[00:00] <soren> The packages should figure it out somehow. I have an idea for this for intrepid.
[00:01] <_CitizenKane_> Hello everyone, I was wondering if there is a good tool to analyze things such as httpd access and error logs, php error log, etc. ?
[00:02] <_CitizenKane_> Ideally, it would also be web based
[00:20] <zul> ivoks: when you are done can you send me your debdiff for bacula
[01:40] <ivoks> zul: sure
[02:45] <kgoetz> right... sudo crashing is probably a bad thing
[02:46] <sommer> kgoetz: did you do a hardy apt-get upgrade?
[02:46] <kgoetz> sommer: yeah (i do expect it to be a known issue)
[02:47] <kgoetz> few other things didnt work in the dist-upgrade itself, but i expect thats unrelated
[02:47] <sommer> I totally just ran into the same issue... :-)
[02:48] <sommer> I thought I'd be clever and reboot, now the darn thing won't come back :\
[02:48] <kgoetz> :|
[02:48] <kgoetz> "interesting" *mentally notes he should be using apt-bugs*
[02:49] <kgoetz> i notice malloc's happening during the dist-upgrade
[02:49] <sommer> I didn't do a dist-upgrade... for that you should use do-release-upgrade
[02:50] <ivoks> sommer: remote server? :)
[02:50] <kgoetz> i dist-upgraded hardy -> hardy.
[02:51] <kgoetz> grr. do-release-upgrade, no man page
[02:51] <kgoetz> i cant open new terminals on the system in screen. interesting
[02:54] <kgoetz> !pastebin
[02:54] <ubotu> pastebin is a service to post multiple-lined texts so you don't flood the channel. The Ubuntu pastebin is at http://paste.ubuntu-nl.org (make sure you give us the URL for your paste - see also the channel topic)
[02:55] <kgoetz> FYI: the mallocs/failed update: http://paste.ubuntu-nl.org/59440/
[02:55] <kgoetz> and an example explosion: http://paste.ubuntu-nl.org/59441/
[02:56] <Jeeves_> root@unknown # zpool status  | grep ONLINE | grep -v raidz | grep -v state | grep -v pool | wc -l 46
[02:57] <sommer> kgoetz: I received the same sudo error
[02:58] <kgoetz> sommer: i ran sudo 3 tijmes, it errored slightly differently each time (probably just different memory addresses).
[02:59] <ivoks> wow, this looks bad
[03:00] <ivoks> don't reboot that machine
[03:00] <sommer> ivoks: you could argue that it's not good :-)
[03:01] <sommer> I'm getting an invalid uuid error... probably because the apt-get upgrade process didn't finish
[03:01] <ivoks> does 'update-grub' works?
[03:01]  * kgoetz doesnt have a root term open either :|
[03:01] <sommer> it's in a initramfs
[03:02] <sommer> I'll try recovery mode
[03:02] <ScottK2> I just had a build die on the buildd's on some kind of UUID error.  I wonder if it's related.
[03:05] <ivoks> ScottK2: dovecot.conf is a conffile?
[03:05] <ivoks> ScottK2: check out /var/lib/dpkg/info/dovecot-common.conffiles
[03:07] <kgoetz> sommer: has it come back yet?
[03:09] <ScottK2> ivoks: soren corrected me a bit on that.
[03:09] <ScottK2> ivoks: Since it's manufactured and not shipped, it's not.  I missed that subtlety.
[03:09] <sommer> kgoetz: nope... not finding partitions by uuid, even in recovery
[03:09] <ivoks> right
[03:09] <ScottK2> ivoks: slangasek still thought it was poor form if actually legal.
[03:10] <sommer> kgoetz: trying boot disk (sigh)
[03:10] <kgoetz> sommer: gl
[03:10] <ivoks> i agree it's poor, but it is legal
[03:10]  * kgoetz cant check the server, pschulz01 has unpluged it
[03:11] <ivoks> and it's the best way to do it; forcing custom changes on everybody sounds... well arogant
[03:11] <ivoks> doing it trough ubuntu's mail-server task sounds like a product
[03:11] <ScottK2> ivoks: Won't current users get the diff thrown at them.
[03:12] <ScottK2> ivoks: Did you read slangesek and mathiaz's conversation on #ubuntu-devel?
[03:12] <ivoks> ScottK2: no, when was it?
[03:12]  * ScottK2 looks
[03:13] <ScottK2> ivoks: It starts with me asking slangesek about it at 22:31 UTC on Wed.
[03:14] <ivoks> ok, i'll read it
[03:17] <kgoetz> sommer: pschulz01: says he hasnt unplugged it, so it must have disconnected for some other reason :| (i'm getting connection refused).
[03:18] <kgoetz> we're turning it off and waiting for news
[03:19] <ivoks> meh... metapackage was my first proposal :/
[03:19] <sommer> kgoetz: weird stuff...
[03:20] <sommer> I've mounted my root partition using a recovery cd, but I don't have a /sbin
[03:20] <pschulz01> sommer: This sound scary
[03:20] <sommer> doh... nm it's target/sbin
[03:20] <kgoetz> sommer: uh... say what? thats ... bizzare?
[03:20] <pschulz01> sounds
[03:22] <sommer> kgoetz: did you have ebox installed?
[03:22] <sommer> I think mine blew up right after upgrading ebox
[03:22] <kgoetz> sommer: i didnt install it, so unless it was a dependancy/automagic no
[03:22] <sommer> kgoetz: ah... probably unrelated then
[03:27] <sommer> sigh... the emails have started
[03:27] <sommer> the libc error is only hardy right?
[03:27] <kgoetz> heh. get ready for a 5 day storm
[03:27] <kgoetz> mine was on hardy.
[03:30]  * sommer updating an edgy test machine
[03:30] <ajmitch> heh, I see what you mean - that's an unhappy person on devel-discuss
[03:32]  * kgoetz senses another 'owch' level of breakage hitting
[03:32] <ajmitch> "it's hardy, you get what you pay for"
[03:32] <kgoetz> yep
[03:33] <pschulz01> ajmitch: "If you break it, you get to keep both bits"
[03:34]  * ajmitch should look on the forums & see how many people have out out death threats already
[03:38] <sommer> edgy updated fine, trying feisty... fyi
[03:58] <sommer> feisty fine as well
[03:59] <sommer> probably only hardy was effected
[04:00] <ivoks> zul: debdiff is in your inbox
[04:14] <owh> Can someone please remind me how I get the full path to an executable in a process list?
[04:19] <sommer> ps -ef or ps aux
[04:21] <owh> sommer: Yeah, but there are processes in my hardy process list, like say: imap-login and [kthreadd], that have no path.
[04:21] <owh> I'm sure I'm missing something :)
[04:23] <owh> In case you're wondering, I'm trying to map running processes to installed packages so I can find which of my init.d scripts belong to those packages, so I can add them to the list to figure out if they need a status option.
[04:24] <owh> I'd rather not manually map 68 processes, but I may have to.
[04:24] <sommer> ah... I'd think there's a ps options somewhere, if not should be some info in /proc
[04:25] <owh> Lemmie see what drops out the bottom if I omit the ones that obviously do have a path.
[04:27] <owh> That halves it, that's a lot better :)
[04:28] <owh> Out of ignorance: Are the processes inside square brackets executables, or internal to the kernel or something else - rather, might they be associated with packages or not?
[04:29] <owh> Examples: [aio/0] [ata/0] [ata_aux] [kacpid] [kacpi_notify] [kblockd/0] [khelper] [kjournald] [kpsmoused] [kseriod]
[04:30] <sommer> I believe those are kernel processes
[04:30] <owh> So for this exercise I can probably ignore them?
[04:34] <sommer> I would think so, but you may want to get a second opinion from someone more knowledgeable about packaging :-)
[04:34] <owh> Chicken :)
[04:35] <owh> I'll start without and work my way down.
[04:35] <sommer> hheheh... I'd just hate for you to waste a bunch of time on my say so
[04:35] <owh> It's ok, I'll find a way to 'help' you in return :)
[04:35] <owh> Thanks for the disclaimer tho.
[04:39] <sommer> np, I do what I can... I do what I can
[04:50] <sommer> anyone know why on hardy I need to use sudo to create self signed cert?  without sudo I get: unable to write 'random state'. from openssl
[04:59] <sommer> aside from there being a permissions issue?
[05:12] <owh> Crap, well that isn't going to work that well. Can someone please check my reasoning?
[05:12] <owh> I'm trying to map running processes to packages to init.d scripts to find if they implement status.
[05:12] <owh> Example, running process, apache2
[05:12] <owh> Obviously there is an init.d script.
[05:13] <owh> The apache2 executable comes from apache2-mpm-prefork
[05:13] <owh> The init.d script comes from apache2.2-common
[05:14] <owh> So, there isn't a map between the apache2 process and the init.d script that controls it.
[05:14] <owh> Of course, I might just have chosen a really bad random example, but I'm thinking not.
[05:14] <owh> Anyone disagree with any of this?
[05:18] <sommer> owh: not exaclty sure what you mean by "map"?
[05:18] <sommer> but your statements seem logical to me
[05:18] <owh> Find a relationship between, ownership.
[05:19] <owh> As in, this process comes from this package which also has this init.d script.
[05:19] <sommer> that's what I thought, I think it will depend package by package
[05:19] <owh> Hence my utterance earlier :)
[05:19] <sommer> more so for applications that are split over more than one package... like apache2
[05:20] <sommer> so ya, seems like you're on the right track to me :-)
[05:20] <owh> I suspect that it will be true for mysql, postgress, php, ldap, etc :(
[05:20] <owh> I didn't really want to get into rdeps :)
[05:21] <owh> Perhaps it won't be all bad, perhaps some/many of the processes come from the same package. It looks like I might have to do this manually.
[05:21] <owh> That's not really something that we can run across all of the distro though at any time.
[05:24] <sommer> there's always a silver lining... heh
[05:25] <owh> Is there any chance that a process that is running that comes from a package, in our case, /usr/sbin/apache2 comes from apache2-mpm-prefork, doesn't depend on a package that provides the init.d, or is that a big assumption that does not hold true?
[05:27] <sommer> I would hazard that if there is it's a bug
[05:27] <sommer> should be safe to assume
[05:28] <owh> So, I should be able to map a process to it's package, then find the depends that have an init.d and make the link, at least identify the init.d (or more if there are others)
[05:29] <owh> I mean for most it might well be the same package, but for apache it obviously isn't.
[05:37] <owh> Apache continues to be fun. It depends on apache2.2-common which provides /etc/init.d/apache2, but it also depends on libc6 which provides /etc/init.d/glibc.sh
[05:38] <owh> ROTFL
[05:39] <owh> Hmm, that didn't work either.
[05:39] <owh> I'll leave you all alone for a bit :|
[05:43] <sommer> owh: go ahead and update... what's the worst that could happen? :-)
[05:43]  * sommer past bed time... sigh
[05:43] <owh> Update?
[05:43] <sommer> hardy libc
[05:43] <owh> Nah, I'm not game :)
[05:43] <sommer> heh... don't blame ya
[05:43]  * owh saw the message to the list and the smack down :)
[05:44]  * owh is writing lovely code: cut -b76- hardy.process.list  | grep ^/ | cut -d' ' -f1 | sort | uniq | while read process ; do dpkg -S $process ; done | grep -v 'diversion by' | cut -d: -f1 | sort | uniq | while read
[05:44] <owh> :)
[05:45] <sommer> owh: party!
[05:45] <owh> It's a Piiiiiiipppppppeeeee
[05:47] <owh> Whoot, I think I have some lift off: cut -b76- hardy.process.list  | grep ^/ | cut -d' ' -f1 | sort | uniq | while read process ; do dpkg -S $process ; done | grep -v 'diversion by' | cut -d: -f1 | sort | uniq | while read package ; do init=`dpkg -L $package|grep /etc/init.d` ; if test "$init" ; then echo $process -\> $package -\> $init ; fi; done
[05:47] <owh> I suppose it should be prefaced with ps -af |
[05:48] <owh> and the hardy.process.list removed
[05:48] <owh> Making it:
[05:49] <kgoetz> you should probably quite the grep
[05:49] <kgoetz> *quote
[05:49] <owh> ps -Af | cut -b76- | grep ^/ | cut -d' ' -f1 | sort | uniq | while read process ; do dpkg -S $process ; done | grep -v 'diversion by' | cut -d: -f1 | sort | uniq | while read package ; do init=`dpkg -L $package|grep /etc/init.d` ; if test "$init" ; then echo $process -\> $package -\> $init ; fi; done
[05:49] <owh> :(
[05:49] <owh> That's wrong
[05:50] <kgoetz> whats it meant to do?
[05:51] <owh> Hold on, I broke some logic.
[05:52] <owh> kgoetz: It's in the process of giving me the init.d scripts that belong to running processes :)
[05:53] <owh> I'm trying to make it too pretty :)
[05:54] <kgoetz> owh: i see... looking in /etc/init.d not complex enough? :P
[05:54] <owh> Well, there's 68 in there and we only really want the ones that have a daemon running.
[05:57] <owh> ps -ealf | cut -b76- | grep ^/ | cut -d' ' -f1 | sort | uniq | while read process ; do dpkg -S $process ; done | grep -v 'diversion by' | sort | uniq | sed 's/://' | while read package process ; do init=`dpkg -L $package|grep /etc/init.d` ; if test "$init" ; then echo $process -\> $package -\> $init ; fi; done
[05:57] <owh> That's better.
[06:09] <owh> I love machine readable output, especially the kind that isn't.
[06:09] <owh> If you're wondering: apt-cache depends samba
[06:16] <owh> ScottK2 Well said on udd
[06:18] <ScottK2> Thanks
[06:50] <kraut> moin
[08:33] <henkjan> hmm, today no dist-upgrading for me
[08:34] <ivoks> :)
[08:34] <ivoks> i just recovered from it
[08:58] <mok0> This mornings libc6 upgrade is corrupt!
[08:59] <soren> It is.
[08:59] <mok0> My friend just smashed his hardy system
[09:00] <soren> An update is in the works.
[09:00] <soren> We need to write up some instructions for recovering, though.
[09:06] <owh> Just as a matter of interest, what's the process of notification for such instructions?
[09:06] <owh> I mean, if I'm a sysadmin, buried in the bowels of a server that is borked because something like this happened, how would I go about finding a solution?
[09:07] <owh> Is there an RSS feed for stuff like this for example?
[09:07] <owh> Or a mailing list?
[09:10] <henkjan> http://ubuntuforums.org/showthread.php?t=722886 has some instructions for recovering
[09:11] <owh> Another way to think of this might be to say: "Imagine this big organisation Ubuntu that has server farms scattered around the planet. (And workstations for that matter.) The people in this room are the back-line support for these devices. How would we notify our remote server teams (people whom we call system administrators) of outages or faults."
[10:17] <AnRkey> are there any nfs heros here?
[10:18] <AnRkey> i can't mount my nfs share with write permissions for love nor pizza
[10:19] <_ruben> AnRkey: sounds like a problem i had in the past .. cant remember if and if so, how, i fixed it tho :(
[10:19] <AnRkey> _ruben, common! i have to know :D
[10:20] <AnRkey> it's driving me nutes
[10:20] <AnRkey> nuts
[10:20] <AnRkey> _ruben, how did you mount the share in your fstab?
[10:42] <mok0> soren: on a live system: for f in ` dpkg -L`; do cp $f /media/blahblah; done
[10:42] <mok0> s/-L/-L libc6/
[10:43] <soren> Or just dpkg --root /media/blahblah -i /tmp/dpkg*.deb
[10:43] <soren> Er..
[10:43] <soren> Or just dpkg --root /media/blahblah -i /tmp/libc6*.deb
[10:44] <mok0> soren: Ah, another dpkg switch I didn't know about :-)
[10:45] <soren> :)
[10:47] <ProbeDot> i need to connect to it via internet, i have 50 xp stations connected to a 2003 server using roaming profiles (desktop,my documents,registery,etc are stored on the server), the internet gateway is the centos server, so i have a local network with local ip's and the internet network with the internet ip,
[10:47] <ProbeDot>  the problem is that i need a laptop using a internet conection to connect to the local area network via internet, i need all windows networking features available, somehow, that laptop must endup connected to the lan with a lan ip, i need to connect a laptop in austria to a local area network in france
[10:47] <ProbeDot> so that the laptop access al network resources of the lan in france, i want the laptop to act as a workstation in the lan, both laptop and lan have internet access, lan has one public ip, laptop same
[10:53] <_ruben> AnRkey: wouldnt know .. my recent attempts involved manual mounting usingn `mount host:/exported/path /mounted/path` .. since i needed to move stuff between two servers i just copied from one to other .. and deleted on the first one afterwards :p
[10:53] <AnRkey> this sounds crazy, but i need it running so i am using samba for now
[10:53] <AnRkey> it works and well.... it works :D
[10:53] <AnRkey> i will mess around with nfs again later
[10:54] <AnRkey> _ruben, thanks for the chat :D
[10:54] <AnRkey> _ruben, where are you? my sister in law wants to know
[11:03] <_ruben> AnRkey: at work .. netherlands .. ;)
[11:14] <sureshot> hey anyone know where extensions(modules) for php is located?
[12:04] <_ruben> hrm .. wonder what would be the appropriate way to compile/install/etc kernel modules that are available as <whatever>-source package only .. iscsitarget-source to be specific
[12:07] <_ruben> ah .. module-assistent
[12:16] <_ruben> bah .. iscsitarget-module doesnt wanna compile .. then again, im trying on desktop edition, not server (just as a test)
[12:22] <sommer> morning
[12:24] <_ruben> mornin sommer
[12:26]  * _ruben slaps Jeeves_ ... nl archive is b0rked :p
[12:26] <Kamping_Kaiser> hehe
[12:27] <_ruben> responds to ping and telnet on port 80 .. but doesnt return anything
[12:27] <henkjan> _ruben: we are investigating
[12:27] <_ruben> ah ok :)
[12:32] <henkjan> _ruben: nl.arhive is working again
[12:32] <zul> hey sommer
[12:32] <mok0> is the libc6 thing fixed in the archives?
[12:33] <_ruben> henkjan: lets test from here
[12:33] <_ruben> henkjan: nice .. what was the prob?
[12:34] <henkjan> _ruben: no idea yet. Jeeves_ went down to the colo
[12:35] <henkjan> server is rebooted
[12:35] <_ruben> ah
[12:35] <_ruben> nasty
[12:37] <henkjan> Mar 13 12:27:00 ftp kernel: [36632.422033] Out of socket memory
[12:37] <henkjan> Mar 13 12:27:03 ftp kernel: [36635.853495] TCP: too many of orphaned sockets
[12:39] <jdstrand> hi faulkes-!
[12:40] <jdstrand> I was asked in the #ubuntu-server meeting yesterday to coordinate with you an requesting that people test the new mysql updates in -proposed
[12:40] <jdstrand> faulkes-: this is mainly for dapper - feisty
[12:40] <jdstrand> faulkes-: see bug #201009 for details
[12:40] <ubotu> Launchpad bug 201009 in mysql-dfsg-5.0 "[mysql-dfsg-5.0] fix for several open vulnerabilities in -proposed" [High,Fix committed] https://launchpad.net/bugs/201009
[12:41] <jdstrand> faulkes-: is there an appropriate place in the forums that this can be arranged?
[12:41] <jdstrand> s/arranged/requested/
[12:45] <_ruben> henkjan: scary
[12:50] <Jeeves_> _ruben: Sorry about that
[12:50] <Jeeves_> It seems there were NFS issues
[12:50] <Kamping_Kaiser> Jeeves_, bugger.
[12:51] <Kamping_Kaiser> Jeeves_, i was going to ask how your going... buried in 'oh noes' atm probably
[12:51] <Jeeves_> Kamping_Kaiser: Yeah, I was rsyncing the iso's to the thumper
[12:51] <Jeeves_> But maybe I forgot a lockd  ...
[12:51] <Kamping_Kaiser> hehe.
[12:52] <Kamping_Kaiser> i almost did that badly today. set a server to an in use IP (read: conflict) then dist-upgraded my hardy and copped the glibc bug, then couldnt change the ip anymore
[13:04] <_ruben> Jeeves_: shit happens (to the best of us)
[13:04] <Jeeves_> Jups
[13:04] <_ruben> sweet .. my initial iscsi test succeeded .. win2k3r2 virtual server as initiator and ubuntu as target
[13:07] <faulkes-> on what hardware?
[13:09] <_ruben> faulkes-: dell dimension 5150 .. as in: one of the workstations under my desk ;) .. more a proof of concept than an actual (read: usefull) test ;)
[13:09] <_ruben> the target that is
[13:10] <_ruben> the next iscsi experiment will be on a machine with 8 sata disks and a raidcore/broadcom raid controller (fakeraid from what ive read, so software raid it'll be)
[13:10] <_ruben> tho that hardware isnt avail yet
[13:22] <faulkes-> ah
[13:22] <faulkes-> just looking at iscsi stuff right now, have a promise iscsi box nobody here has ever touched I'm going to hook up to dell 1955 blades
[13:23] <faulkes-> should be interesting to play with
[13:28] <_ruben> 1955s .. we're looking into those aswell .. for longer term plans as vmware esx hosts mainly
[13:29] <_ruben> we'll be getting some dell san stuff (borrowed from dell) to play with in a month or so .. need to finish off some other projects first
[13:29] <_ruben> really looking forward to dive into the virtualization and san stuff
[14:18] <zul> mathiaz: I uploaded samba-3.0.28a this morning
[14:18] <mathiaz> zul: awesome
[14:27] <jdstrand> mathiaz, soren, nijaba: just used ubuntu-vm-builder for a gutsy guest
[14:27] <soren> amd64 or i386?
[14:27] <jdstrand> and got grub 'Error 15' on boot
[14:27] <jdstrand> amd64
[14:27] <jdstrand> would you like to see the invocation?
[14:27] <soren> You need to pass --compontents main,restricted for the kernel install to function.
[14:28] <jdstrand> sudo ubuntu-vm-builder -a amd64 -d ./gutsy_server64 --domain strandboge.com --hostname gutsy-server64 --mirror http://192.168.122.1/ubuntu --addpkg openssh-server -s gutsy --vm kvm --user james --part ./partfile --kernel-flavour server --components main,restricted,universe,multiverse --ip=192.168.122.5 --libvirt qemu:///system
[14:28] <soren> Oh.
[14:28] <jdstrand> I did
[14:28] <soren> Is your mirror up-to-date?
[14:28] <soren> I mean *really* up-to-date?
[14:28] <faulkes-> hrmm, that reminds me, I need to turn VT on in the bios's
[14:28] <soren> Oh, it's gutsy.
[14:28] <soren> Never mind.
[14:28] <jdstrand> as of 07:00 UTC
[14:29] <jdstrand> the host is hardy
[14:29] <jdstrand> the guest is gutsy
[14:29] <soren> --ip=192.168.122.5 works? Wow. NEver tried that :)
[14:29] <jdstrand> soren: I can't speak to that
[14:29] <soren> I've always seperated them by spcaes.
[14:29] <nijaba> soren: I put that part in and I did test it ;)
[14:30] <soren> With '='? Ok, cool.
[14:30] <nijaba> soren: ah, no, always using space
[14:30] <soren> jdstrand: I'd much rather see the output of the build process :)
[14:31] <jdstrand> let me try without the '='-- I had some strange stuff that I was going to report separately
[14:33] <jdstrand> soren: interesting--
[14:33] <jdstrand> Searching for GRUB installation directory ... found: /boot/grub
[14:33] <jdstrand> findfs: Unable to resolve 'UUID=3812df5a-550f-48e9-8046-12d584257ec2'
[14:33] <jdstrand> Cannot determine root device.  Assuming /dev/hda1
[14:33] <jdstrand> This error is probably caused by an invalid /etc/fstab
[14:34] <jdstrand> but I'll paste the whole thing momentarily
[14:34] <soren> Don't worry about that one.
[14:36] <jdstrand> soren: it's booting
[14:36] <jdstrand> guess it *really* didn't like the '='
[14:38] <jdstrand> soren: had this http://paste.ubuntu-nl.org/59478/ at the end, which was disconcerting
[14:38] <jdstrand> soren: I am sure the openssh part was due to /dev/pts
[14:38] <jdstrand> well, *sure* is strong
[14:39] <jdstrand> I bet it is because of it "Can not write log, openpty() failed (/dev/pts not mounted?)"
[14:39] <mathiaz> jdstrand: I've seen that error too - it's because openssh cannot be started
[14:39] <jdstrand> mathiaz: yes
[14:40] <jdstrand> mathiaz: I was just saying it may not have been able to start bec of that
[14:40] <jdstrand> the last two lines made me think something wet wrong too
[14:40] <mathiaz> jdstrand: oh - I think it was not able to start because there is already an ssh daemon running on the host
[14:40] <jdstrand> just fyi
[14:41] <jdstrand> mathiaz: oh-- that makes sense
[14:41] <jdstrand> mathiaz: I was thinking that it was virtualized already, but of course it wasn't yet
[14:41] <mathiaz> jdstrand: I briefly mentionned it to soren in a meeting - apparently there is trick to make it work anyway
[14:42] <mathiaz> jdstrand: You'd run in the same problem when installing apache2 which couldn't be started as it cannot bind to port 80
[14:42] <jdstrand> it more just a usability thing for others, it clearly works well :)
[14:42] <jdstrand> mathiaz: yes I understand now
[14:42] <mathiaz> jdstrand: forget about the last comment - it could bind to port 80 as the install is run as root
[14:43] <jdstrand> mathiaz: not sure about that
[14:43] <soren> mathiaz: Unless you have a web server running already..
[14:43] <jdstrand> mathiaz: when I did testing in chroots of server daemons, I could only do one at a time
[14:44]  * jdstrand assumed mathiaz meant that it would be the same for any service that was already running on the guest
[14:45] <jdstrand> s/guest/host
[14:49] <Probedot_> i need to configure apache to allow access to http:\\url.com\dummy\download , But to block access ,to http:\\url.com\dummy on port 80,can you do it?
[14:49] <Probedot_>  (i do need access to dummy on 443, i have a site where the root folder is a login to a console i need blocked,but a subfolder of that folder is open to clients to access files from,  I dont want people to access the console but still have access to the downloads, with Virt hosts,is that how i set up ports? like if i want it to work via 443,do i use *:443
[14:49] <jdstrand> soren: where are the options for dnsmasq configured? I checked /etc/libvirt/* and /etc/default/libvirt-bin
[14:49] <jdstrand> soren: I'd like to use a different address range
[14:49] <jdstrand> soren: well, really just 'smaller'
[14:50]  * Probedot_ waits
[14:50] <jdstrand> so I can have static addresses not get in the way
[14:51] <soren> jdstrand: That's in your libvirt network xml.
[14:51] <soren> jdstrand: Use virsh to dump/redefine.
[14:51] <mathiaz> jdstrand: try a net-dumpxml from virsh to see how to set the range
[14:52] <jdstrand> ah /etc/libvirt/qemu/networks
[14:52] <jdstrand> got it
[14:52]  * jdstrand didn't dig deep enough
[14:52] <jdstrand> soren: sick of me and virtualization yet? ;)
[14:53] <jdstrand> (I know you are)
[14:53] <mathiaz> soren: re bug 194237 - why do you want to install usplash ?
[14:53] <ubotu> Launchpad bug 194237 in ubuntu-vm-builder "don't install usplash by default" [Undecided,New] https://launchpad.net/bugs/194237
[14:53] <mathiaz> soren: it's not installed by default on a server
[14:55] <soren> It's shiny.
[14:55] <soren> I thought I made that quite clear :)
[14:55] <soren> jdstrand: Not at all!
[14:56] <_ruben> hehe
[14:56] <mathiaz> soren: well... I'd expect a better argument thatn it's shiny.
[14:56] <jdstrand> :)
[14:56]  * jdstrand has *always* appreciated the shiny argument
[14:56] <mathiaz> soren: as I mentionned in the bug, there are some artifact left on the console when using vnc.
[14:57] <mathiaz> soren: that is the reason why I've filed the bug
[14:57] <mathiaz> soren: it's annoying to have orange pixels in the middle of the screen. it's not... shiny ;)
[14:58] <jdstrand> soren: I'd also mention that whatever loads kvm_intel is not using log_action_ for its output. as a result, that line is not in alignment with the other boot messages
[14:58] <jdstrand> log_action_ is usplash firendly
[14:58] <jdstrand> friendly
[14:59] <jdstrand> other log_ functions are not
[14:59] <Probedot_> i need to configure apache to allow access to http:\\url.com\dummy\download , But to block access ,to http:\\url.com\dummy on port 80,can you do it?
[14:59] <Probedot_>  (i do need access to dummy on 443, i have a site where the root folder is a login to a console i need blocked,but a subfolder of that folder is open to clients to access files from,  I dont want people to access the console but still have access to the downloads, with Virt hosts,is that how i set up ports? like if i want it to work via 443,do i use *:443
[15:01] <mathiaz> !servergui
[15:01] <ubotu> Sorry, I don't know anything about servergui - try searching on http://ubotu.ubuntu-nl.org/factoids.cgi
[15:07] <soren> mathiaz: Well, that's the bug we should be fixing then.
[15:17] <Jeeves_> Anyone from canonical here?
[15:18] <Jeeves_> If so. Can you gently push someone from the mirror-team towards #ubuntu-server? :)
[15:34] <Jeeves_> I'm not allowed to paste NSFW techpron here, right? :)
[15:37] <Jeeves_> Nobody here highlighting on NSFW? :)
[15:45] <nijaba> soren, mathiaz: any reco on how I should be implementing the templates in u-v-b?
[15:47] <soren> Jeeves_: What do you need mirror gusy for?
[15:47] <soren> guys
[15:49] <Jeeves_> soren: Already talking to elmo. Thanks though
[15:49] <nealmcb> !servergui
[15:49] <ubotu> Ubuntu server does not install a desktop environment or X11 by default in order to enhance security, efficiency and performance.  !eBox provides a GUI system management option via a web interface.  See https://help.ubuntu.com/community/ServerGUI for more background and options.
[15:49] <nealmcb> mathiaz: there you go!
[15:50] <jdstrand> mathiaz: when automating vm creation, what tool are you using?
[15:54] <mathiaz> jdstrand: custom scripts build around ubuntu-vm-builder :)
[15:55] <mathiaz> jdstrand: These are just shell scripts that call ubuntu-vm-builder with the right paths
[15:56] <jdstrand> mathiaz: are you assigning ip addresses or just using dhcp?
[15:57] <mathiaz> jdstrand: well - I'm hoping to use dhcp in fact
[15:57] <mathiaz> jdstrand: dnsmasq has this wonderfull feature to provide dns names for served ips
[15:58] <mathiaz> jdstrand: so I plan to integrate this in my environement so that I don't need to know which ip address I need to connect to
[15:59] <mathiaz> jdstrand: I aim at the following workflow: ubuntu-vm-builder host1; wait for 5 minutes ; ssh host1
[15:59] <jdstrand> mathiaz: where are the names exposed? (I am thinking of the traditional dynamic dns updates where you still end up having to know the hostname)
[15:59] <jdstrand> mathiaz: oh I see-- yes, that would rock
[15:59] <jdstrand> mathiaz: keep me posted :)
[15:59] <mathiaz> jdstrand: I've already that setup with vmware-server
[16:00] <mathiaz> jdstrand: I've modified vmware-server to use dnsmasq instead of dhcpd
[16:11] <nealmcb> ScottK, soren - good straightforward thoughtful comments on the u-d-d list - thanks
[16:22]  * soren hugs ScottK2
[16:22] <AnRkey> i want to import about 140 users into ubuntu
[16:23] <AnRkey> does anyone have any idea's for an import solution?
[16:27] <nealmcb> what does "break=bottom" do on the kernel boot line, anyway?
[16:29] <soren> It breaks out of the initramfs at a specific point.
[16:30] <soren> Most interestingly: after the root device has been mounted.
[16:30]  * ScottK2 hugs back.
[16:31] <nealmcb> soren: thanks.  I don't see it in Documentation/kernel-parameters.txt - is it grub-specific or ubuntu-specific or something?
[16:33] <soren> Neither :)
[16:33] <soren> It's from Debian.
[16:33] <soren> See /usr/share/initramsf-tools/init
[16:33] <soren> Er..
[16:33] <soren> /usr/share/initramfs-tools/init
[16:33] <soren> Look for "maybe_break bottom"
[16:33] <soren> That's where it'll bail out if you pass break=bottom.
[16:38] <sommer> mathiaz: thanks for adding links to the roadmap doc review section... I should have thought about that
[16:45]  * jdstrand -> lunch
[17:02] <zul> ivoks: around?
[17:03] <ivoks> zul: yes
[17:03] <zul> ivoks: I had a go through it looks ok. I added some stuff to it, changelog is here. http://pastebin.ca/941320
[17:03] <zul> Im going to ask for a FFE exception and merge the version from debian as well
[17:04] <ivoks> i noticed one thing tough... and i don't have an idea how to resolve it
[17:04] <ivoks> if someone has bacula-dir-mysql|pgsql before and now updates this new version
[17:04] <ivoks> postinst will ask for setting up new database
[17:05] <zul> so if bacula-version is older than blah then exit fi or something like that?
[17:05] <ivoks> ok, that was obvious :D
[17:05] <soren> In postinst: if [ "$1" = configure ]; then if [ -z "$2" ]; then echo "Ooh, a fresh install."; else echo "Oh, an upgade from $2"; fi; fi.
[17:06] <zul> soren: that works :)
[17:07] <soren> Or even better:
[17:07] <ivoks> zul: s/contorl/control :)
[17:07] <ivoks> in changelog
[17:07] <zul> yeah havent uploaded it yet :)
[17:08] <soren> In postinst: if [ "$1" = configure ]; then if dpkg --compare-versions 1.2.3-4ubuntu5 -lt "$2"; then echo "We're upgrading past version 1.2.3-4ubuntu5. Let's do some cool stuff."; fi; fi
[17:08] <zul> if dpkg --compare-versions "$2" lt 2.2.8-5ubuntu1 \ then do your bits else what soren said fi
[17:09] <soren> Er.. Yeah, I should have used -gt, probably.
[17:09] <soren> Er.. "gt", not "-gt".
[17:09] <zul> soren: great minds think alike
[17:09] <soren> So what zul said :)
[17:11] <ivoks> horrible debdiff
[17:12] <zul> ivoks: yeah it was fun...I just split the files up to make it easier
[17:12] <ivoks> :)
[17:12] <Theo__> how do i get libxerces26?
[17:12] <Theo__> I typed "apt-get install libxerces26" and it did not work
[17:13] <ivoks> Theo__: try 27 or 28
[17:13] <Theo__> ok
[17:14] <Theo__> Thank you
[17:14] <Theo__> it worked
[17:14] <jdstrand> oh that's fun
[17:15] <jdstrand> install dapper with ubuntu-vm-builder and boot it up and it goes into memtest86
[17:15] <zul> jdstrand: whhaaaa?
[17:15] <jdstrand> looking at grub boot menu, there is no kernel entry
[17:16] <mathiaz> jdstrand: you should check the build log and see which version of the kernel was asked to be installed
[17:16] <jdstrand> mathiaz: I was just now
[17:17] <mathiaz> jdstrand: I ran into a similar problem when tryinb to install hardy on amd64 - it tried to install linux-virtual which doesn't exist on amd64
[17:17] <jdstrand> I was trying to remember if -server is on dapper
[17:17] <ScottK2> No.  It's not.
[17:17] <mathiaz> jdstrand: I think so
[17:17] <ScottK2> OK.  Maybe I misremember, but I think not.
[17:18] <ScottK2> IIRC when I had dapper servers I just installed from the alternate ISO.
[17:18] <mathiaz> http://packages.ubuntu.com/dapper/linux-server
[17:18] <soren> jdstrand: I fixed that, too.
[17:18] <ScottK2> Fair enough.
[17:18] <soren> jdstrand: You can do:
[17:18] <soren> jdstrand: --flavour amd64-server
[17:19] <soren> jdstrand: if you're still on amd64, that is. If not, it's something else, which I haven't seen before.
[17:19] <jdstrand> soren: this is for an amd64 vm
[17:19] <mathiaz> jdstrand: well - linux-server on dapper exists only on i386
[17:19] <jdstrand> soren: is it only dapper or all that I should redo?
[17:19] <soren> jdstrand: Right. It's called amd64-server
[17:20] <jdstrand> ok cool
[17:20] <soren> jdstrand: dapper and edgy at least.
[17:20] <ivoks> is mail-server-task good name for binary meta package?
[17:20] <soren> Not sure about feisty.
[17:20] <soren> ivoks: No.
[17:20] <soren> :)
[17:20] <ivoks> :)
[17:20] <jdstrand> edgy worked...
[17:20] <soren> Because it's not a task.
[17:20] <ivoks> then just mail-server
[17:20] <mathiaz> jdstrand: linux-server was added to edgy
[17:20] <ivoks> one could thnik that's just MTA
[17:20] <ivoks> ubuntu-mail-server?
[17:21] <mathiaz> ivoks: if you're talking about sasl, I'd create a package named postfix-sasl-enable or something like that
[17:21] <mathiaz> ivoks: I'm still not sure how we should do this.
[17:21] <ivoks> mathiaz: that name would be misleading cause we include more that just sasl configuration
[17:22] <mathiaz> ivoks: actually no.
[17:22] <mathiaz> ivoks: the plan would be to move the dovecot stuff directly in the default dovecot configuration file
[17:22] <mathiaz> ivoks: which leaves only the postconf calls in the postinst script
[17:23] <ivoks> mathiaz: even the maildir and sasl setup in dovecot?
[17:23] <mathiaz> ivoks: we could enable sasl by default in dovecot
[17:23] <ivoks> mathiaz: the changes we propose are stricly for postfix
[17:23] <Theo__> Does anyone know how to setup a MTA server?
[17:24] <ivoks> so, if someone uses exim, dovecot wouldn't even start
[17:24] <jdstrand> ok, that is much more fun-- it's booting :)
[17:24] <ScottK2> Theo__: Yes.
[17:24] <Theo__> I started the server but cannot connect
[17:25] <mathiaz> ivoks: why ?
[17:25] <ivoks> dovecot creates socket in /var/spool/postfix
[17:25] <mathiaz> ivoks: well - that's not so wrong
[17:26] <mathiaz> ivoks: I still wonder if we should move this to a different path
[17:26] <mathiaz> ivoks: and then set postfix to use this new path
[17:26] <ivoks> postfix is jailed
[17:26] <ivoks> so it must be inside /var/spool/postfix
[17:26] <ivoks> or... we can bind mount
[17:26] <lamont> having dovecot allow multiple paths to the file would fix things...
[17:27] <ivoks> lamont: iirc, it actualy allows that
[17:27] <mathiaz> I was more worried about the effect of having dovecot putting something in /var/spool/postfix
[17:27] <mathiaz> if that works with lamont - it's all good
[17:28]  * lamont has no objections to packages delivering things into the chroot to let them work.  well, assuming that they aren't utter crack.
[17:28] <lamont> and if the admin installs dovecot onto a box with postfix, then he probably means to have it work.
[17:28] <lamont> so, yeah, +1
[17:28] <ScottK2> Theo__: What did you install, what Ubuntu release are you using, and more exactly what's the trouble?
[17:29] <ivoks> no, dovecot doesn't support multiple SASL clients - so, only one socket
[17:29] <Theo__> I have ubuntu server edition, libxerces26 and libstdc++6
[17:29] <Theo__> . When I start the server there are no errors. I cannot connect to it through lan from another computer
[17:29] <ScottK2> Personally I think it's quite reasonable to have the default dovecot configuration be the one that works with postfix.  It's our primary MTA.
[17:30] <mathiaz> ivoks: well - we could make sure that dovecot is integrated with the postfix
[17:30] <mathiaz> ivoks: I agree with ScottK2 ^^
[17:30] <ivoks> ScottK2: then dovecot should depend on postfix and other way around
[17:30] <mathiaz> ivoks: not needed
[17:30] <ivoks> cause postfix won't start if it can't connect to dovecot's socket
[17:31] <mathiaz> ivoks: postfix isn't configured by default to use sasl
[17:31] <ivoks> i see where you're heading
[17:31] <ScottK2> ivoks: I think depends postfix|mail-transport-agent is quite reasonable.
[17:31] <mathiaz> ivoks: that the role of the new package - postfix-sasl-configure
[17:31] <ivoks> ok
[17:32] <ScottK2> Dovecot does need an MTA, but the MTA doesn't necessarily need Dovecot.
[17:32] <mathiaz> ScottK2: dovecot really needs an MTA installed ?
[17:32] <Theo__> what
[17:32] <ScottK2> mathiaz: Does it?
[17:32] <ScottK2> mathiaz: Sorry, speculating again.
[17:32] <ivoks> it does not
[17:33] <ScottK2> OK.  Nevermind about that bit then.
[17:33] <Theo__> MTA=Multi Theft Auto?
[17:33] <ivoks> ok, so postfix by default doesn't support sasl
[17:33] <mathiaz> !mta | Theo__
[17:33] <ubotu> Theo__: A Mail Transfer Agent (MTA) is the server software that sends and queues mail. The default MTA (and !MDA) on Ubuntu is !postfix ("exim" is also officially supported). See also !MailServer and !MUA
[17:33] <ScottK2> Theo__: Mail Transport Agent
[17:33] <ivoks> and dovecot by default provides sasl socket?
[17:33] <mathiaz> ivoks: yes.
[17:33] <ivoks> one important thing
[17:34] <ivoks> if we pre-setup dovecot for sasl socket for postfix (/var/spool/postfix)
[17:34] <mathiaz> ivoks: how would dovecot react if the path to the sasl socket doesn't exist ?
[17:34] <ivoks> and postfix isn't installed
[17:34] <ivoks> dovecot will fail to start
[17:34] <ScottK2> That would be bad.
[17:34] <ivoks> but...
[17:34] <mathiaz> ivoks: that means dovecot needs to create /var/spool/postfix or whatever.
[17:34] <ivoks> mathiaz: or that... right
[17:35] <ivoks> or... bind directories
[17:35] <ivoks> dovecot starts socket on /var/run/dovecot/auth
[17:35] <mathiaz> that's the bit I'm worried about
[17:35] <ivoks> and postfix init script binds it to /var/spool/postfix/private/dovecot-auth
[17:35] <ivoks> but that's so... hacky :/
[17:36] <ivoks> it would be much easier if dovecot would start tcp socket :/
[17:37] <mathiaz> ivoks: so another option is to look into dovecot src code and make it exit properly if the path to the socket doesn't exists
[17:38] <mathiaz> ivoks: if /var/spool/postfix/private/ doesn't exits, don't start the sasl daemon
[17:38] <ivoks> it's stupid that it doesn't act like that already
[17:38] <mathiaz> s/exits/exiSt/
[17:38] <ivoks> cause it only exports auth socket
[17:38] <ivoks> it's not like it uses it
[17:43] <ivoks> is it wise to leave a unused socket which can be used for probing usernames?
[17:55] <sommer> !servergui
[17:55] <ubotu> Ubuntu server does not install a desktop environment or X11 by default in order to enhance security, efficiency and performance.  !eBox provides a GUI system management option via a web interface.  See https://help.ubuntu.com/community/ServerGUI for more background and options.
[17:55] <sommer> heh
[17:56] <nealmcb> :-)
[17:58] <ivoks> hm...
[18:00]  * jdstrand thinks it is just *awesome* that he can use multi-cpu in kvm
[18:15] <AnRkey> i want to import about 140 users into ubuntu
[18:15] <AnRkey> does anyone have any idea's for an import solution?
[18:15] <AnRkey> i want to make the username and passwords the same
[18:18] <nealmcb> AnRkey what about home directories and stuff - just make new ones, or migrate from somewhere?  uids?
[18:18] <nealmcb> and what do you have for them now - passwd and shadow files from some other system?
[18:18] <AnRkey> make new ones
[18:19] <AnRkey> i was going to make a small script with adduser
[18:19]  * nealmcb nods
[18:19] <nealmcb> so you know all their passwords?
[18:19] <AnRkey> cant figure out how to add the password to the adduser script
[18:19] <AnRkey> well i want to make the login and password the same
[18:19] <AnRkey> it's a small internal network
[18:20] <AnRkey> it's just for a mail server
[18:20] <nealmcb> ok - so not really an import, exported from somewhere else - you just want to make them up from scratch
[18:26] <AnRkey> well i have the list from the old server
[18:26] <AnRkey> the old server was setup in the same way
[18:26] <AnRkey> login and pass where also the same
[18:26] <AnRkey> we want to make the migration to the new server seamless
[18:30] <ivoks> don't use adduser then
[18:31] <ivoks> use useradd and set up password with chpasswd
[18:32] <AnRkey> thanks ivoks, looking into that
[18:32] <nealmcb> ivoks: cool - that is what I was searching for - chpasswd allows AnRkey to use the old passwords from the old shadow file also
[18:33] <AnRkey> nealmcb, use old shadow file?
[18:33]  * nealmcb hates the idea of 140 users all knowing that their peers password is the same as the userid
[18:33] <AnRkey> nealmcb, that's just it, they don't know
[18:33] <AnRkey> :D
[18:33] <AnRkey> most of these people are farmers and have no idea what copy and paste is
[18:34] <nealmcb> the -e and -m options of chpasswd look like they'll take the sort of stuff you can get out of /etc/shadow
[18:34] <nealmcb> so you can just say that their password hasn't changed
[18:34] <AnRkey> shweet, thanks nealmcb ivoks
[18:34] <nealmcb> :-)
[18:35] <ivoks> np
[18:36] <AnRkey> looks like i might get some sleep tonight
[18:36] <AnRkey> :P
[18:38] <nealmcb> huh - I guess I misread the chpasswd man page on -m:  "Use MD5 encryption instead of DES when the supplied passwords are not encrypted"  - the "not" is surprising.   I guess it knows md5 vs des from the way it is encrypted?
[18:45] <AnRkey> yeah that does not look like the correct switch
[18:45] <AnRkey> i have the old shadow file so it looks like i'll be fine
[19:06] <nealmcb> soren, sommer: I added the "break" option to https://help.ubuntu.com/community/BootOptions - the mysteries of initramfs are slowly unveiling themselves :-)
[19:20] <sommer> nealmcb: that's good stuff, I know like 2.5 boot options :-)
[19:20] <sommer> anyone have a favorite sql reference or website?
[19:25] <mathiaz> sommer: I've found the Mysql User Manual good at that usually
[19:26] <kirkland> sommer: not a website, but I bought this book in 1999, and learned everything I've needed to know about SQL from it.  I still pull it off the shelf occasioinally:  http://www.amazon.com/Special-Using-SQL/dp/0789719746/ref=sr_1_3?ie=UTF8&s=books&qid=1205436337&sr=8-3
[19:27] <sommer> mathiaz: yep, I've linked to that already, looking for more references for the database section
[19:27] <sommer> kirkland: cool, thanks I'll take a look
[19:27] <mathiaz> sommer: and you can even find it in ubuntu - a simple apt-get install mysql-doc will make you happy
[19:27] <sommer> oooh, that's a good point
[19:28] <kirkland> sommer: also, i'm more of a postgres junkie, so i tend to use http://www.postgresql.org/docs/ for db specific questions
[19:28] <sommer> kirkland: ah, I'll link that as well
[19:28] <mathiaz> sommer: and may be mention sqlite also
[19:28] <kirkland> sommer: you can go straight to the appropriate manual, if you want
[19:29] <mathiaz> sommer: these are the main databases
[19:30] <sommer> ya, I was thinking about an sqlite section... I'm a little hesitant to add a bunch of new content so close to SF
[19:30] <mathiaz> !servergui
[19:30] <ubotu> Ubuntu server does not install a desktop environment or X11 by default in order to enhance security, efficiency and performance.  !eBox provides a GUI system management option via a web interface.  See https://help.ubuntu.com/community/ServerGUI for more background and options.
[19:30] <sommer> Matt East was worried about translation time
[19:31] <sommer> if not for hardy it'll definitely be there for ibex :-)
[19:38] <jdstrand> faulkes-: ping re mysql testing
[19:50] <nealmcb> sommer: Is the virt section draft for the server guide somewhere where I can point folks at it?
[19:51] <sommer> nealmcb: you bet: http://doc.ubuntu.com/ubuntu/serverguide/C/virtualization.html
[19:51] <sommer> I commited an update last night so that's out of date, until the update process runs
[19:52] <sommer> not sure when that is exaclty though
[19:54] <nealmcb> sommer: cool - I should have known that - got confused by the versions again....
[19:55] <sommer> np... mathiaz also added links to the roadmap :-)
[20:11] <nealmcb> sommer: is the vmware line in http://doc.ubuntu.com/ubuntu/serverguide/C/jeos.html et al. still necessary, or is it running well in kvm also?  Or perhaps the following sections should be renamed to "vmware-xxx", since jeos seems a more generic concept?
[20:14] <sommer> nealmcb: that's a good question... I like where your head's at!
[20:16] <sommer> personally I haven't tested jeos with kvm, but I believe from what others have said is that it works fine
[20:16] <sommer> nijaba or soren can probably shed more light on that one
[20:17] <sommer> my plan is to do some testing this evening... so probably know more than :-)
[20:17] <nijaba> sommer: I'm in a meeting right now.  should be avail in 30 min or so...
[20:18] <sommer> cool
[20:22]  * nealmcb is also on a phone call now - off in 40 minutes or so
[20:35] <jdstrand> mathiaz: I have theory about bug #105457
[20:35] <ubotu> Launchpad bug 105457 in mysql-dfsg-5.0 "mysqd_safe high cpu usage" [Low,Triaged] https://launchpad.net/bugs/105457
[20:35] <jdstrand> I updated the bug accordingly, but don't really have time to look into it now
[20:36] <jdstrand> let's go with hypothesis instead
[21:14] <nijaba> sommer, nealmcb: I think we need to revisit the Jeos tutorial to a/include usage of ubuntu-vm-builder (even though it is universe) b/amend it for kvm usage as well
[21:18] <nealmcb> nijaba: right on!
[21:18] <nealmcb> and I suppose libvirt builds are a third possibility
[21:18] <nealmcb> but I haven't tried that stuff
[21:20] <nijaba> nealmcb: ubuntu-vm-builder now include a libvirt option...
[21:21] <nealmcb> nijaba: great!
[21:21]  * nealmcb is so out-of-it, but hoping to find time to get back up to speed....
[21:22] <nijaba> I am currently working on adding some templating to the xml file.
[21:24] <nealmcb> I'm talking with a guy who is having trouble with vmware and hardy - not in partner repo now?  missing headers when he tries it straight from vmware?
[21:25] <nijaba> nealmcb: vmware-server for hard should be in the repo sometimes around GA
[21:28] <nealmcb> Good
[21:33] <nealmcb> at some point, jeos descriptions for xen and perhaps even containerized solutions would be nice also, either on the wiki or in the server guide
[21:35] <nijaba> nealmcb: as xen / openvz are community supported solutions, I'd vote for wiki
[21:35]  * nealmcb nods
[21:36] <nealmcb> nijaba: so is canonical supporting vmware on ubuntu?
[21:37]  * nealmcb wonders about vmware in the server guide for the same reason
[21:37] <nijaba> nealmcb: no, we are working with vmware on having ubuntu (server, including Jeos) certified on ESX
[21:37] <nijaba> nealmcb: in the same principle as another hw vendor in fact
[21:38] <nealmcb> so should that be on the wiki instead also?
[21:38] <nijaba> nealmcb: I am not following you..  which "that"?
[21:38] <nealmcb> vmware
[21:39] <nealmcb> have the server guide focus on supported open source solutions
[21:39] <nijaba> nealmcb: vmware(-server) is not part of the official doc, only running Jeos on vmware is
[21:39] <nijaba> afaik
[21:40] <nijaba> and it is not a "canonical" reasoning, more a "main" vs "universe" one
[21:42] <kirkland> mathiaz: are you here?
[21:42] <kirkland> mathiaz: owh and I have a few questions about the init script lsb work, and how best to document this in LP
[21:42] <kirkland> mathiaz: here's what I'm thinking....
[21:42] <kirkland> mathiaz: we create one LP bug for general initscript LSB compliance
[21:42] <nealmcb> I may be confused on just what is needed to run jeos on vmware - is the necessary stuff in main? universe? partner?
[21:43] <kirkland> mathiaz: target that bug for Intrepid, really
[21:43] <kirkland> mathiaz: and file individual bugs against each package with a non-LSB compliant init script
[21:43] <nijaba> nealmcb: I think the necessary stuff is in main (and you can add open-vm-tools from universe if you want)
[21:44] <kirkland> mathiaz: ideally, the LSB tests would tell us which are and are not compliant, we still need to try their tests
[21:44] <nealmcb> so the partner stuff is just extra fancier stuff for building vms?
[21:44] <kirkland> mathiaz: as owh, and i, and anyone else wants, we chip away at those bugs, most are pretty simple fixes, great little work items for new developers
[21:45] <nijaba> nealmcb: yep, that's how I would view it
[21:45] <kirkland> mathiaz: we will document in a wiki page exactly how to fix up an init script
[21:45] <nealmcb> and it is vmware-server?
[21:45] <kirkland> mathiaz: it's fairly formulaic, actually
[21:45] <mathiaz> kirkland:  I like the "great little work for new developers" :)
[21:45] <kirkland> mathiaz: i don't mean to trivialize it.....
[21:45] <kirkland> mathiaz: but it shouldn't be that scary
[21:45] <mathiaz> kirkland: oh no - I really think it's difficult
[21:46] <mathiaz> kirkland: heu... *not difficult*
[21:46] <kirkland> mathiaz: why so?
[21:46] <nijaba> nealmcb: vmware-server is in partner repo, so that's really canonical's pmau
[21:46] <kirkland> mathiaz: oh :-)
[21:46] <nijaba> s/pmau/play
[21:46] <kirkland> mathiaz: also, I think we throw FFe out the window entirely--the number of init scripts affected by this is tremendous
[21:46] <owh> nijaba: What keyboard are you typing on?
[21:46]  * mathiaz remembers how an azerty keyboard looks like
[21:46] <nijaba> owh: frenglish
[21:47] <kirkland> mathiaz: so we (the community) just cips away at them over time
[21:47] <kirkland> mathiaz: perhaps some make it into Hardy
[21:47] <kirkland> mathiaz: hopefully more make it into hardy+updates
[21:47] <kirkland> mathiaz: and ideally we're lsb-compliant by intrepid
[21:47] <kirkland> mathiaz: (profit)
[21:47] <kirkland> mathiaz: owh: sound like a plan?
[21:48] <owh> kirkland: mathiaz: wfm
[21:48] <mathiaz> kirkland: well - the init scripts changes won't make it in hardy-updates
[21:48] <kirkland> mathiaz: if so, this is what I'll write up in the roadmap, per meeting action yesterday
[21:48] <kirkland> mathiaz: no?  security issues only make it in there?
[21:48] <mathiaz> kirkland: they won't qualify for a StableReleaseUpdate
[21:48]  * kirkland goes look up StableReleaseUpdate guidelines
[21:49] <mathiaz> !sru
[21:49] <ubotu> Stable Release Update information is at http://wiki.ubuntu.com/StableReleaseUpdates
[21:49] <mathiaz> !sru | kirkland
[21:49] <ubotu> kirkland: please see above
[21:50] <mathiaz> kirkland: then rather than filing bugs about lsb-compliant scripts, I'd rather concentrate on "implement status action" for relevant scripts.
[21:50] <mathiaz> LSB-compliant scripts requires more than just a status action
[21:50] <kirkland> mathiaz: really?  okay.  that's easier
[21:50] <mathiaz> kirkland: you need the headers, the correct exit code
[21:51] <owh> mathiaz: Well, the init.d scripts that come with hardy tasksel full install all include the lsb_functions which indicates that the author is at least aspiring to compliance.
[21:52] <nealmcb> nijaba: I just sent you some mail - perhaps you can help us get to more clarity....
[21:52] <kirkland> mathiaz: just sourcing /lib/lsb/init-functions and using pidofproc gets the right exit code
[21:53] <mathiaz> kirkland: right - but there are other places in the init script that may use the wrong exit code  - see apache2 init script
[21:53] <kirkland> mathiaz: right, agreed.
[21:53] <mathiaz> owh: correct. But they may be not completly compliant.
[21:53] <kirkland> mathiaz: so what you're saying is concentrate on implementing the status section
[21:54] <mathiaz> owh: and some of the changes are actually made by Ubuntu.
[21:54] <mathiaz> owh: and we still carry the delta with debian.
[21:54] <mathiaz> kirkland: yes - just the status action.
[21:54] <kirkland> mathiaz: what if we added a function to /lib/lsb/init-functions
[21:54] <owh> mathiaz: Yes, and changing the exit codes might break code that relies on the "faulty/non compliant" exit codes.
[21:54] <nijaba> nealmcb: would you like me to reply here or by email?
[21:54] <owh> kirkland: That's what I wanted to talk about :)
[21:54] <mathiaz> kirkland: it should be easier to implement, more targeted and easier to get a FFe for it.
[21:55] <kirkland> mathiaz: statusof $DAEMON $NAME
[21:55] <owh> Now we're talking :)
[21:55] <kirkland> mathiaz: I think most of them should look a lot like:
[21:55] <kirkland> +status)
[21:55] <kirkland> +	pidofproc /usr/sbin/cron >/dev/null
[21:55] <kirkland> +	status=$?
[21:55] <kirkland> +	if [ $status -eq 0 ]; then
[21:55] <kirkland> +		log_success_msg "Cron is running."
[21:55] <kirkland> +	else
[21:55] <kirkland> +		log_failure_msg "Cron is not running."
[21:55] <kirkland> +	fi
[21:55] <kirkland> +	exit $status
[21:55] <kirkland> +	;;
[21:55] <kirkland> (Sorry for the code drop, it's small though)
[21:55] <kirkland> DAEMON=/usr/sbin/cron
[21:55] <kirkland> NAME=Cron
[21:56] <kirkland> and status just becomes
[21:56] <kirkland> a call to statusof
[21:56] <nealmcb> nijaba: by mail is probably better, and I can forward to our colorado list as appropriate
[21:56] <kirkland> owh: mathiaz ?
[21:56] <kirkland> mathiaz: that might be FFe-able?
[21:56] <mathiaz> kirkland: yes - that'd make sense.
[21:56]  * nealmcb hopes he hasn't confused his buddies
[21:57] <mathiaz> kirkland: I'd have a quick look at the existing init scripts that have a status action to see they would fit the pattern.
[21:57] <nealmcb> nijaba: thanks!!
[21:57] <owh> kirkland: Yup that's what I intended to get to :)
[21:57] <kirkland> owh: you da man
[21:57] <kirkland> owh: we're on the same page, i like that \o/
[21:57] <owh> kirkland: I can't get credit, I thought of it first, but you wrote it down first :)
[21:57] <mathiaz> kirkland: that would bring more weight to the FFe
[21:58] <kirkland> owh: if you want the credit, you hack the patch to /lib/lsb/init-functions :-)
[21:58] <mathiaz> kirkland: I'd also send an email to ubuntu-devel outlining the reason to do this
[21:58] <owh> ROTFL - mine!
[21:58] <mathiaz> kirkland: you'd get more input from all the other developers.
[21:59] <kirkland> mathiaz: sounds good, probably should be done ASAP, i assume?
[21:59]  * owh is already downloading source :)
[21:59]  * kirkland downloads from his local mirror at 1Gbps :-)
[22:00] <mathiaz> kirkland: yes.
[22:00] <kirkland> :-P owh
[22:00] <kirkland> mathiaz: owh: give me a minute, let me draft something and pastebin it
[22:00] <mathiaz> kirkland: try to come up of list of packages that you'd change, which one would benefit from it
[22:00]  * owh has a vSat terminal with 1Mbit that can be setup anywhere, so there is no need to be indoors :)
[22:01] <mathiaz> kirkland: once you've got a list of packages to fix, look at bug 199014 for a way to track the work done on it.
[22:01] <ubotu> Launchpad bug 199014 in emesene "python-xml removal: please drop/replace (build) dependencies" [Medium,In progress] https://launchpad.net/bugs/199014
[22:02] <kirkland> mathiaz: cool
[22:10] <nijaba> nealmcb: reply in our mail.  hope it is clear...
[22:12] <skwashd> hi, are server related hardy issues dealt with here or #ubuntu+1?
[22:12] <skwashd> i would prefer not to waste my time explaining it all here only to be told "go next door"
[22:15] <nealmcb> nijaba: great - many thanks.  so while ubuntu is focussed on kvm, as free software, canonical is also continuing to work with others to improve how other vm technologies work on the platform
[22:16] <nijaba> nealmcb: exactly
[22:16] <skwashd> nealmcb: pity kvm is currently useless for real world stuff in hardy
[22:17] <mathiaz> skwashd: could you be more explicit ?
[22:17] <nijaba> skwashd: you should talk about this with soren, I do not think he would agree. Me neither.  Now, it is clear that we have some work to do before release, but wait, and you'll see...
[22:17] <mathiaz> skwashd: I use kvm on a daily basis and it works great.
[22:17] <skwashd> mathiaz: sure ... bridged networking has been broken for almost 2 weeks
[22:18] <skwashd> and trying to get any help to confirm the problem is like pulling teeth
[22:18] <skwashd> this is on hardy
[22:18] <nijaba> skwashd: make sure the bug is reported (and this one is) and it will be fixed
[22:18] <nijaba> skwashd: hardy is alpha...
[22:19] <skwashd> nijaba: is it ... i can't find it on launchpad
[22:19] <skwashd> nijaba: i know the state of hardy
[22:19] <skwashd> someone on #kvm has suggested that tap io is going to the console ...  but no suggested fix
[22:19] <nijaba> skwashd: I think it is, but you should check and report if not
[22:20] <skwashd> nijaba: i am still trying to track down exactly where the problem is ... and report the bug against the right package
[22:20] <nijaba> skwashd: ok, thanks for the search then :)
[22:20] <skwashd> and i have given launchpad a good workout .... haven't seen anything relevant
[22:21] <nijaba> skwashd: you should maybe go talk about it on #ubuntu-virt.  I know kees and soren have been troubleshooting bridge networking together last week
[22:22] <skwashd> nijaba: ok ... will do
[22:22] <skwashd> #ubuntu* chans breed quicker than a couple without a tv
[22:22] <nijaba> skwashd: I like the metaphore :)
[22:23] <skwashd> nijaba: i took the religious part out - coc compliance :)
[22:28] <nealmcb> skwashd: lol - and thanks for helping make kvm better!
[22:28] <skwashd> nealmcb: np :)
[22:28] <skwashd> i love the idea of kvm
[22:28] <nealmcb> nijaba: thanks - I forwarded that on with refs to a few more things you mentioned, etc
[22:29] <nijaba> nealmcb: I just saw that (along with the typos in my reply).
[22:30] <nealmcb> nijaba: oh - I didn't notice them.  just my own "expert advice from an expert" inexpertitude....
[22:30] <nealmcb> :-(
[22:31] <nijaba> nealmcb: no worry, I should have taken the time to proofread, and I think it is still quite understandable ;)
[23:03] <kirkland> mathiaz: are you still around?
[23:11] <kirkland> ScottK2: hi, are you around?
[23:11] <kirkland> ScottK2: I was looking for some feedback on the init script stuff, and I thought you might have an opinion on a draft of an email i was about to send to ubuntu-devel
[23:20] <kirkland> all: this is an email I'm drafting to ubuntu-devel regarding the "status" action in init scripts.  I'd like any feedback from we-the-server-guys before I drop it on ubuntu-devel: http://ubuntu.pastebin.com/m6eb6cdbd
[23:27]  * ScottK2 looks
[23:29] <ScottK2> kirkland: My suggestion would be to discuss this with release management before you send the mail and get one big FFe for the project.  If they say no, make it a spec for Hardy +1.
[23:30] <kirkland> ScottK2: okay, good idea.  That's what I was hoping for, one blanket FFe.
[23:31] <kirkland> ScottK2: only for a list of defined, agreed-upon services for Hardy
[23:31] <kirkland> ScottK2: and a full spec for Hardy+1
[23:31] <ScottK2> Right.
[23:31] <kirkland> ScottK2: very good, thank you for your input
[23:31] <kirkland> ScottK2: I'll run that list of suggested packages to be affected by this too, if that's cool
[23:32] <ScottK2> I'd definitely have that list handy when you go talk to slangesek.
[23:42] <steve176> Hi. How do I get tomcat to start after reboot? sudo /etc/init.d/tomcat start works from the command line and I have S71tomcat symlinks in /etc/rc2.d -> /etc/rc5.d
[23:43] <mathiaz> kirkland: re your email
[23:43] <mathiaz> kirkland: I wouldn't mention "a great number of package"
[23:43] <mathiaz> kirkland: I'd say that the target for hardy are just daemons that are shipped on the ubuntu-server isos.
[23:44] <kirkland> mathiaz: k
[23:44] <kirkland> mathiaz: i was trying to avoid defining the list in this email
[23:44] <kirkland> mathiaz: and do that in the launchpad bug instead
[23:45] <mathiaz> kirkland: right - good idea.
[23:45] <kirkland> mathiaz: try to keep this mail focused on the functionality itself
[23:45] <kirkland> mathiaz: see if there are arguments with that
[23:45] <mathiaz> kirkland: but mentionning a huge amount of potential packages can be scary at this stage of the release.
[23:46] <mathiaz> kirkland: so the tactic would be to suggest what we try to aim for hardy ;)
[23:46] <sommer> steve176: sudo update-rc.d tomcat5.5 default
[23:46] <kirkland> mathiaz: perfect, agreed.
[23:47] <mathiaz> steve176: if you have a symlink in rc3.d it should start on boot.
[23:47] <steve176> sommer: usage error
[23:47] <steve176> sommer: do you mean defaults?
[23:47] <sommer> steve176: yep, sorry defaults
[23:48] <steve176> mathiaz: thats what I thought. One thing I thought of - maybe JAVA_HOME / CATALINA_HOME (exported by /etc/profile) aren't available that early during boot
[23:48] <mathiaz> steve176: you shouldn't need to setup for start at boot. This should be done when the package is installed.
[23:48] <infinity> mathiaz: rc2.d, even.  Our default runlevel is 2, unless someone mangled something very recently...
[23:48] <steve176> mathiaz: Not a package - downloaded and expanded the tar
[23:48] <owh> Tsk Tsk
[23:49] <infinity> steve176: /etc/profile is most likely not sourced on boot at all.
[23:49] <infinity> steve176: I'd imagine your init script will want to set its environment.
[23:49] <steve176> So the only way is to hard code tomcat home in /init.d/tomcat
[23:50] <mathiaz> steve176: is there a reason why you're not using tomcat from the ubuntu repository ?
[23:50] <steve176> mathiaz: I didn't think to check there was one
[23:50] <infinity> steve176: /etc/profile is only sourced for login shells.
[23:50] <owh> 5.5
[23:50] <steve176> infinity: thx, that's bound to be part of the problem
[23:51] <mathiaz> steve176: apt-get install tomcat5.5 will do all the work for you
[23:51] <kirkland> mathiaz: you can see a diff here: http://ubuntu.pastebin.com/pastebin.php?diff=m6820f91d
[23:51] <owh> steve176: Specifically version: 5.5.25-5ubuntu1
[23:51] <kirkland> mathiaz: see if that verbage is better
[23:51] <steve176> mathiaz: sometimes it's nice to do it by hand - that way I'll always have control of the version
[23:52] <mathiaz> kirkland: wfm
[23:52] <kirkland> mathiaz: okay, so, i want to file the launchpad bug first, so that I can reference it in the email
[23:53] <kirkland> mathiaz: i was going to file against lsb-base, is that what you recommend?
[23:53] <kirkland> owh has a debdiff patch
[23:53] <kirkland> mathiaz: http://ubuntu.pastebin.com/m3f24c004
[23:53] <kirkland> mathiaz: that's for the common function
[23:54] <mathiaz> kirkland: I'd send the email first.
[23:54] <kirkland> mathiaz: okay
[23:54] <mathiaz> kirkland: to ubuntu-devel@lists.ubuntu.com
[23:55] <steve176> mathiaz, infinity, sommer: looks like that fixed it. cheers
[23:55] <mathiaz> kirkland: your message may be moderated because you're not a developer yet.