=== meritorious is now known as StrangeCharm
tarvidpuzzling over dhcpd01:02
tarvidthe version I have is isc-dhcpd-4.1.1-P101:02
tarvidthere is a howto for dhcpd3 - https://help.ubuntu.com/community/dhcp3-server01:03
tarviddo I just pretend it is dhcpd3?01:04
=== jjohansen is now known as jj-afk
twbOK, when I want something to permanently appear on shell logins (i.e. /etc/motd), and not be deleted by any fancy-pants auto-rewriting motd junk, where do I put it?01:37
qman__that used to be the case01:43
qman__but recently an update killed my motd on all my lucid servers01:43
twbFor now I have manually put it in both motd and motd.tail, since apparently there's no /usr/sbin/update-motd and simply logging in again doesn't trigger the pam-based update-motd gank01:45
twb(The purpose of the message is, perhaps ironically, to remind sysadmins that this machine doesn't deal with rebooting properly.)01:45
qman__mine broke a few weeks ago but I haven't had time to look into it01:46
qman__my systems barely qualify as multiuser though, so it's not a big deal, just annoying01:47
twbIt's the library server at a prison :-(01:49
uvirtbot`New bug: #812642 in mysql-5.1 (main) "package mysql-server-5.1 5.1.54-1ubuntu4 failed to install/upgrade: el subproceso instalado el script post-installation devolvió el código de salida de error 1" [Undecided,New] https://launchpad.net/bugs/81264201:51
uvirtbot`New bug: #812646 in cloud-init (main) "Race in DataSourceNoCloudNet with kvm" [Undecided,New] https://launchpad.net/bugs/81264602:11
chrismsnzHi, we're running Ubuntu 10.04 LTS on some supermicro servers, and we're having some stability problems when the servers come under load02:15
chrismsnzif the problem is triggered (sometimes due to high memory usage) the system/kernel appears to go into some sort of zombie mode where it's still accepting network connections but not responding to anything02:16
chrismsnzleaving them to time02:17
chrismsnzIf I were to investigate this problem further, what is the best way to start collecting data on this failure to provide the maximum information for filing a bug02:17
=== zz_ng_ is now known as ng_
=== ng_ is now known as zz_ng_
qman__it may not be a bug at all, if the servers are simply running out of memory02:39
qman__it would make perfect sense that the kernel could accept connections, but then have the application unable to do anything02:39
chrismsnzin most cases I've seen the OOM killer has been activated02:40
twbchrismsnz: responding to ping but not doing anything else is normal behaviour when its resources are exhausted02:40
qman__setting proper limits would solve the problem02:40
twbchrismsnz: particularly if it's e.g. thrashing swap instead of simply OOM-killing02:40
twbqman__: unfortunately it's... nontrivial to set per-proc RAM limits02:40
qman__or finding out why you're running out of memory02:40
chrismsnzI will investigate02:42
lifelessulimit is goo02:42
lifelessalso consider making sure your swap is only large enough to hold a few seconds IO02:42
chrismsnzSwap is configured to be 50% of RAM02:43
lifeless(that is, big enough for truely unused pages to page out, but small enough you can't get completely wedged)02:43
chrismsnz8gb RAM 4 GB swap02:43
lifelesschrismsnz: thats going to -hurt-02:43
lifelessunless you've got a 1GB San or something02:43
chrismsnzother servers are configured as 4gb RAM 2 gb swap02:44
qman__it doesn't scale like that02:44
lifelessso, ask yourself what you want the swap to hold02:44
qman__said servers have the same disk bandwidth02:44
qman__the same disk seek times02:44
twbFWIW I don't use swap at all02:44
qman__so they should have the same swap02:44
qman__also, that02:44
twbUnder 2.6, default swap behaviour is so shit that it's not helpful02:45
qman__unless the server is ancient and can't be upgraded, I don't use swap anymore02:45
chrismsnzI think that revisiting the swap configuration is probably a good idea02:45
qman__and I don't make any swaps bigger than 1GB02:46
chrismsnzthe swap usually sits with 500mb of pages in it with the memory not even close to being full (minus buffers/cache)02:46
chrismsnzthat's probably a good place to start02:46
chrismsnzI also recall configuring our high memory servers (36gb +) to be very stingy with swap via the swappiness tweakable in /proc02:47
twbWow, swap is actually used in normal operation?02:47
twb(...of that box)02:47
chrismsnztwb: I believe rarely used pages are swapped out as a matter of process02:47
chrismsnzand the freed ram used to power io cache and buffers02:48
twbchrismsnz: only if vm.swappiness is high or whatever02:48
chrismsnzit's 60/100 by default on ubuntu02:48
chrismsnzI have zero'd it before given certain workloads and high memory systems02:48
chrismsnzperhaps I'm placing too much trust in the kernel to swap out unused pages02:49
chrismsnzand should lower this value02:49
chrismsnzhere is the current memory usage of an identically configured server02:49
chrismsnz             total       used       free     shared    buffers     cached02:49
chrismsnzMem:          7994       7744        250          0       1019       349002:49
chrismsnz-/+ buffers/cache:       3234       476002:49
chrismsnzSwap:         4767       1238       352902:49
chrismsnzas you can see... swap is being used even though there's 4-5gb of free memory02:50
twbI dunno man02:50
TheEvilPhoenixi think you shouldnt be pasting into here02:51
chrismsnzthanks for your help i'll bbl02:51
qman__my file server still has a swap because I set it up long ago02:51
qman__said swap is 0 used02:51
qman__current uptime is 21 days, but it's been up much longer than that in the past02:51
qman__granted it's using ~1.5GB instead of ~4GB (out of 8GB)02:53
qman__but I've not seen it swap since back when it had 2GB in it02:53
qman__IMO, if a server spends any measurable amount of time swapping under normal load, something is wrong02:56
q_a_z_steveHow do I go about setting up a linux server with the ability of providing a network boot option to clients on the LAN?03:03
lifelessyou need a tftp server for that03:07
Tommy_nmwlifeless: what?03:07
lifelessTommy_nmw: I was answering someone else03:08
Tommy_nmwlifeless: sorry bro03:08
=== vDubG__Gone is now known as vDubG
UrsinhaDaviey: does this help you with the changelog parsing thing you mentioned me at the Rally? http://rsalveti.wordpress.com/2011/07/19/launchpadlib-creating-a-changelog-for-a-ppa/03:37
chrismsnztwb and qman__: what does 'cat /proc/sys/vm/swappiness' say on those systems if you don't mind my asking?03:43
chrismsnzI just went to check out the crashed server - there was some sort of kernel debug output on the screen, it was too verbose for me to read and the system was unresponsive so I could not properly read it03:44
twb60 on the one I pasted03:46
=== vDubG is now known as vDubG__Gone
chrismsnztwb: thanks. Time to start auditing my applications03:47
chrismsnzhaha, guess he shouldn't have been pasting03:48
aarcaneso I'm running maverick.  I need to update samba to the version in natty.  I added the sources deb-src to my sources.list.d in a .list file, but they're not showing up after an update.  is there something else I need to do to be able to build the package I need ?04:45
lifelessSpamapS: oh hai05:17
uvirtbot`Launchpad bug 812691 in lxc "failure building lucid container" [Undecided,New]05:17
maxagazhow to get the MAC of a remote machine ?06:45
trapmaxeg. arp -a <ip>06:46
jmarsdenmaxagaz: ping it and check your arp table afterwards, or do something like ssh user@remotemachine.example.com /sbin/ifconfig06:50
twbjmarsden: itym "ip n"07:04
jmarsdentwb: For a machine on the same subnet, sure, you can do that, if you remember all the ip subcommands.  I remember arp, it has been around longer and is portable to windows and old BSD boxes etc etc, which ip n is not.07:07
DavieyGoooooood Morning!07:10
twbDaviey: I've yet to see any evidence07:12
Davieytwb: you'd be suprised!07:19
=== smb` is now known as smb
Ursinhabuenos dias, Daviey07:34
parapanhi fellows > need some help in relation with nomachine nx client/server operation ....is someone available ???07:35
=== zz_ng_ is now known as ng_
DavieyUrsinha: o/08:03
UrsinhaDaviey: o/08:03
jamespagemorning all08:04
Ursinhamorning jamespage08:06
jamespagehi Ursinha08:06
lynxmanmorning guys08:17
lynxmanUrsinha: what are you doing up?08:17
lynxmanDaviey: ping08:18
Ursinhabon dia lynxman :)08:20
Ursinhalynxman: I'm working :P08:20
lynxmanUrsinha: bom dia! :)08:20
lynxmanUrsinha: so soon? wow08:20
lynxmanand by soon I mean early08:20
Ursinhalynxman: actually it's not too early, but a tad late :P08:21
Ursinhayou can sleep when you're dead08:22
lynxmanUrsinha: I can tell you for sure, one leads to the other, somehow :D08:22
Davieylynxman: hola08:22
lynxmanDaviey: hola! o/08:23
lynxmanDaviey: just wondering if you had any time this week to have a look at the package? :D08:23
Davieylynxman: Can you remind me of which one?  I can't see it in the sponsorship queue.08:24
lynxmanDaviey: Sent you the dsc, wait a sec...08:24
lynxmanDaviey: https://launchpad.net/~lynxman/+archive/ppa/+files/mcollective_1.2.1-0ubuntu2.dsc08:24
lynxmanDaviey: just lots of fixes08:25
Davieyon it, thanks08:25
lynxmanDaviey: you're a star (*)08:25
Davieylynxman: bugs with attached branches or debdiffs really do make this easier.08:26
lynxmanDaviey: will have a debdiff next time :)08:26
Davieylynxman: looks fine to me, the only thing i am going to change is the Maintainer field to be Ubuntu Developers rather than MOTU.08:29
lynxmanDaviey: great :)08:29
lynxmanDaviey: as said, just small fixes to control and added debconf-po support08:29
lynxmanDaviey: also cleaned it in Lintian as much as I could08:29
Davieylynxman: nice job!08:30
lynxmanDaviey: thanks, trying hard :)08:31
Davieylynxman: uploaded08:33
lynxmanDaviey: thank you very very much \o/08:34
Davieylynxman: no, thanks for the patch!08:37
lynxmanDaviey: no no, thank you ;)08:39
Davieyjamespage: I'm right in saying there is no point in a bouncycastle merge?08:40
jamespageDaviey: not ATM08:40
jamespageI wanted to get the unit testing enablement back into debian - then we can just sync08:41
jamespagethe changes that have been made in Debian are the last set of Ubuntu changes - but not all of them08:41
Davieyjamespage: ah, dandy08:41
=== ng_ is now known as zz_ng_
shal3rHow to enable remote root login on lucid? I'm getting "Please login as the ubuntu user rather than root user"08:48
twbshal3r: root logins are allowed by default.08:49
twbshal3r: you simply can't using single-factor authentication, because root doesn't have a password.08:50
=== mendel__ is now known as mendel_
=== mcahornsirup_ is now known as mcahornsirup
shal3rtwb, i know it's not allowed by default. I'm looking for way to enable it because i need to add this server to my backup system08:57
twbUse multi-factor auth.08:59
Davieyshal3r: check /etc/ssh/sshd_config for PermitRootLogin.09:00
twbDaviey: wait, isn't it without-password by default?09:02
twbAh, as at lucid it is "PermitRootLogin yes" by default.09:03
Davieytwb: Yeah.. i'm assuming he's already set a passwd.09:03
twbSo he SHOULD NOT set a password, but should instead use multi-factor auth (passphraseful SSH keys), since password-based authentication sucks donkey balls.09:04
Daviey!root | shal3r09:05
ubottushal3r: Do not try to guess the root password, that is impossible. Instead, realise the truth... there is no root password. Then you will see that it is 'sudo' that grants you access and not the root password. Look at https://help.ubuntu.com/community/RootSudo09:05
Daviey(i hate factoids btw)09:05
twbDaviey: they exist to save me retyping the same thing each time :-P09:05
maswantwb: I'm not going to argue stronly for passphraseful ssh keys, since automation might want passphraseless keys. You just have to remember that security then is a strict tree, where anyone compromising the root will get control of all downstreams servers09:05
twbmaswan: granted, although a passphraseless key is back to single-factor auth, i.e. you lose09:06
Davieyless of a loss than passwd based auth IMO.09:07
twbNow, fun fact09:07
* Daviey moves to the edge of his seat.09:07
twbIf you have access to the private SSH key, you can brute force its passphrase as fast as you like, WITHOUT ever contacting any server that uses it in its authorized_keys file.09:07
maswantwb: If you are going to insist on two-factor auth, you should have two independent factors. Like ssh key to login and password to sudo, imho.09:08
Davieymaswan: tools like rsnapshot etc, i don't believe can use sudo.09:08
shal3ri'm using key file + IP filter for this09:08
maswanDaviey: Yeah, there are exceptions, especially when you need automated systems. But then the master server that has direct and automatic root access to all the clients should require strong auth (and not run any random vulnerable services either)09:10
twbDaviey: strictly, they can, but it's a fucking pain09:11
twbAnd probably even harder if their use of sudo wasn't NOPASSWD:09:11
Davieyinteresting.. i should look at doing that.. i've always just used root for that.09:13
* Daviey spies jamespage going on a Debian ITP spree.09:14
jamespagejust catching up with myself :-)09:14
twbDaviey: I use root with rrsync09:14
mendel_someone with OCR experience here?09:14
twb(Which is in /usr/share/doc/rsync/scripts/rrsync.gz)09:15
twbe.g. command="/usr/local/bin/rrsync /var/www/",no-agent-forwarding,no-port-forwarding ssh-dss ... foo@example.net09:15
twb(That's in ~root/.ssh/authorized_keys, you understand.)09:16
Davieymendel_: There are 3 main types of people here, server developers and sysadmins; and those that cannot make up their mind and call themselves devops.  The combined knowledge here is enough to take over the world. so try,09:16
ubottuPlease don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-)09:16
mendel_currently using tesseract-ocr, curious to learn about commercial alternatives or better ways to integrate tesseract..09:17
mendel_currently grayscaling + tiff the image09:17
* Daviey glazes over.09:17
twbmendel_: this is a bad place to ask about proprietary solutions.09:17
mendel_but as server guys some of you must have experience with it.. that's why I try09:18
twbtesseract probably has its own channel, too09:19
lynxmanDaviey: devops == can't make up their minds? hah :)09:33
Davieylynxman: :)09:38
lynxmanDaviey: I guess you're right... can't make my mind about it09:39
=== mcahornsirup_ is now known as mcahornsirup
* jamespage expects from kickback for filing that may ITPs for jenkins forks :-)10:05
lynxmanjamespage: wise to expect it indeed10:06
Davieyjamespage: isn't kickback a good thing?  Like when you get referal bonus?10:06
jamespagehowz that?10:07
* jamespage goes for coffee10:08
Davieyjamespage: I want a coffee.10:13
Davieyand a pony.10:13
* jamespage gets his thermos out 10:13
jamespagecould be a while tho - quite a few hours to your house :-)10:13
jamespagecan't help with the pony10:14
jamespagelynxman: might not be to bad - rather a heated debate going on about systemd on debian-devel ATM10:17
lynxmanjamespage: oh... systemd :)10:17
jamespageso my ITP's might not get that much attention10:17
lynxmanjamespage: I need to subscribe to debian-devel, now that I think of10:17
RoyKhm,.. how can I check when a prosess was started?10:18
MACscrI have an ubuntu server running in a xen guest. I have the guest setup to mount the drives as sda1 and sda2. This is setup in fstab and the guests cfg, yet when it boot, its showing xvda1, etc. Any ideas why this is happening? Are the newer kernels forcing this?10:22
lynxmanRoyK: if you run "ps aux" it has a "Start" column that gives you a rough idea of the process start time10:22
RoyKps -eo pid,lstart,cmd was a bit more precise :)10:23
lynxmanMACscr: Xen drives will show as xvdaN as far as I know10:23
lynxmanRoyK: showoff :P10:23
MACscrlynxman: why the hell is that being forced though. Thats pretty retarded =/10:23
lynxmanMACscr: well talk to the xen guys about it, it does make sense to me though10:24
DavieyMACscr: we are tracking the similar/same(?) issue with our ec2 images. bug #78493710:25
uvirtbot`Launchpad bug 784937 in linux "/mnt not mounted, swap not used, disk is xvde" [High,Confirmed] https://launchpad.net/bugs/78493710:25
MACscrDaviey: so what im seeing from that is that were shit out of luck and its an intended "feature"10:30
MACscrmeans i have to completely destroy all my backups and start from scratch10:30
MACscrwhat a load of bs10:30
uvirtbot`New bug: #812806 in clamav (main) "package clamav-base 0.97 dfsg-2ubuntu1 failed to install/upgrade: subprocesso script post-installation instalado retornou erro do status de saída 1" [Undecided,New] https://launchpad.net/bugs/81280610:31
DavieyMACscr: If you have to destroy your backups, you are probably doing it wrong.10:33
MACscrDaviey: its a commercial app, dont have a choice. They are crappy that way10:34
MACscrits r1soft10:34
MACscri dont think its an issue with their 3.0 product, but that has its own load of bugs, so im not upgrading, nor is a lot of people10:35
DavieyI seem to remember there was a feature of xen that allowed the format to be changed.10:35
DavieyBut i could be mistaken, remembering something else10:35
MACscrDaviey: it used to be an option, but i think these newer kernels are removing the option10:35
MACscrDaviey: just found this on the net "As of kernel 2.6.32-131, all sda* device nodes that appear in your vm***.cfg will be remapped to xvde* inside the vps. The previous behavior had been to map sda* to xvda*."10:44
DavieyMACscr: good stuff.10:45
MACscrno. Complete bullshit if you ask me10:45
MACscrthats TB of data im going to have to rebackup, let alone guests i have to manually change their fstab, guests configs and who knows what else10:46
jamespageDaviey: would you be good to sponsor another NEW package?10:55
Davieyjamespage: No, but i would be GREAT to do it.11:05
DavieyMACscr: apparently there is a patch around that can solve this issue for you11:12
DavieyAlso, booting with --scsi might work?11:12
MACscrDaviey: seems like its going to be an issue with all my guests in the future, so might as well change them all now so i dont have to worry about workarounds. Still sucks though11:16
DavieyMACscr: yeah :(11:18
jamespageDaviey: \o/ lp:~python-jenkins-developers/+junk/python-jenkins-packaging - ta11:20
Davieyjamespage: you really need to apply for MOTU :)11:24
jamespageDaviey: its on my TODO list11:24
jamespageapologies for being a PITA11:24
Davieyjamespage: don't be sorry.. it's no problem at all.11:25
Davieyjamespage: looks perfect, builds fine - uploaded.11:26
jamespageDaviey: w00t - thanks v much11:26
uvirtbot`New bug: #812847 in cobbler (universe) "Importing ubuntu desktop no distro in webui" [Undecided,New] https://launchpad.net/bugs/81284711:42
=== zz_ng_ is now known as zz_zz_ng_
=== Ursinha is now known as Ursinha-bbl
chris_99hi, i'm just wondering if there is a way to upgrade from 32 bit to 64 bit13:05
chris_99using apt13:05
joschichris_99: no13:06
chris_99so i'd just have to do a clean install then13:07
=== utlemming_ is now known as utlemming
utlemmingsmoser: ping13:13
utlemmingready to test the AMI?13:14
smoser--> iceflatline (~iceflatli13:15
smoserits not ready yet13:15
smosericeflatline, that was paste failure, sorry.13:16
iceflatlineno worries :)13:17
smoserutlemming, the build isn't done yet.13:17
smoserit looks like it will be ~ 1.5 hours. its in ebs publishing stage.13:17
utlemmingsmoser: k, so we'll hit this after the block of meetings then13:18
smoseror... we can be ready and start it during.13:18
smoserthe tests take a while to run13:18
smoserbasically, try to have the tests ready to go, just waiting on the published-ec2-daily.txt file13:19
utlemmingthat sounds like a prudent plan13:19
smbsmoser, hm... there seems to be hope that you get back xvda for oneiric. I think upstream just accepted that patch...13:25
smb(we just should be glad that ec2 does not use hda* for pvm...)13:26
utlemmingsmb: do you have the commit for that?13:27
smbutlemming, not yet. Just got an email from Konrad, saying he pulled two of my three patches13:28
dassoukiwhere does ubuntu-server store mono ?14:14
jpdsdassouki: /usr/lib/mono ?14:15
dassoukijpds: thanks i guess the issue is that the installer can't find it, and i'm trying to follow the suggestion to use pkg-configure to find it http://paste.pocoo.org/show/442216/14:17
jpdsdassouki: Installed libmono-dev ?14:19
dassoukijpds: thanks :) i think tht was it14:20
TuxBrothersomeone has expierence with failing NFS on netbooting an Ubuntu Live CD?14:21
TuxBrother(using PXE boot)14:22
=== cmagina_ is now known as cmagina
=== oCean_ is now known as oCean
h4lfl1nghaving an internet connection sharing issue. I have my primary setup as a dhcp server and it's giving ips to my secondary but no internet on the secondary15:35
chris_99could anyone explain the advantages of using LVM when installing15:42
chris_99or just using a std. filesystem15:42
patdk-wkyou can't compare those two15:43
patdk-wkthe advantages of using LVM vs partitions, you can though15:43
chris_99sorry yeah15:43
chris_99vs. partitions i meant15:43
patdk-wkpartitions have to be a continous solid chunk of disk space15:44
patdk-wkLVM it can be fragmented, split over several disks, or even moved between disks15:44
patdk-wkthey can also be snapshot to make backup easier15:44
chris_99hmm, so theres no disadvantages to using15:44
chris_99like the FS's won't be any slower?15:44
patdk-wkanother layer of indirection?15:45
patdk-wknot measurably15:45
patdk-wkwell, if you have snapshots existing, then it can noticably slow thing down15:46
chris_99i'm basically installing on several blades, each can only have one physical disk15:49
chris_99so maybe LVM isn't the best option in this case?15:49
patdk-wkusing lvm or not, greatly depends what you plan to do with the system down the road15:52
patdk-wkdo you need to make consistant backups of the fs? use lvm15:53
patdk-wkwill you want to shrink/grow the size of it, use lvm15:53
patdk-wkotherwise, probably not worth using lvm15:53
chris_99i'll give it a shot, as i like the idea of making backups at specific time points15:53
chris_99will be interesting to learn a new tool anyway15:54
chris_99appreciate your advice!15:54
smoseradam_g, you there?15:54
smoserlooking at bug 81255315:54
uvirtbot`Launchpad bug 812553 in nova "LXC instance fails to start (dup-of: 805083)" [Undecided,New] https://launchpad.net/bugs/81255315:54
uvirtbot`Launchpad bug 805083 in nova ""libvirtError: internal error cannot determine default video type" when using UML" [High,Fix committed] https://launchpad.net/bugs/80508315:55
adam_gsmoser: i am15:58
h4lfl1nghas anyone setup internet connection sharing before? with a dhcp server?15:58
smoserso i'm looking at the bug above15:58
smoserand the fix that is in that merge proposal15:58
smoserand i dont see how it could fix the issue that i'm seeing16:00
smoseri can dig deeper, but that fix definitely looks broken (see the last comment there)16:00
smoseradam_g, ^16:01
=== Ursinha-bbl is now known as Ursinha
adam_gsmoser: hmm ok, what about '--vnc_enabled=false' in nova.conf as a workaround?16:02
smoserthat might work. i can test it.16:03
astrostltcp6       0      0          :::*                    LISTEN      16101/java16:05
astrostlthat's from netstat -tlnp, on a server running tomcat.  it doesn't even report that it's listening on tcp, just tcp6, and it lists a v4 address in a v4 "local address" field.  is this a netstat bug?16:05
astrostlv4 address in a v6 "local address" field, i mean16:06
adam_gsmoser: works well for me. not sure why chucks commit for that and Bug #749973 wouldnt be working16:06
uvirtbot`Launchpad bug 749973 in nova "libvirtError: internal error cannot determine default video type" [High,Fix released] https://launchpad.net/bugs/74997316:06
smoserchucks commit has broken logic, adam_g16:08
smoser>>> for s in ['lxc', 'uml']:16:10
smoser...   if s != 'lxc' or s != 'uml':16:10
smoser...     print "would use vnc"16:10
smoserwould use vnc16:10
smoserwould use vnc16:10
smoseradam_g, ^16:10
smoserlooking at the diff of: https://code.launchpad.net/~zulcss/nova/lp805083/+merge/6689616:10
adam_gsmoser: yah, duh. im awake now.16:14
smoseri will un-dupe that bug and submit a merge proposal to actually fix it16:15
adam_gsmoser: ok. what about Bug #812548 ? have you tried using flat manager instead? that seems to be configuration error16:18
uvirtbot`Launchpad bug 812548 in nova "bridge not set up correctly with LXC and all-in-one system" [Medium,New] https://launchpad.net/bugs/81254816:18
smoseradam_g, how do i do that? it probably is configuration error.16:18
smoseror 'luser' error, one of the 216:18
smoserbut really, you should get a better failure one way or the other.16:19
adam_gsmoser: i editted the script you were using to do it. you need to specificy network_manager and flat_network_bridge in nova.conf (i believe before you do 'nova network create')16:19
adam_gi agree that the failures coul dbe more graecful and docs could be better, especially surrounding nova-network flavors16:21
uvirtbot`New bug: #809400 in php5 (main) "Cannot compile any version of PHP I want on Lucid due to dependencies in apache2-prefork-dev" [Undecided,New] https://launchpad.net/bugs/80940016:25
juliuxdoes somebody know if it is possible that virt-install is not terminating bevor the instances is setup full?16:36
ahs3adam_g, smoser: so i'm trying to set up the same thing (openstack on a single machine, with LXC) and running into the same problems as the bugs you've been filing...16:47
smoserahs3, so i think we're stuck on the gateway bug now that is preventing it16:49
smoserbug 80776416:49
uvirtbot`Launchpad bug 807764 in nova "KeyError: 'gateway6' - caused by unchecked hash key reference" [Undecided,Confirmed] https://launchpad.net/bugs/80776416:49
adam_gahs3: which bug?16:49
adam_gim putting together a merge for that one now16:49
Davieysmoser: have you fixed it yet?16:49
Davieyah good, adam_g is saving the day16:49
smoserthats because adam_g rules and smoser drools16:50
ahs3adam_g: 812553 -- happens in natty, too16:50
Davieybug 81255316:51
uvirtbot`Launchpad bug 812553 in nova "LXC instance fails to start" [Medium,In progress] https://launchpad.net/bugs/81255316:51
ahs3the patch for 807764 seems to work around that and at least let you get further16:51
* Daviey wonders why smoser is fixing it in Ubuntu packages rather than just upstream16:52
ahs3adam_g: merge for which one?  807764?16:53
adam_gahs3: yes16:54
smoserDaviey, we have this nice tool 'ubuntu-bug'16:56
smoserthat i use to open bugs with16:56
smoseri collects information about the system16:56
smoserand it opens bugs against ubuntu packages16:56
ahs3smoser: can you get lxc to start a guest inside a guest, independent of libvirt and openstack?  /me is prolly doing something dumb...17:00
smoserahs3, well... i'm fairly sure that should work17:03
smoserbuti 'm not set up to test it *right now*17:04
smosersorry to not be helpful17:04
ahs3smoser: np.  i'm trying it as we speak17:10
Davieysmoser meet serue_17:10
smoserwho the heck is serue_17:11
smoserserue_, Daviey thinks i should bug you about some software a guy named hallyn is in charged with17:11
Davieysmoser: duh, it's ircname  : Unknown17:11
smoseri'm looking for some help for bug 80085617:12
uvirtbot`Launchpad bug 800856 in cloud-init "resizefs module causes problems on LXC containers" [Medium,Triaged] https://launchpad.net/bugs/80085617:12
serue_smoser: did i not reply?17:17
serue_i did17:18
serue_smoser: what do you think about the upstart job i proposed?17:18
=== serue_ is now known as hallyn
hallynkirkland: hey, are you ready?  where did you want to meet?  mumble?17:41
kirklandhallyn: I'm sorry, I had another meeting scheduled on top of this one17:41
kirklandhallyn: can we do it as soon as this one is over?  <1 hour?17:41
hallynwe can try.17:42
hallyn(I'd locked the kids in a closet for this (j/k), so they may prevent a meeting, but let's try17:43
hggdhall: it seems the 10.04.3 20110719.1 server images (and alternate) are broken17:43
hggdhinstallation fails with "no kernel modules found"17:45
=== manjo` is now known as manjo
olokki1hello ppl, how can i change the default commandline for force check filesystem on reboot? i cant seem to find this info anywhere18:00
olokki1(i know how to force the check, i just want to change the options)18:01
SpamapSolokki1: the options for the fsck ?18:19
SpamapSolokki1: it does a very quick check by default.. if you want to do the full check, that should be manual usually.18:19
olokki1SpamapS: yes sure, but is there a way to change the default?18:27
olokki1SpamapS: i just want to do a -D on a remote system18:28
=== zz_zz_ng_ is now known as zz_zz_zz_ng_
SpamapSolokki1: I don't think you cna change the options no18:29
SpamapSolokki1: remote consoles are useful for this sort of thing. :)18:30
=== med_out is now known as medberry
hallynkirkland: I'm here if you're ready18:31
olokki1SpamapS: yeah i wish i had payed for a kvm18:31
kirklandhallyn: nearly done18:31
kirklandhallyn: okay18:32
SpamapSolokki1: if you can tie the lost performance to money, I bet you can justify the KVM easily. :)18:35
kpettitAnybody know a editor/tool that can re-format HTML so it's all consistant?18:36
shauno'tidy' ?18:36
kpettitAlot of the HTML I have to look at is all squished up on a few lines, makes it hard to look through18:36
kpettittidy?  I don't know what that is, I'll look it up18:37
shaunoit'll argue with you if the html is malformed, but it's otherwise handy for tidy -indent -ashtml to turn generated blobs back into something readable18:38
kpettitah cool.  Just installed it.18:39
kpettitshauno, ahhhh that's much better.  Thanks, works like a charm18:41
SpamapSlifeless: when did you want to chat about Ensemble?18:52
DavieyRoAkSoAx: Yo'll.. how are those cobbler bugs looking?19:35
uvirtbot`New bug: #813110 in php5 (main) "CVE-2011-1938" [Undecided,New] https://launchpad.net/bugs/81311019:36
uvirtbot`New bug: #813115 in php5 (main) "CVE-2011-2202" [Undecided,New] https://launchpad.net/bugs/81311519:36
philsfhi, I'm used to installing debian and ubuntu on desktop-like hardware  and I'm trying to install on a DELL R710  for the first time. I'm puzzled that the four HDs appear as one device (sda) apparently set by hardware RAID. I'd like to manage each HD individually, how can I do that?19:39
RizVia the raid utility19:40
philsfI'm familiar with mdadm, but I don't know how to access the hw raid19:40
RizWatch the console as the server boots, it's ctrl+e or something like that19:40
RizYou can then manage your disks19:40
RizOr isntall dell open manage19:40
philsfRiz, are you familiar with this set? is this the default?19:41
RizTypically, unless otherwise stated. Dell will send the server preconfigured.19:41
RizI'm not sure what they default to for the 710's...likely a raid 5 setup.19:41
philsfforgive my ignorance, what does this dell open manage do?19:42
philsfI don't think I have a manual, only some CDs, and no PDFs within19:42
RizOpenmanage will let you work with the raid card, lets you do your virtual disks, replace drives and what not19:43
Rizwith that said, it's far better to use hardware raid vs software19:43
RizIMHO that is19:43
philsfand hw RAID is accessible through Ctrl-E in boot?19:45
RizSomething like that, watch the console, it'll tell y19:45
philsfit is, but it mostly let's me configure NIC stuff19:46
philsfiDRAC6 config utility, is that it?19:46
RizErm, wait19:48
philsfoh, found it. It's Ctrl-R, instead19:48
Rizno, iDRAC is your managment card, another very usefull thing19:48
ppetrakiphilsf, you'll see multiple "tabs" for virtual disks and physical disks19:48
ppetrakiphilsf, not the worlds greatest UI19:49
philsfppetraki, yup, I see that now, in a tree-like UI19:49
ppetrakiphilsf, are you sure you want to do this?19:49
philsfppetraki, why not? should I use all HDs as one single device?19:49
smoseradam_g, around ?19:50
uvirtbot`Launchpad bug 812539 in cloud-init "FQDN does not get set correctly in /etc/hosts" [Medium,Triaged]19:50
ppetrakiphilsf, well, if you have a sizable cache on the controller, performance, and if battery backed, data integrity19:50
ppetrakiphilsf, and dell will support you, soon as you say the words "MD" they'll give up on you19:51
philsfppetraki, what if I need to remove one disk? how would I substititute it19:51
adam_gsmoser: hey19:52
ppetrakiphilsf, the chassis has drive beaconing, it'll blink the bad disk for you19:52
smoseri'm looking at that first hunk19:52
ppetrakiphilsf, err, the slot that is19:52
philsfppetraki, sorry, that's not what I meant19:52
smoserhostname = cloud.get_hostname()19:52
philsfwith my mdadm background, I was planning to use the 4 available HDs in two batches of RAID119:52
adam_gsmoser: ok19:52
smoserif that was None, then wouldn't 'fqdn = cloud.get_hostname(fqdn=True)' == None?19:53
ppetrakiphilsf, sure ok19:53
philsfppetraki, I was hoping I could do the same with the hw controller19:53
ppetrakiphilsf, it had better be able to do that :)19:53
smoseras the only way it returns None is if local-hostname isn't available19:53
philsfppetraki, so what exactly are you recommending me *not* to do? I'm confused19:54
philsfppetraki, not to use mdadm, you mean?19:54
ppetrakiphilsf, the question to ask is what do you gain from using md? To me the answer is "portability".19:54
smoseroh. wait. i'm reading it wrong, adam_g19:55
adam_gsmoser: fqdn = cloud.get_hostname(fqdn=True) only when local-hostname is available, otherwise use other means19:55
philsfppetraki, I will gladly use whatever controller Dell wants me to use. I just don't like this default linear config19:55
ppetrakiphilsf, oh, I didn't know it was linear, that controller should atleast be able to do RAID 0/119:55
philsfsorry, raid519:55
smoserso then my only sisue at this point, adam_g is that local-hostname on eucalyptus is an ip address19:56
philsfppetraki, what I have now is 4x140GB disks, and 1x408GB raid5 virtual drive19:56
lifelessSpamapS: hi19:57
ppetrakiphilsf, *looking up specs*19:57
philsfI can't tell from this config utility if this already has mirroring configured for the disks19:57
ppetrakiphilsf, so which PERC do you have?19:58
adam_gsmoser: is it an ip or a string constructed like ip-192-168-5-1?19:58
smoseran IP19:58
smoserwhich is then turned into a string by cloud-init19:58
philsfppetraki, perc i/619:58
smoser(right after your 'if fqdn' statment)19:58
ppetrakiphilsf, ah ok19:59
ppetrakiphilsf, so it appears to support every RAID level under the sun19:59
ppetrakiphilsf, and has 256M cache20:00
adam_gsmoser: sec20:00
philsfgood. I know raid5 allows mirroring, but I never used it. does it always have mirroring, or is it optional?20:00
ppetrakiphilsf, RAID5 is distributed parity, can tolerate 1 fault and operate in a degraded state20:01
adam_gsmoser: better? http://paste.ubuntu.com/647593/20:01
adam_gsmoser: actually, ill rework it in a bit20:02
philsfhmm, I think I know understand how this is broken down: 4x140GB disks would totalize 560GB, but I have a 408GB virtual disk. that leaves around 150GB for a parity "disk" and probably some hidden rescue partitions20:02
philsfdoes that sound correct?20:02
ppetrakiphilsf, it really depends on your application, do you value read perf over write perf?20:03
philsfppetraki, TBH I don't expect much traffic here, so I'll probably operate on very low demand, given the cpu availability20:04
philsfppetraki, this will become some mail and webservers, mostly20:04
philsfI'd like good write performance, for mail access and maybe an FTP/SMB in a forseable future20:05
philsfppetraki, does that mean I should trade raid5 for, say, raid1?20:05
ppetrakiphilsf, then I'd take RAID 10 over 520:05
philsfppetraki, and will Dell support frown if I change this default?20:06
lifelessSpamapS: I have a call with flacoste now, then Lynne and I go see the midwife from 930 (its 800 now) till 1030. Anytime outside those constraints.20:08
ppetrakiphilsf, perhaps, a RAID 10 is considered a "mirrored strip" or RAID 0 + 1. So you could concat two of your disks, and mirror them, giving you 280G20:08
ppetrakiphilsf, RAID 1 would limit you to the size of the physical disk, so 140G mirrored. Do you have any spares?20:09
ppetrakiphilsf, ideally this should be planned around a spare or two, resident in the chassis. Unless you like running down to the datacenter :-)20:09
RoAkSoAxDaviey: haven't yet worked on them I'm working on ensemble atm20:10
ppetrakiphilsf, the easiest config to manage would be a RAID 1 (140G) with two hotspares20:10
ppetrakiphilsf, I'm pretty sure the dell firmware automagically puts the spare into service, check the docs20:11
RoAkSoAxDaviey: i wanna finish some stuff with ensemble first and then will look at the cobbler stuff20:11
ppetrakiphilsf, also, when you go the vdisk route, the firmware will watch for physical disk problems and report back via the admin tool. less work for you20:12
DavieyRoAkSoAx: okay.. great stuff.20:12
philsfppetraki, sounds brilliant20:12
ppetrakiphilsf, and you get to know which disk is in which slot :-), unless you plan to write serial numbers down on the front of each cage20:12
philsfgoodbye mdadm :)20:12
ppetrakiphilsf, at least in this case, I love it and it has its place20:13
philsfppetraki, as I said earlier, I don't think the performance demand will require much fiddling, for me to choose raid 10 over 5. and this default setup seems sound and provides me the best space offer for the buck20:13
philsfppetraki, I was just worried that it might be raid0, with no mirror, but I hadn't done the math then.20:14
philsfppetraki, thanks very much for the clarifications, and the pointers20:15
ppetrakiphilsf, Dell usually defaults to some sort of HA setup, better to provide "safe" defaults then get a customer call about lost data20:15
=== fenris_ is now known as Guest70853
philsfwill see now the link on performance20:16
philsfhmm, writes are really much faster. sounds good to me20:17
jnsl_this might be more of a apache question, but nonetheless.. can i rewrite urls in a folder from site.com to site-com ?20:19
philsfppetraki, thanks again, will sleep on all that20:20
jMCgjnsl_: #httpd20:20
ppetrakiphilsf, no problem, glad I could help20:21
lifelesshallyn: is there a workaround for bug 802985 ?20:42
uvirtbot`Launchpad bug 802985 in eglibc "[lucid] /var/lib/dpkg/tmp.ci/preinst: 399: arithmetic expression: expecting EOF: "3.0-0-generic"" [High,Triaged] https://launchpad.net/bugs/80298520:42
smoseradam_g, http://paste.ubuntu.com/647621/20:43
lifelessif we can't build a lucid container in oneiric, we'll be rather wedged lxc wise in the Launchpad team20:43
smoserbut i wonder if we should not just have the cloud do the platform.node() magic20:44
=== Guest70853 is now known as ejat
hallynlifeless: cjwatson had mentioned a few, I can't remember which was the most palatable20:46
hallynlifeless: I guess I've just been creating lucid containers under natty instead :(20:47
lifelessis there some escalation process I should use to say 'this is going to be very important to us' ?20:48
hallynlifeless: ping cjwatson and bump the debootstrap bug up to critical?20:48
lifelessdoing in -devel20:49
hallynstgraber: are you around by chance?20:49
stgraberhallyn: yep20:49
hallynstgraber: would you mind, when you get a chance, sanity-checking my debdiff for lxc on bug 813075 ?20:50
uvirtbot`Launchpad bug 813075 in lxc "lxcguest should provide a way to tell if this is lxc container" [Medium,In progress] https://launchpad.net/bugs/81307520:50
hallynstgraber: mainly worried about tossing in lots of new names in the upstart job namespace and in /usr/bin20:51
hallynbut i think it's overall a very useful feature, so probably worth it20:51
hallyneveryone wants to know if they're in the matrix20:51
lifelessred pill, blue pill20:51
stgraberhallyn: you could probably use "grep -q" but other than that, looks good20:53
stgrabermy current way of doing it is: [ -f "/proc/1/cgroup" ] && grep -vq "/$" /proc/1/cgroup && echo "It's lxc"20:54
stgraberbut it's not always true :)20:54
hallynstgraber: right, my main complaint about that is that initramfs or systemd can screw you :)20:55
hallyngrep -q.  feh i always forget about that :)20:55
hallynstgraber: cool, thanks.  Would you mind sponsoring that too?  :)20:55
hallynoh nm20:56
stgraberhallyn: yep, I can sponsor it20:57
hallynstgraber: thanks!  the whole shebang with using grep -q is at http://people.canonical.com/~serge/lxc_0.7.4.2-0.3ubuntu4-package20:58
stgraberhallyn: just wondering, did you check that lxc-is-container is executable?21:00
hallyni tested on an ec2 instance...21:00
hallynshould that being enforced by the packaging?21:01
hallyn(actuall, i'm guessing that dh_install already makes sure anything in usr/bin gets +x, but i could be wrong)21:02
stgraberyeah, that's the bit I'm not too sure about :) checking here quickly21:02
stgraberhallyn: uploaded21:06
hallynstgraber: thanks!21:07
hallyni must run for a bit, bbl21:08
smoseradam_g, could you try: http://paste.ubuntu.com/647637/21:10
smoserand sanity check that too?21:10
adam_gsmoser: sorry, was at lunch.21:13
adam_gill test it now21:13
smoseryeah, take alook21:14
smoseri will look later or tomororw21:14
RoAkSoAxkirkland: ping?21:41
kirklandRoAkSoAx: pong!21:41
RoAkSoAxkirkland: howdy!! I seek your advice and help with a postrm in cobbler21:41
kirklandRoAkSoAx: heh, sure21:42
kirklandRoAkSoAx: should i branch the latest code?21:42
RoAkSoAxkirkland: i don't think it would be necessary :)21:42
kirklandRoAkSoAx: okay, hit me21:42
RoAkSoAxkirkland: so when purging the package it fails as per bug #80590121:43
uvirtbot`Launchpad bug 805901 in cobbler "cobbler failed to purge" [High,Confirmed] https://launchpad.net/bugs/80590121:43
RoAkSoAxkirkland: I changed the cobbler.postrm to:http://pastebin.ubuntu.com/647663/21:43
RoAkSoAxkirkland: however, now it doesn't fail, but it gets stuck and does not return21:44
RoAkSoAxkirkland: log setting -x: http://pastebin.ubuntu.com/647661/21:44
RoAkSoAxkirkland: any ideas of why it might be?21:45
kirklandRoAkSoAx: your set -x log looks like it completed, no?21:46
RoAkSoAxkirkland: yes, the set -x log looks like completed, but never returns21:46
RoAkSoAxkirkland: this is what is in /var/lib/dpkd/info/cobbler.postrm : http://paste.ubuntu.com/647664/21:46
kirklandRoAkSoAx: lines 5 and 22 are redundant21:47
kirklandRoAkSoAx: so i don't think you need lines 5 and 621:48
RoAkSoAxkirkland: if I don't set lines 5 and 6 it fails as per the bug description21:49
kirklandRoAkSoAx: erm21:49
RoAkSoAxkirkland: seems like apache2 is the problem21:51
RoAkSoAxkirkland: i just killed apache processes and the removal resumed21:51
kirklandRoAkSoAx: hmm21:51
kirklandRoAkSoAx: missing configuration?21:51
RoAkSoAxbut got stuck again in "Building database of manual pages ..."21:51
RoAkSoAxkirkland: might be indeed21:52
TREllisDaviey: have you connected glance to swift before?21:52
TREllisbah 11pm in the UK21:53
TREllisanyone setup glance connected to swift backend before?21:53
RoAkSoAxkirkland: yeah it is apache's problem21:54
TREllissorted it, no worries22:02
RoAkSoAxkirkland: I think we don't even really need cobbler22:03
=== kentb_ is now known as kentb-out
kirklandRoAkSoAx: heh, what do you propose in place of cobbler?22:04
RoAkSoAxkirkland: err I meant "cobbler.postrm"22:05
RoAkSoAxkirkland: as it's only there to restart apache, but cobbler binary itself doesn't mess with apache2. Cobbler-web does though22:05
kirklandRoAkSoAx: heh :-)22:07
kirklandRoAkSoAx: right22:07
kirklandRoAkSoAx: i'd agree witht hat22:07
RoAkSoAxkirkland: hehe22:10
hggdhhallyn, adam_g: is the test rig available?22:43
hggdh(need to test UEC on Lucid)22:44
quentusrexAnyone know of a way to test the system memory without rebooting the machine?23:13
quentusrexI have a remote box that seems to have faulty memory, but I need to know before I go to the datacenter if it is the memory or if the system has other problems.23:14
lifelessSpamapS: yo23:15
SpamapSlifeless: howdy23:21
lifelessSpamapS: is it too late for you ?23:22
lifelessmidwife ran late; ended up asking for a statistical analysis of $various $options23:22
SpamapSlifeless: not at all, I've got about 45 min left23:23
SpamapSlifeless: sure, clintfewbar23:24
r4___im running clonezilla SE on ubuntu 10.04 LTS 32bit machine, i created an image of ubuntu 11.04 64-bit. when i restore the image, the machine cannot boot...(writing what happens)23:24
r4___what happens when i try to boot...a really fast msg pops up on the screen that reads "error: file not found" and goes away23:24
r4___i suspect it's a grub problem with the mbr23:25
r4___because when the image is restored, it tries to install grub from the restored OS but fails and installs grub from the running OS clonezilla SE provides23:25
r4___i feel that since clonezilla is running a 32bit OS and the image is a 64bit OS that it causes a problem23:25
r4___does this sound reasonable or could it be something else?23:25
jMCgr4___: "feel" and "reasonable" doesn't "feel right"23:30
r4___jMCg: i was right..reinstalled grub from a live enviroment and it booted :)23:31
r4___crappy issue though... :(23:31
r4___Does ubuntu 10.04 LTS use grub or grub2?23:52
twbd-i still asks you with priority=low, but if you ask for grub it pulls in the upgrade compat package :-/23:53
twb(As at lucid, I mean.)23:53
r4___hrm...odd issue nontheless.23:54
twbr4___: perhaps if you described it...23:54
r4___read up23:54
r4___oh sorry23:54
r4___you just joined23:54
twbAh, sorry.23:55
twbCan you pastebin or /msg me the scrollback?23:55
r4___sure can23:56
r4___check query23:58
twbOK; AFAIK grub doesn't care if the OS is 32 or 64-bit.  If the versions in lucid and clonezilla are very different, that could cause problems.  It would be useful to know if (the installed) grub is configured to not display anything, as is the lucid default -- you have to hit shift at exactly the right time to get it to display anything.23:59
twbMost likely I think is just that clonezilla installed grub incorrectly.23:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!