20090310

Agenda

Items we will be discussing:

  • Review ACTION points from previous meeting.
  • Review progress made on the specification listed on the Roadmap.

  • Support for likewise-open krb5 patch.
  • Launch screen by default in ec2? -- DustinKirkland

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

Minutes

Samba Bug Day

mathiaz reminded that next Thursday is dedicated to triaging samba bugs. Ubuntu Server members are welcomed to join the effort and invited to join #ubuntu-bugs to give a hand to bug triagers.

Postfix and Dovecot integration

ivoks wrote a blog post about the dovecot/postfix integration. mathiaz mentioned bug 339966 reported by a user that had an existing working dovecot server and installed the dovecot-postfix package. That resulted in a broken system since his customized dovecot configuration had been preempted by the default dovecot-postfix.conf from the dovecot-postfix package. There was some discussion about how to prevent that but no final solution has been found.

Exchange support for Evolution

ivoks made some tests with the new evolution-mapi plugin. Although the openchange mapi clients are working evolution is unable to connect to the exchange server. ivoks is not sure what the problem is - further debugging is required.

Support for likewise-open krb5 patch

mathiaz gave an update on the likewise-open patch he had applied to the krb5 package in Jaunty. One of the Debian maintainer contacted him to inquire about the future of this patch. The upcoming version of MIT kerberos, 1.7, includes a different implementation of the GSS service than the one provided by the likewise-open patch. Implementations are API incompatible. The Debian maintainer suggested to rework the patch to not change the public header and include the types and prototypes in the likewise package instead. That would give a better chance of avoiding supporting the additional public functions.

There was some discussion about whether likewise-open 5 would be included in Jaunty (which is the main reason for including the krb5 patch in Jaunty). Packages are almost ready but it's rather late in the release cycle to include them in Jaunty.

Planet ubuntuserver

dantalizing mentioned he started to aggregate Ubuntu Server related blogs on The Completely Unofficial Ubuntu Server Planet.

KVM backport in hardy

kirkland announced he had backported the userspace part of KVM 84 to hardy. Packages are available in the ubuntu-virt team PPA. Anyone having long-standing issues with kvm-62 on hardy should try that package and let the Ubuntu Server team know how it works in #ubuntu-virt.

QEMU 0.10.0 in Jaunty

kirkland reported that a new version of qemu (0.10.0) had been released upstream. He plans to push it to jaunty if the Feature Freeze Exception is accepted by the MOTU release team.

Agree on next meeting date and time

There was some discussion about the time of the meeting since it clashes with the TB meeting and/or the Kernel Team meeting. It was decided to stick to 15:00 UTC.

Next meeting will be on Tuesday, March 17th at 15:00 UTC in #ubuntu-meeting.

Log

[15:03] <mathiaz> #startmeeting
[15:03] <kirkland> mathiaz: good luck with that
[15:03] <Brazen> o/
[15:03] <mathiaz> kirkland: I know
[15:03] <kirkland> :-)
[15:03]  * ball sits down and shuts up
[15:03] <mathiaz> kirkland: but it also help for writting the minutes later :)
[15:04] <mathiaz> so today's agenda: https://wiki.ubuntu.com/ServerTeam/Meeting
[15:04] <ivoks> ok
[15:04] <mathiaz> Previous meeting minutes: https://wiki.ubuntu.com/MeetingLogs/Server/20090303
[15:04] <mathiaz> [TOPIC] SRU bug tracking
[15:04] <mathiaz> ivoks: ^^
[15:04] <mathiaz> ivoks: ready to discuss that?
[15:04] <ivoks> nope
[15:04] <ivoks> :)
[15:04] <mathiaz> ivoks: ok - let's move on then
[15:04] <ivoks> i'll add it to agenda when redy
[15:05] <ivoks> ready
[15:05] <mathiaz> [ACTION] ACTION: ivoks to add to the server team agenda an item about better SRU management.
[15:05] <mathiaz> [TOPIC] Postfix and Dovecot integration
[15:05] <mathiaz> ivoks: thanks for blogging about it
[15:05] <ivoks> right, we got nice feedback
[15:05] <ivoks> people like it so far...
[15:05] <mathiaz> ivoks: there was a bug related to the package
[15:05] <ivoks> was it?
[15:06] <mathiaz> ivoks: bug 339966
[15:06] <uvirtbot> Launchpad bug 339966 in dovecot "dangerous action: dovecot-postfix force-installs new conf file" [Low,Confirmed] https://launchpad.net/bugs/339966
[15:06] <mathiaz> I'm not sure how we should handle that.
[15:06] <ScottK> If it's accurate, that's not Low, IMO.
[15:06] <mathiaz> I've been discussing wit the reporter
[15:07] <ivoks> i see...
[15:07] <mathiaz> ScottK: I don't think dangerous is the correct word
[15:07] <ivoks> i'll work on a fix for this
[15:07] <mathiaz> ivoks: what would be the plan?
[15:08] <ivoks> i have to think about it
[15:08] <ivoks> check if dovecot.conf is changed
[15:08] <mathiaz> I'm not sure we should/could support upgrading an existing dovecot system to integrate it with postfix
[15:08] <ScottK> It can be known if the existing config is modified or not, so I think this is solvable.
[15:08] <mathiaz> not really - dovecot.conf is modified by -pop and -imap when they get installed
[15:09]  * nijaba realizes the meeting is here... o/
[15:09] <ScottK> Hmmm
[15:09] <ivoks> mathiaz: well, we can workaround it
[15:09] <ScottK> OK.  Then I'm glad ivoks is going to solve it.
[15:09] <ivoks> mathiaz: ignore ^protocols and then check
[15:09] <cemc> modified but not ignored completely
[15:09] <ivoks> if user changed protocols, he still has not working dovecot
[15:09] <ivoks> or... hm...
[15:09] <mathiaz> ivoks: OTOH I'm not sure if we should support this in the dovecot-postfix package
[15:10] <cemc> if used had imap support, and he decides to install pop3, will that break imap ?
[15:10] <ivoks> mathiaz: i understand your point of view and i agree
[15:10] <mathiaz> the dovecot-postfix package is a different kind of package since it encapsulate a script
[15:10] <mathiaz> It doesn't really ship new files - it just modifies existing configuration
[15:10] <ivoks> mathiaz: maybe we should warn user during preinst
[15:11] <ivoks> mathiaz: adding 'if you have already working dovecot, purge this package'
[15:11] <mathiaz> ivoks: I thought about that - but you cannot detect if you're installing a brand new system or if the system is an pseudo-upgrade of a running dovecot system
[15:11] <ivoks> mathiaz: we just let everybody know that?
[15:11] <ivoks> mathiaz: it's a good way to advertise all features that comes with it :)
[15:12] <mathiaz> ivoks: right - so I think we should update the description of the package
[15:12] <ivoks> or that...
[15:12] <mathiaz> ivoks: also - I think that the postinst modifies the dovecot.conf file
[15:12] <mathiaz> ivoks: to stick a comment in it
[15:12] <ivoks> nope
[15:12] <ivoks> that's in source
[15:12] <ivoks> dovecot.conf is modified during build
[15:13] <mathiaz> ivoks: hm - right.
[15:13] <mathiaz> ivoks: so may be doing this in the postinst will force a ucf merge of dovecot.conf
[15:13] <ivoks> urgh...
[15:13] <mathiaz> ivoks: that way the admin will have to see the new comment added in dovecot.conf during postinst which would have helped in the situation
[15:14] <mathiaz> ivoks: I don't know if that's the correct way to do it though
[15:14] <ivoks> give me couple of days to thing about this
[15:14] <ivoks> think
[15:14] <mathiaz> ivoks: right - could you post your solution to the bug?
[15:14] <ivoks> sure
[15:15] <mathiaz> I'm not sure about the right way to tackle this issue
[15:15] <mathiaz> let's move on
[15:15] <mathiaz> [TOPIC] Samba bug day
[15:15] <ivoks> me neither, so i have to think about it for a while :)
[15:16] <mathiaz> as you may have noticed if you read the planet next Thursday will be dedicated to triagging samba bugs
[15:16] <mathiaz> ttx: thanks for blogging about it
[15:16] <methods> how do i install an older version of package ?
[15:16] <mathiaz> nijaba: still on track to blog about it tomorrow?
[15:16] <nijaba> mathiaz: sure
[15:16] <nijaba> mathiaz: blog is ready
[15:17] <mathiaz> I'll prepare also a post to be published on Thursday
[15:17] <nijaba> waiting ti tomorrow 9am for launch
[15:17] <mathiaz> nijaba: awesome - thanks :)
[15:17] <ttx> mathiaz: who will be running the show from QA ?
[15:17] <genii> methods: Specify version on cli apt-get. eg:   sudo apt-get install something=specific-version              "specific-version"  is one which can be reported by apt-cache policy <packagename>
[15:18] <mathiaz> ttx: you
[15:18] <mathiaz> ttx: and me :)
[15:18] <incorrect> when logging in over ssh i have a log delay before the bash prompt appears
[15:18] <mathiaz> ttx: oh - QA - noone special
[15:18] <incorrect> I am not sure why its on some systems but not others
[15:18] <ttx> mathiaz: ok
[15:18] <mathiaz> ttx: the whole triagger community is there
[15:19] <mathiaz> ttx: we should be in #ubuntu-bugs to give a hand to the triagger though
[15:19] <ttx> I'll be there.
[15:19] <nijaba> incorrect: failure to do a reverse dns check is generally the issue
[15:19] <mathiaz> any developer is welcome in #ubuntu-bugs too
[15:20] <mathiaz> to help out with samba bugs
[15:20] <mathiaz> let's move on
[15:20] <mathiaz> [TOPIC] Exchange support for Evolution
[15:20] <incorrect> nicetry, I've set DNS to no, it look to be that some nodes can't talk to the ldap server for some weird reason
[15:20] <kinnaz> https://bugs.launchpad.net/ubuntu/+source/lustre/+bug/229821 <--- anyone has figured fix for what ?
[15:20] <uvirtbot> Launchpad bug 229821 in lustre "lustre-source build fails in hardy" [Undecided,New]
[15:20] <mathiaz> ivoks: ^^ how did this worked out?
[15:20] <ivoks> still nothing
[15:20] <mathiaz> ivoks: no time or not working?
[15:20] <ivoks> evolution crashes
[15:21] <mathiaz> ivoks: did you report a bug?
[15:21] <ivoks> i tested mapiprofile app today and it worked (i guess)
[15:21] <ivoks> mathiaz: not yet, i see there's new evolution-mapi in archive
[15:21] <ivoks> so i'll test with it and then report
[15:21] <ivoks> mathiaz: i guess it's cause of this particular exchange setup
[15:21] <mathiaz> ivoks: great. Keep seb128 in the loop if you find some bugs
[15:21] <mathiaz> ivoks: hm ok.
[15:22] <mathiaz> ivoks: what is mapiprofile?
[15:22] <ivoks> mathiaz: eovlution-mapi uses libmapi library
[15:23] <ivoks> mathiaz: there are also cli tools for that 'openchangeclient'
[15:23] <ivoks> i'm not sure evolution calls openchangeclient directly or trough library
[15:23] <mathiaz> ivoks: ok - so the openchange client tools are working correclty
[15:23] <mathiaz> ivoks: which means that the issue is in the evolution-mapi plugin
[15:23] <ivoks> ................i'm not sure.... yet
[15:23] <mathiaz> ivoks: ok
[15:23] <ivoks> mathiaz: output is strange
[15:24] <ivoks> lots of OK OK OK, and then error connecting
[15:24] <mathiaz> ivoks: seems that it needs more investigation
[15:24] <ivoks> i'll have to dig more into it
[15:24] <mathiaz> ivoks: cool. thanks
[15:24] <mathiaz> That's all I had from last week minutes
[15:24] <mathiaz> is there anything else to add wrt to last week meeting?
[15:25] <mathiaz> nope - let's move on then
[15:25] <mathiaz> [TOPIC] Support for likewise-open krb5 patch
[15:26] <mathiaz> so I've uploaded likewise-open patch to krb5 in jaunty
[15:26] <mathiaz> and I've received an email from the debian maintainer
[15:27] <mathiaz> the situation is that the next version of mit krb5 (1.7) will provide the same functionality (GSSAPI) but with a different implementation than the one done by likewise
[15:27] <mathiaz> however for jaunty we will ship krb 16 with the GSSAPI support from likewise-open
[15:28] <ttx> when is 1.7 scheduled ?
[15:28] <mathiaz> ttx: end of april
[15:28] <mathiaz> the likewise-open patch introduced symbols you've
[15:28] <mathiaz> exposed in your shared libraries and public headers that have diverged
[15:28] <mathiaz> from the krb5 upstream.
[15:29] <ttx> do we need the GSSAPI support for anything else than likewise-open 5 ?
[15:29] <mathiaz> that may mean we'd have to maintain API compatibility in coming releases
[15:29] <mathiaz> ttx: no - not in jaunty
[15:29] <ttx> hmm.
[15:30] <mathiaz> ttx: the reason I put the patch in jaunty is because of likewise-open
[15:30] <ttx> mathiaz: I'm not sure likewise-open 5 will make it to jaunty. If it doesn't, it would really make sense to back out that patch.
[15:31] <mathiaz> ttx: when would we know when it doesn't?
[15:32] <ttx> mathiaz: very soon. My packaging is almost ready, Ffe should follow
[15:32] <mathiaz> one proposal is to change the krb5 patch slightly
[15:33] <ttx> but it's a complete rearchitecture, not a small update. So we are quite late oin the cycle for it
[15:33] <mathiaz> to avoid supporting additional public functions
[15:33] <mathiaz> ttx: so my question is if we should look into changing the likewise-open krb5 patch to maintain API compatibility with upstream
[15:34] <mathiaz> ttx: which means modifying likewise-open
[15:34] <mathiaz> ttx: to support the modified krb5 patch
[15:36] <ttx> mathiaz: yes, probably.
[15:37] <mathiaz> ttx: ok
[15:37] <mathiaz> ttx: it seems we should discuss this a bit more with upstream
[15:38] <mathiaz> let's move on
[15:38] <ttx> mathiaz: definitely.
[15:39] <mathiaz> [TOPIC] Launch screen by default in ec2
[15:39] <mathiaz> kirkland: ^^
[15:39] <kirkland> mathiaz: zul has asked that i postpone this discussion until he's back from vacation
[15:39] <mathiaz> kirkland: ok.
[15:39] <kirkland> mathiaz: sorry
[15:39] <mathiaz> kirkland: np
[15:39] <mathiaz> [TOPIC] Open Discussion
[15:39] <mathiaz> anything else to add?
[15:40] <dantalizing> i've got something
[15:40] <ScottK> I discovered Debian Bug 518524 today
[15:40] <uvirtbot> Debian bug 518524 in amavisd-new "Fails to detect message with multiple virus payloads as infected" [Grave,Fixed] http://bugs.debian.org/518524
[15:40] <ivoks> oh, lol
[15:40] <ScottK> I've got a merge prepared and asked ubuntu release if I should upload it now or after Alpha 6.
[15:40]  * ScottK is waiting for an answer.
[15:40] <ScottK> It only affects Jaunty.
[15:40]  * ScottK is done.
[15:40] <ivoks> after alpha sounds ok, imho
[15:41] <ivoks> after all, it's a development release :)
[15:41] <ScottK> It's a quick build, so depending on where they are, I could see it either way.
[15:41] <mathiaz> ScottK: well - does it block the release of alpha6?
[15:42] <ScottK> mathiaz: No, so it isn't critical if it waits.
[15:42] <mathiaz> ScottK: I don't think so - so it can wait for after alpha6
[15:42] <mathiaz> dantalizing: yes
[15:43] <dantalizing> i just started aggregating rss on my own from multiple people, and it has evolved.
[15:43] <dantalizing> i picked up ubuntuserver.org and put a planet on it
[15:43] <dantalizing> http://planet.ubuntuserver.org/
[15:43] <dantalizing> i know there is an exisiting wp site
[15:43] <dantalizing> just wanted to throw it out to the server team
[15:43] <dantalizing> fyi
[15:44] <dantalizing> and if yall had some specific desire for it
[15:45] <jbernard> kirkland merged my changes to update-motd to add inotify support, fyi
[15:45] <mathiaz> dantalizing: great - thanks.
[15:45] <kirkland> jbernard: but i haven't uploaded it to jaunty yet :-)
[15:45] <jbernard> ill file an Ffe today
[15:45] <kirkland> but yes, jbernard did some great work to get update-motd to be able to run either in a cron-base, or an inotify based mode
[15:46] <kirkland> \o/
[15:46] <mathiaz> jbernard: awesome. Thanks for the good work!
[15:47] <mathiaz> anything else to add before we wrap up?
[15:47] <kirkland> mathiaz: 2 small things from me
[15:47] <kirkland> mathiaz: i've backported kvm-84 to build on hardy, it's available in the ~ubuntu-virt PPA
[15:48] <ivoks> er... including kernel part?
[15:48] <kirkland> mathiaz: anyone having long-standing issues with kvm-62 on hardy ... i suggest you try that package and let us know how it works for you in #ubuntu-virt
[15:48] <kirkland> ivoks: not yet, i'll work on that next
[15:48] <kirkland> ivoks: userspace only, thanks for the clarification
[15:48] <ivoks> ;)
[15:48] <kirkland> mathiaz: and second, qemu has finally release 0.10.0 (after nearly a year)
[15:49] <kirkland> mathiaz: i'm merging that now, would like to try to get that into universe for jaunty, will probably need an FFE
[15:49] <kirkland> mathiaz: but there are lots of bugs fixed
[15:50] <mathiaz> kirkland: you'd have to ask the motu-release team - ScottK would probably be able to help reviewing the FFe.
[15:50] <ScottK> For server issues I can decide.  File a FFe bug and I will review it.
[15:50] <kirkland> mathiaz: yep, 'tis why i'm mentioning it here, since it seemed that ScottK was around
[15:50] <kirkland> ScottK: thanks, will do
[15:51] <kirkland> mathiaz: all from me
[15:51] <mathiaz> great. Anything else to add?
[15:53] <mathiaz> [TOPIC] Agree on next meeting date and time
[15:53] <mathiaz> so as some of you have noticed we're currently clashing with the TB meeting
[15:53] <mathiaz> the kernel team meeting has also moved one hour forward
[15:54] <soren> Has Canada switched to DST as well?
[15:54] <mathiaz> soren: yes
[15:54] <soren> Or have we really moved it just for the benefit of the USAnians?
[15:54] <mathiaz> soren: canada and usa are in DST
[15:54] <genii> No, us canucks are forced to suffer also
[15:54] <mathiaz> OTOH the TB will also move back an hour in 4 weeks
[15:55] <mathiaz> and since TB meetings only happen every other week we would just conflict once more
[15:55] <mathiaz> so my proposal is to leave the server meeting at 15:00 UTC
[15:55] <soren> Tell me again why we moved the meeting?
[15:56] <ivoks> to make it more interesting :)
[15:56] <soren> ivoks: It's not working :)
[15:56] <mathiaz> soren: because it's the same time for northamerican
[15:56] <mathiaz> soren: and it was a better time for europeans
[15:56] <soren> mathiaz: Which makes it different for *everyone* else.
[15:56] <ivoks> northamericans, raise your hand; everybody else, raise both of them
[15:57] <sommer> o/
[15:57] <mathiaz> soren: yes - but everything will be back to the regular schedule in 4 weeks (for european)
[15:57] <Brazen> o/
[15:57] <jbernard> o/
[15:57] <soren> mathiaz: You're not making a very convincing argument :)
[15:58] <mathiaz> anyway - my point being that we'd have only one more conflict with the TB
[15:58] <mathiaz> and after that we'd be back to the regular schedule with TB (every other week), server, kernel teams meeting
[15:59] <ScottK> o/
[15:59] <soren> You could use the exact same arguments for keeping it at the same time relative to UTC.
[15:59] <soren> In three weeks it'd be normal for USAnians and Canadians.
[16:00] <mathiaz> soren: nope -because we'd have a weekly conflict with the kernel team meeting
[16:00] <soren> ...and all the while, it's been normal for Europeans.
[16:00] <soren> See, *that's* a (somewhat) useful argument.
[16:00] <mathiaz> 15:00 UTC -> 1 conflict with the TB meeting (in 2 weeks), 16:00 UTC -> 3 conflicts with the kernel team
[16:01] <soren> Alright.
[16:01] <mathiaz> so next week, in #ubuntu-meeting at 15:00 UTC?
[16:01] <ivoks> ok
[16:02]  * soren still grumbles that we have to bow to the kernel team's acceptance of American daylight savings time imperialism and not the other way around :)
[16:03]  * mathiaz points soren to #ubuntu-kernel
[16:03] <ivoks> let's kill DST
[16:03] <ivoks> it's usless anyway
[16:03] <ScottK> mathiaz: slanagasek told me to go ahead and upload, so it's done.
[16:03] <Brazen> I leave my lights on all day anyway
[16:03] <ball> I mostly run servers at UTC and let workstations calculate their offset based on that.
[16:04] <mathiaz> allright folks - thanks for attending.
[16:04] <mathiaz> happy alpha6 testing
[16:04] <mathiaz> and see you all next week, in #ubuntu-meeting at 15:00 UTC
[16:04] <ivoks> we need to have lights on car all the time :)
[16:04] <mathiaz> for the Ubuntu server team meeting
[16:04] <mathiaz> #endmeeting

MeetingLogs/Server/20090310 (last edited 2009-03-10 21:00:08 by mathiaz)