[00:57] <zanzacar> ok thanks
[03:10] <lkthomas> hey guys
[03:11] <lkthomas> when I see number of processes on server, it shows around 420 processes running
[03:11] <lkthomas> while the loading on each CPU doesn't go more than 5% and it's 16 cores
[03:11] <lkthomas> how could I judge if I need to buy more servers in this case ?
[03:23] <PryMar56> lkthomas, do you count kdmflush ? ignore those processes
[04:04] <lkthomas> PryMar56: so how could I count it correctly ? :P
[06:25] <zanzacar> I think ubuntu-desktop got stuck installing half way through and I am not sure how to continue. Does anyone have any thoughts?
[06:48] <zanzacar> what is sane.d used for?
[07:55] <SpamapS> smoser: https://bugs.launchpad.net/ubuntu/+source/iputils/+bug/1313550 <-- when you wake up. ;)
[07:55] <uvirtbot> Launchpad bug 1313550 in iputils "ping does not work as a normal user on trusty tarball cloud images." [Undecided,New]
[07:56] <SpamapS> utlemming: https://bugs.launchpad.net/ubuntu/+source/iputils/+bug/1313550 for you too
[08:10] <lordievader> Good morning.
[08:12] <smb> jamespage, I know smoser has been actioned with this. But I am wondering whether this needs a bit more attention, given that its a regression in Precise and even worse one that causes issues with running clusters. (bug 1185756)
[08:12] <uvirtbot> Launchpad bug 1185756 in drbd8 "drbd8-utils not compatible with linux-lts-raring kernel in 12.04" [High,Fix released] https://launchpad.net/bugs/1185756
[08:12] <jamespage> smb, morning
[08:13] <jamespage> SpamapS, ah - that sounds familiar - we had a similar issue with the iso
[08:13] <jamespage> SpamapS, smoser: the tarball needs to be created with an extra flag to maintain extended attributes
[08:16] <smb> jamespage, Good morning. Hm, looks like I should at least set back the status to in progress or triaged. Unfortunately the regression was reported in the backport bug and I could not get anyone to open a separate bug.
[08:17] <SpamapS> jamespage: I tried to coerce tar to do it but tar seems to do basically nothing iwth --xattrs :-/
[09:49] <caribou> I need help in fixing bug 1313602
[09:49] <uvirtbot> Launchpad bug 1313602 in nova-cloud-controller "Nova-cloud-controller charms failed to sync ssh keys between compute nodes" [Undecided,Confirmed] https://launchpad.net/bugs/1313602
[09:49] <caribou> more precisely on how the authorized_keys / known_hosts are exchanged between the nova-cloud-controller & nova-compute units
[09:50] <caribou> I'm having problems in understanding the requirements wrt those two files
[10:01] <mojibake8> im trying to setup 2 servers in virtual box, 1 a web server and the other a database server
[10:02] <mojibake8> how should i configure the network adatpters for each?
[10:02] <mojibake8> such that they have static ips
[10:02] <vedic> I have installed ubuntu server 12.04 . I had to try a GUI app therefore I have installed xfce4. But Ubuntu is starting in text mode only. I have to type startx manually to run gui. How to change this bahevior?
[10:04] <mojibake8> vedic: i did a search, http://ubuntuforums.org/showthread.php?t=2003543
[10:05] <mojibake8> 2nd post says: install display manager or click the link in the 2nd post which asks you to modify bash_profile
[10:05] <vedic> mojibake8: I have updated runlevel to 5 already but still it starts in terminal mode
[10:14] <vedic> mojibake8: That doesn't help
[10:15] <vedic> I have added "[[ -z $DISPLAY && $XDG_VTNR -eq 1 ]] && exec startx"  to .bashrc file but doesn't work
[10:16] <vedic> Anybody? Why doesn
[10:16] <vedic> Why doesn't gui starts despite xfce4 installed and runlevel set to 5
[10:17] <mojibake8> if im not wrong bashrc is used on boot
[10:18] <jamespage> caribou, what do you need to know?
[10:19] <caribou> jamespage: I'm trying to figure out what needs to be passed in authorized_keys/known_hosts between the controller and each compute node
[10:19] <jamespage> caribou, ah - this is the #ETOMANYNODES error right
[10:19] <vedic> mojibake8: Trying this: http://ubuntuforums.org/showthread.php?t=1385698
[10:20] <caribou> jamespage: the one where relation-set(known-hosts...) explodes with "argument list too long" on the O/S side
[10:20] <jamespage> caribou, the base64 encoded parameter gets to be to long
[10:20] <jamespage> caribou, indeed
[10:20] <caribou> jamespage: yep, that one
[10:20] <caribou> jamespage: so right now, the whole file gets encoded in base64.
[10:20] <jamespage> caribou,you need to fragment the keys and hosts on the relation from the nova-cloud-controller side, and re-assemble them on the remote side
[10:20] <jamespage> nova-compute in this case
[10:21] <caribou> jamespage: so one relation-set/get for each line in each file ?
[10:21] <jamespage> caribou, that would make sense
[10:21] <caribou> jamespage: ah, that helps; I was not sure if the whole file needed to be passed
[10:21] <jamespage> caribou, well it does right now because that's how its been coded
[10:22] <jamespage> you would need to implement code on both sides of that relation
[10:22] <caribou> jamespage: this is the additional info I'm after : do we need to pass the complete files or only part of each one is suficient
[10:22] <jamespage> caribou, that's not really the right question
[10:23] <jamespage> caribou, you need to assemble a complete file on the compute nodes - how its gets passed is up to the charms
[10:23] <caribou> jamespage: ok, let me try a few things and I'll show you what I get if that's ok with you
[10:26] <vedic> mojibake8: I link I shared has worked
[10:27] <vedic> should I install lightdm or gdm?
[10:32] <jamespage> smb, still reading that bug btw
[10:34] <smb> jamespage, Unfortunately a mix between requesting a backport of drbd to handle newer kernels and that backport failing. I think my last debdiff is working good enough (hopefully)
[10:34] <vedic> I have installed lightdm after installing xfce4 on Ubuntu server. But now the screen doesn't accepts my login password
[10:35] <vedic> Its simply appears like it has accepted the password but it comes back within 1 or 2 seconds
[11:14] <Patrickdk> vedic, so? did you try #ubuntu
[11:14] <vedic> Patrickdk: I have removed lighdm and just added shell script in .bashrc to start startxcfe4
[11:14] <vedic> Its works now but without login manager
[11:15] <Patrickdk> vedic, it's out of scope, that is gui stuff, not -server stuff
[11:15] <vedic> Patrickdk: yea, I guess you are right. I should have posted this in #ubuntu
[11:15] <vedic> Thanks for point this
[11:15] <cwillu_at_work> vedic, you'll probably want to investigate /etc/x11/xsession.d/
[11:16] <vedic> cwillu_at_work: I see
[11:54] <stetho> Hi all. Does anyone know if there's a way that during a PXE install (or other method) using a KickStart file I can specify the install of a different kernel (3.2.0-56-generic in this case) and make it the default for grub to boot from? I'm currently scripting an apt-get install of the kernel and then rewriting the grub config with cat and I think I'm just asking for trouble doing this BUT I can't see a better way. Any suggestions?
[12:10] <rbasak> hallyn: around? I'm seeing a gettimeofday syscall fail after a clone() call with CLONE_NEWIPC | CLONE_NEWNET. Do you have any tips?
[12:10] <hallyn> rbasak: fail how?
[12:10] <rbasak> I've also noticed that it's using a pointer to a structure that existed before the clone. I'm wondering if the cloned process considers that invalid or something.
[12:10] <rbasak> hallyn: SIGSYS
[12:11] <rbasak> hallyn: https://bugs.launchpad.net/ubuntu/+source/vsftpd/+bug/1219857
[12:11] <uvirtbot> Launchpad bug 1219857 in vsftpd "vsftpd connections fail on amd64: "500 OOPS: child died"" [High,Confirmed]
[12:11] <rbasak> hallyn: [ 2967.985773] type=1326 audit(1398685961.714:32): auid=1000 uid=65534 gid=65534 ses=2 pid=1713 comm="vsftpd" sig=31 syscall=96 compat=0 ip=0x7ffff7ffadbc code=0x0
[12:12] <rbasak> I have it reproducing under gdb. Though for some reason I couldn't get gdb to stop at the SIGSYS itself - I had to reproduce down the stack to find the call.
[12:12] <hallyn> have you reproduced with a handwritten testcase?
[12:12] <rbasak> No. I wondered if I was missing something obvious.
[12:12] <hallyn> short answer is no, clone shouldn't be affecting validity of memory at all
[12:13] <hallyn> oh, you're using seccomp
[12:13] <hallyn> why not just add gettimeofday to the seccomp whitelist?
[12:13] <rbasak> So it is something obvious ;)
[12:14] <rbasak> I'll track that down - thanks!
[12:14] <hallyn> i'm trying to remember whether the seccomp filter only takes effect after a fork or not
[12:14] <hallyn> s/fork/clone
[12:15] <hallyn> ok - thx, i'm in an early-morning-without-coffee-after-weeklong-break funk :)
[12:21] <pmatulis> stetho: you want the installer to run a different kernel or the target system to have a different kernel?
[12:24] <m_tadeu> hi...crontab defines SHELL=/bin/sh....that seems to pose a problem on rinnung bash scripts, am I correct?
[12:24] <m_tadeu> s/rinnung/running
[12:25] <ogra_> if you have bash scripts you should explicitly use "#! /bin/bash"
[12:25] <ogra_> always ...
[12:25] <ogra_> (or just make your scripts POSIX compliant without any bashisms ... which is the better approach anyway)
[12:28] <m_tadeu> ogra_: In deed I'm using #!/bin/bash...but I'm getting the same errors as if I  run "sh myscript"....so, just confirming this
[12:28] <ogra_> $SHELL shouldnt override the hashbang line
[12:30] <m_tadeu> well it seems to be doing that, somethow
[12:31] <m_tadeu> I'll change the script to sh, instead of bash
[12:33] <coreycb> hallyn, rbasak:  seccomp takes effect immediately but I think there might also be an option to have it take effect after the next exec
[12:34] <rostam> HI in our lab I have a server & client systems. Our client system  have a disk with two partitions, sda1 & sda2. I have installed ubuntu 12.04 LTS on sda1.  The client system boots from sda1. I like to know if I can install another instance of ubuntu on sda2. Using a kickstart file and whatever is needed on the server system?  Thx
[12:41] <newbie123> h
[12:41] <hallyn> coreycb: near as i can tell it should take effect as soon as seccmp_load() is called
[12:41] <hallyn> i don't see an option for deferring.  but presumably vsftpd does that right before the clone :)
[12:46] <coreycb> hallyn, yeah I'm not seeing anything for deferring.  it might have been an option in earlier patches that was nacked.
[12:58] <stetho> pmatulis: Sorry, went away from my desk. I want the target system to have a different kernel.
[12:59] <pmatulis> stetho: a more recent kernel?
[12:59] <stetho> No, an older kernel - 3.2.0-56-generic to be precise.
[13:01] <pmatulis> stetho: just install the old package you want, and remove the recent one, with late_command .  shouldn't that work?
[13:01] <zul> hallyn:  i was going to do libvirt 1.2.3 but im going to wait until 1.2.4 is out
[13:01] <xutl> how to run ubuntu server live as like ubuntu desktop from dvd/usb ?
[13:02] <adac> Guys, I'm trying to install supervisor via docker on precise, but since about 2 or 3 days I get the following error: https://gist.github.com/anonymous/11371085
[13:02] <adac> did something change on distro side?
[13:02] <xutl> !live
[13:02] <ubottu> The Ubuntu Desktop CD is a "LiveCD" which can be run without altering existing files on your harddrive. Especially useful for testing your hardware's compatibility, it also includes an install option.
[13:02] <stetho> pmatulis: It doesn't seem to. I've tried it and get the grub> prompt when the system reboot. Obviously I've taken this to mean it doesn't work rather than I'm doing something wrong.
[13:03] <stetho> :-)
[13:04] <pmatulis> stetho: pastebin the late_command
[13:06] <hallyn> zul: ok
[13:09] <stetho> pmatulis: I don't have it to hand anymore, it was something along the lines of d-i preseed/late_command string in-target apt-get remove -y linux-image-3.2.0-60-generic
=== sz0 is now known as sz0`
=== sz0` is now known as sz0
=== sz0 is now known as sz0`
=== Ursinha-afk is now known as Ursinha
=== sz0` is now known as sz0
[13:25] <m_tadeu> can anyone point me to a good tutorial on how to use dash arrays? google is not helping that much
=== sz0 is now known as sz0`
=== sz0` is now known as sz0
[13:25] <zul> jamespage/coreycb: ok juno is open
=== sz0 is now known as sz0`
=== sz0` is now known as sz0
[13:26] <coreycb> zul, thanks
[13:29] <pmatulis> stetho: dunno then, i haven't tried it myself
[13:44] <rbasak> utlemming: hey, any eta on utopic cloud images, please?
[13:44] <rbasak> (I mean dailies, of course)
=== nszceta_ is now known as nszceta
[14:15] <stetho> Another question about PXE:  I've got a number of servers with a single disk for the OS and two disks in RAID 1 for data. I'm trying to prepare the disks as part of a PXE boot - /dev/sda works but Ubuntu give the RAID a name like isw_chhgbfgage_RAID1 where chhgbfgage seems to be a completely random string so I can't refer to it as /dev/mapper/isw_chhgbfgage_RAID1 literally. Before I write a script that gets the contents of /dev/mapper and figures
[14:15] <stetho>  out the array name, is there a "better" way to do it?
[14:18] <hadifarn_> what was that package for Apache that optimize images? it was made by google
[14:19] <stetho> modpagespeed
[14:20] <patdk-wk> I don't know why anyone would use that for images
[14:20] <patdk-wk> it's so much easier to just optimize the images, one time
[14:20] <patdk-wk> than on every load
[14:21] <hadifarn_> cheers stetho
[14:21] <hadifarn_> apt-get package name is different?
[14:26] <stetho> I don't think there's a package for it. There's a .deb for it on Google code though. Download it and dpkg-i
[14:27] <stetho> http://developers.google.com/speed/pagespeed/ apparently
[14:43] <cwillu_at_work> hmm, how do you go about getting a recent debootstrap on lucid?
=== Lcawte|Away is now known as Lcawte
[14:59] <axisys> how do I list the linux images that are lower than current kernel version here? http://dpaste.com/1801794/
=== Ursinha is now known as Ursinha-afk
[15:07] <cwillu_at_work> ah, you can run it from the source tree
[15:19] <smoser> SpamapS, i really would have appreciated you opening that bug say 3 weeks ago
[15:19] <smoser> ;)
[15:21] <rostam> HI in our lab I have a server & client systems. Our client system  have a disk with two partitions, sda1 & sda2. I have installed ubuntu 12.04 LTS on sda1.  The client system boots from sda1. I like to know if I can install another instance of ubuntu on sda2. Using a kickstart file and whatever is needed on the server system?  Thx
[15:24] <xtbman> Hey, I installed ubuntu server, and my network adapter doesn't show an IP address, and pinging my router returns an unknown host error. Is there a guide for configuring the network?
[15:25] <smoser> xtbman, man interfaces
[15:25] <xtbman> smoser: thanks I'll give it a shot.
[15:37] <smoser> hallyn, stgraber your thoughts on https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/1313550/comments/5 would be appreciated.
[15:38] <uvirtbot> Launchpad bug 1313550 in maas "ping does not work as a normal user on trusty tarball cloud images." [High,Confirmed]
[15:38] <smoser> especially wrt backwards compat (ie, getting that functional on precise)
[15:39] <tom[]> is this the right place to get php 5.4 on 12.04lts? https://launchpad.net/~ondrej/+archive/php5-oldstable
[15:50] <patdk-wk> tom, how do you mean, right place? there is no *supported* php 5.4 on 12.04
[15:56] <hallyn> smoser: not sure what you're asking.  You just want the ubuntu-cloud lxc template to re-set the xattr?
[15:56] <hallyn> oh, i see
[15:57] <hallyn> i think the bug should be marked as affecting tar as well, as a query about backporting the xattr support...
[15:57] <hallyn> though i assume it's a feature and so not possible
[15:57] <tom[]> patdk-wk: i understand about canonical not including 5.4 in 12.04's packages. what i mean by "right place" is, do y'all think this is a good choice? is there ant consensus on using Ondřej Surý's ppa when you need 5.4 on a 12.04?
=== Ursinha-afk is now known as Ursinha
[15:59] <smoser> hallyn, i'd consider it SRU'able as a necessary function for using newer things.
[16:00] <hallyn> so you agree with marking that bug as affecting tar, fix released in trusty, confirmed on prcise?
[16:00] <smoser> sure.
[16:00] <tom[]> s/ant/any/
[16:00] <smoser> hallyn, do you have any cycles to look at what that patch would be ?
[16:01] <hallyn> and if you like i can look this afternoon at what it'll actually take to do it, no idea about the
[16:01] <hallyn> yup
[16:01] <patdk-wk> tom, that doesn't include *many* php modules, so the question is, does that even have enough php for you
[16:01] <smoser> thanks.
[16:01] <hallyn> pls assign the precise tar task to me
[16:01] <patdk-wk> and then, do you think he will maintain security updates for you
[16:01] <hallyn> (or i'll do it in a few mins)
[16:02] <patdk-wk> personally, I wouldn't even think about pushing 5.4 onto 12.04, with it being in 14.04 currently
[16:02] <patdk-wk> you just upgrade to 14.04 lts, and get security updates for 5years on it
[16:02] <tom[]> patdk-wk: you already answered my next question
[16:03] <tom[]> i was nervous about 14.04's newness and its hype about openstack
[16:03] <patdk-wk> I know atleast in my ppa, I sync security updates as much as possible
[16:03] <patdk-wk> I haven't had any real issues with 14.04
[16:03] <patdk-wk> just a few minor things, that aren't production impacting
[16:04] <tom[]> but i guess i should be less nervous of that than of unsupported builds
[16:04] <patdk-wk> well, there are like 2 security issues a year, generally with php
[16:04] <patdk-wk> enough to be concered about
[16:05] <tom[]> the seasoned php user is almost as scared of php security as she is of linux
[16:05] <tom[]> ;)
[16:12] <xtbman> I figured out my networking issue, thanks.
[16:16] <stgraber> smoser: I'm off today, I'll try and set aside some time to think about it tomorrow. I'm pretty surprised I didn't notice this before, I guess I'
[16:16] <stgraber> I'm mostly running older releases in containers using lxc-download...
[16:17] <stgraber> if we only support 14.04 on 12.04 and won't support 14.10 on 12.04, we may consider just setting the setuid flag again in a SRU which would be a quick and dirty fix for this
[16:18] <stgraber> and likely much more easier than backporting the tar change (I also have the concern that busybox's tar likely doesn't support those flags so LXC will need to do some detection...)
[16:20] <smoser> lxc download would also be broken, yeah. i affected lxc as the cloud template.
[16:27] <Havenstance> okay so my drive is full, how can i figure out what files made it fill up?
[16:27] <patdk-wk> dunno
[16:28] <patdk-wk> but you can start with finding out what folders have the most space and/or inodes
[16:28] <patdk-wk> exactly what is it full of? data or files?
[16:28] <Havenstance> I think its full of old wireshark caputures
[16:28] <Havenstance> but I don't know where to look for the previous saves?
[16:30] <jak3000> hi all, i configured my vsftpd service on ubuntu-server, when i try acces i get a error: ftp: connect :Connection refused  why? how to solve the port21 is opened. and the service is running
[16:33] <rbasak> jak3000: which release?
[16:33] <Havenstance> patdk-wk, I found it thanks
[16:34] <jak3000> rbasak how to know wich is the release?
[16:34] <rbasak> jak3000: lsb_release -a
[16:34] <rbasak> jak3000: also, which version? "dpkg-query -W vsftpd"
[16:35] <jak3000> 12.04.3
[16:35] <rbasak> Ah, not sure then, sorry. I asked because I'm working on a vsftpd bug right now, but it only affects >= 13.10.
[17:00] <toyotapie> I need to write a bash script that can take the top of a page from two different tiff files and create a new tiff file with a single page that is half one page and half the other.
[17:00] <toyotapie> is tiffcrop the right tool ?
[17:06] <SpamapS> toyotapie: tried imagemagick?
[17:07] <toyotapie> I am reading through the man page of imagemagick right now.
[17:08] <xtbman> My server is working, such that other computers on my network can access it. I want non-local computers to access the home page, too. Is there a guide that explains how to set this up?
[17:09] <xtbman> Right now, if I type 10.0.0.4 into a network computer's browser, it goes to my apache page on the server. So, somehow I have to set up an IP for outside computers to connect to.
[17:09] <toyotapie> I think it's the append flag I need to use with convert :)
[17:10] <sarnold> xtbman: simple mechanism is to port forward port 80 from your router to your 10.0.0.4:80
[17:10] <sarnold> xtbman: http://portforward.com/english/routers/port_forwarding/  :)
[17:10] <xtbman> sarnold thanks
[17:10] <Havenstance> can someone recommend a good packet capturing program? Something similar to wireshark but one that would save the files in alot smaller file
[17:11] <sarnold> Havenstance: compress the files with gzip?
[17:11] <Havenstance> Except wireshark doesn't have the features I want anyway
[17:12] <Havenstance> I want to be able to see how much data a particular IP has used over say a 30 day window
[17:12] <Havenstance> then I'd open wireshark and capture the packets coming from that machine.
[17:12] <Havenstance> So maybe I'm looking for something to use in conjunction with wireshark
[17:13] <sarnold> Havenstance: ah you want something completely different then :)
[17:13] <Havenstance> well, just something that would capture the packets but rather log the mbps usage?
[17:13] <sarnold> Havenstance: check out 'ntop' and 'darkstat'. I've used ntop before and didn't really love it, but it did okay..
[17:14] <Havenstance> perfect, i'll run a google search thanks sarnold
[17:14] <sarnold> Havenstance: just apt-cache show ntop and apt-cache show darkstat -- see if you like the sound of them :) hehe
[17:14] <Havenstance> thanks
[17:15] <jak3000> how to open permanently the port 3306 ?
[17:16] <jonascj> Hi all. I run an sftp server (using openssh) ans users are chrooted to /srv/sftp/{username}. What is the prettier solution for this setup regarding the users home directory and the creation of /srv/sftp/{username} (these users only have sftp access, no shell access): 1) to specify /srv/sftp as base dir during user creation and specify an empty skeleton dir (e.g. no files copied to the new home dir), or 2) manually create /srv/sft/{usernam
[17:18] <sarnold> jonascj: check out adduser(8) -- the bit about /usr/local/sbin/adduser.local looks useful to you :)
[17:19] <jonascj> The end result should be the an empty /srv/sftp/{username} directory. What the users homedir (in /etc/passwd) is doesn't really matter to me. In openssh matching I can chroot to /srv/sftp/%u or %h. I just have to choose the optoin based on whether I set the users homedir to /srv/sftp/{username} or if I set it to something else
[17:19] <jonascj> sarnold: thanks, but adduser isn't that great for commandline adding users (from a script), is it?
[17:20] <jonascj> or do adduser function just like useradd if I want to? normally one just does "adduser joe" and then you are asked a series of questions (password, room number etc.)
[17:20] <zul> jamespage:  looks like git is on dep wait for utopic
[17:21] <sarnold> jonascj: you can provide e.g. --gecos command line options to specify many things.. I'm not sure if it'll skip prompting entirely or not though...
[17:23] <jonascj> sarnold: then I think I'll have to go with the useradd command instead of adduser :)
[17:23] <jonascj> since it is automatic creation of triggered by something
[17:24] <sarnold> jonascj: makes sense :) if you're already scripting it up, a bit more to create new directories isn't the worst thing :)
[17:25] <jonascj> it looks like --gecos disables any interactive part of adduser, but then why user adduser vs. useradd?
[17:25] <sarnold> jonascj: for the automatic and cheap /usr/local/sbin/adduser.local execution :)
=== lazypower-travel is now known as lazyPower
[17:26] <jonascj> sarnold: oh, and adduser.local holds commands which are run every time a user is added?
[17:27] <sarnold> jonascj: yeah
[17:27] <jonascj> sarnold: okay, point taken. that could be useful in some situations.
[17:28] <sarnold> jonascj: yeah, more useful if you're doing the creating by hand. if you're already scrpiting it up, not as magical :)
[17:28] <SpamapS> smoser: --include-xattrs=* ... don't you just love GNU people? "What? You turned on a feature, AND you want it to actually default to being .. on.. when you turn it.. on? CRAZY"
[17:29] <jonascj> but really I was just more interrested in the "sematics" of the situation. If the users homedirectory have no special meaning (e.g. it holds no profile informaiton like .profile or .bashrc etc.) should I just set it to /tmp to emphasize that it holds no persistent files.
[17:29] <smoser> SpamapS, yeah, i figured that part out.
[17:30] <sarnold> jonascj: I'd make sure that only the user in question owns the directory and has write access to the directory.
[17:30] <smoser> the fun part is going to be getting that fixed in 12.04 consumption. (ie, 12.04 systems using 14.04 roots)
[17:30] <sarnold> jonascj: that assumption is baked into thousands of programs, it'd be best to not screw that one up :) hehe
[17:32] <jonascj> sarnold: Then I might as well use the -m option of useradd (to have the homedir created) and "-b /srv/sftp". How then should I ensure that the newly created dir is empty? Specify my own skel-directory ? or empty out the /etc/skel dir?
[17:32] <sarnold> jonascj: I'd probably go with the own skel directory, that feels more flexible for long-term modification
[17:35] <xtbman> sarnold it worked! :)
[17:35] <sarnold> xtbman: nice :)
[17:36] <jonascj> sarnold: thanks - those a great suggestions :9
[17:36] <jonascj> * :)
[17:36] <sarnold> jonascj: have fun! :)
[18:02] <jonascj> hmm, the folder to which sftp users are chrooted needs to be owned by root
[18:02] <jonascj> so if I decide to chroot them to their home (/sftp/srv/{username}) then their home needs to be owned by root...
[18:03] <jonascj> I do not know if I can put additional ACL on top of that to give the users permission again.
[18:04] <jonascj> I know I need to get ACL up and running at some point because normal sftp users should be chrooted to /srv/sftp/{username} but sftp admins should be chrooted to /srv/sftp and have read and wirte access to the /srv/sftp/{username} folders...
[18:15] <RoyK> jodh: seems strange. we're using sftp with rssh at work, and users' homedirs are definetely owned by the users' themselves
[18:16] <patdk-wk> yep
[18:17] <patdk-wk> if you want to use openssh chroot, you have to chroot to a like /home
[18:17] <patdk-wk> you could replace ssh with like, proftpd and use the sftp module
[18:18] <patdk-wk> you could use a custom sftp program
[18:18] <patdk-wk> you could go apparmor crazy :)
[20:20] <qhartman> anyone know how to force the order of resolv.conf entries via modification of the /etc/network/interfaces file?
[20:21] <qhartman> I'm looking for something like modifying the dhclient.conf config and adding a "prepend domain-name-servers", but I don't know how to change that file with the curtin installer.
[20:39] <SpamapS> smoser: so should we expect tarballs w/ xattrs soon on cloud-images.ubuntu.com ?
[20:40] <SpamapS> utlemming: ^ ?
[20:40] <utlemming> SpamapS: shortly. I just merged the chnage in
[20:40] <SpamapS> utlemming: w00t
[20:41] <smoser> again, SpamapS it' would have been nicer if you did this 3 weeks ago!
[20:41] <smoser> (but thanks for raising it now)
[20:42] <SpamapS> smoser: I agree. I did actually test trusty. Just not with nagios.
[20:42] <SpamapS> Or the other 3 binaries that use the feature. :-P
[20:47] <smoser> SpamapS, what are the other 3 binaries ?
[20:48] <SpamapS> smoser: arping, gnome-keyring-daemon ... something else
[20:51] <fellayaboy> hey hows it going guys
[20:51] <fellayaboy> i have a question.  when i boot into pxe server. i can see my entries but when i select them they do absolutely nothing...it wont run the operating system
=== mjohnson151 is now known as mjohnson15
[20:52] <fellayaboy> im using tftp-hpa & nfs
[20:55] <fellayaboy> where can i see logs about my pxe server
[21:14] <m_tadeu> I'm stuck in initfstab...can anyone tell me how to edit fstab from there?
[21:17] <sarnold> m_tadeu: "initfstab"? I'm not sure I've heard of it..
[21:17] <m_tadeu> sarnold: sorry, long day...it's initramfs....
[21:20] <sarnold> m_tadeu: ah, okay; so, you've got a shell?
[21:20] <sarnold> m_tadeu: does cat /etc/fstab show you your fstab or the one from the initramfs environment?
[21:22] <m_tadeu> sarnold: I'm afraid /etc/fstab it doesn't exist at this point
[21:23] <sarnold> m_tadeu: okay; cat /proc/mounts -- do you see your root filesystem mounted anywhere else at the moment?
[21:24] <m_tadeu> sarnold: yes...rootfs is mounted in /
[21:24] <sarnold> m_tadeu: oh, interesting. I expected /etc/fstab to your fstab in that case.
[21:25] <m_tadeu> sarnold: the real root should be a LVM volume and the device is /dev/Root
[21:26] <m_tadeu> I changed the volume name, but probable it didn't change fstab
[21:27] <sarnold> m_tadeu: hrm, I haven't done LVM before :/ no idea there
[21:29] <m_tadeu> sarnold: first time here too
=== markthomas_ is now known as markthomas
=== markthomas_ is now known as markthomas
[23:03] <mgw> We have these lines in the upstart script for salt-minion:
[23:03] <mgw>   # invoke salt-minion via su so that /etc/environment is read
[23:03] <mgw>   exec su -c salt-minion
[23:03] <mgw> it doesn't seem to work on Trusty though
[23:04] <mgw> did something change?
=== sz0_ is now known as sz0`
=== sz0` is now known as sz0_
=== Lcawte is now known as Lcawte|Away
[23:59] <keithzg> Hmm. Ran into a bug where logrotate is complaining about mysql, turns out it's a known issue when you've stopped your mysql service . . . and has been around, unfixed, since 2009. Yeesh. https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=544013
[23:59] <uvirtbot> Debian bug 544013 in mysql-server "mysql-server-5.1: logrotate script cannot handle stopped mysqld" [Normal,Open]
[23:59] <keithzg> Wonder if MariaDB solves that :P