sudo acronym-define MIR
[16:48] <ScottK-palm> Maybe someone could make a list for the server packages.
[16:48] <ScottK-palm> Main Inclusion Report
[16:49] <kirkland> lukehasnoname: https://wiki.ubuntu.com/MainInclusionProcess
[16:49] <lukehasnoname> thanks ScottK-palm kirkland
[16:49] <mathiaz> ScottK-palm: good point. that would give us some real data on the amount of work needed
[16:50] <ScottK-palm> mathiaz: MIR is more of a Canonical question than Ubuntu.
[16:50] <ivoks> ScottK-palm: how come?
[16:50] <ScottK-palm> Because it affects their support contracts.
[16:51] <nealmcb> well, it affects the support contracts of anyone who supports main
[16:51] <ivoks> right
[16:52] <nijaba> ScottK-palm: it affects mainly Ubuntu maintenance commitment
[16:52] <nijaba> the support contract is a separate thing
[16:52] <ivoks> the moment we don't include something in main cause it affects somebody's contract, i'll leave :)
[16:52] <ScottK-palm> :)
[16:52] <nijaba> if canonical decides to support a subset of main, nothing forbids us to do so
[16:53] <ScottK-palm> True.
[16:53] <nijaba> and that is not really something that we should consider here
[16:53] <nealmcb> nijaba: right
[16:53] <ivoks> let's eliminate fear of canonical and move on :)
[16:53] <nijaba> however, maintenance of packages has some impact
[16:54] <mathiaz> maintenance has an impact on the Ubuntu Security team
[16:54] <nijaba> and the security team is not the last one impacted
[16:54] <ScottK-palm> Someone please get a policy and let us peons know.
[16:55] <nealmcb> moving right along....
[16:55] <mathiaz> let's move on
[16:55] <mathiaz> [TOPIC] Open discussion
[16:55] <ScottK-palm> Yep....
[16:55] <mathiaz> anyone wants to add something ?
[16:56] <ivoks> postfix-dovecot is on it's way
[16:56] <zul> mathiaz: thanks for getting openldap uploaded btw
[16:56] <soren> ivoks: Cool.
[16:56] <mathiaz> ivoks: \o/
[16:56] <mathiaz> zul: np
[16:56] <soren> ivoks: I'm looking forward to seeing the implementation.
[16:56] <sommer> yay for openldap :)
[16:57] <ivoks> soren: https://bugs.launchpad.net/ubuntu/+source/dovecot/+bug/164837
[16:57] <ubottu> Launchpad bug 164837 in tasksel "Dovecot SASL for postfix" [Undecided,Invalid]
[16:57] <ScottK-palm> Do we have people who watch upstream support forums for dovecot,openldap, etc.?
[16:57] <ivoks> funny... invalid means something completly different in croatian :D
[16:58] <mathiaz> [TOPIC] Agree on next meeting date and time
[16:58] <kirkland> ivoks: there's an alternate meaning in English too
[16:58] <mathiaz> same time, same place, next week ?
[16:58] <sommer> mathiaz: sure
[16:58] <ivoks> +
[16:58] <ScottK-palm> Ubuntu specific stuff comes up on postfix-users mail list frequently.
[16:58]  * kirkland is on vacation next week ;-)
[16:59] <nijaba> +1
[16:59] <ivoks> ScottK-palm: like?
[16:59] <ScottK-palm> I think someone should be watching other upstream venus too.
[16:59] <zul> kirkland: nyeah nyeah nyah :P
[16:59] <mathiaz> allright - so same place, same time
[16:59] <ivoks> ScottK-palm: i'm on bacula's dev list
[16:59] <mathiaz> see ya next week and happy iso-testing

MeetingLogs/Server/20080624 (last edited 2008-08-06 16:59:37 by localhost)