[00:13] <YankDownUnder> Got a server (brand new) that literally every time I power it down, I've got to re-install grub2....any clues?
[00:16] <SpamapS> YankDownUnder: you're probably not installing grub to the right place.
[00:16] <fidelix> Hey, how do i mv/cp files recursively overwriting files and folders on the destination?
[00:17] <YankDownUnder> SpamapS, I'd have assumed that (as in this case) I've done it the same as I always do for this type of build... (grub-install /dev/sda => MBR?)
[00:17] <YankDownUnder> fidelix, cp -Rf
[00:17] <fidelix> This does not work.
[00:18] <YankDownUnder> fidelix, cp -rf ?
[00:19] <fidelix> I'll check
[00:19] <fidelix> The first one worked.
[00:20] <fidelix> Thanks YankDownUnder
[00:20] <YankDownUnder> fidelix, Yer welcome mate
[00:25] <SpamapS> YankDownUnder: depends on what device your system is actually booting from.
[00:25] <SpamapS> YankDownUnder: its not a hardware raid card by any chance, is it?
[00:27] <YankDownUnder> SpamapS, RAID is disabled on the mobo...all is fine when I actually do a warm boot, but from a cold boot, it freaks out...
[00:40] <SpamapS> YankDownUnder: oh, that is actually pretty odd. I wonder if the controller is somehow screwing with writes to the MBR because of the RAID.
[01:04] <ryoohki> how do i tell upstart which services to add for startup and remove unwanted services that are starting now?
[01:06] <fluvvell> I'm wondering if hiddev is availble in the server kernel, or what the options are for usb based ups - I use apcupsd but I have no /proc/bus/usb/devices  - anybody familiar in this area?
[01:06] <fluvvell> Im using 8.04
[01:25] <YankDownUnder> SpamapS, There's a thought...have to check this out - I'll be in front of the server next week - so have to see what I can find out via remote...hate having to go through this all the time when the machine is powered down - it's going to be taken down at least two more time in the next month before it's got a perm home...
[01:41] <tucemiux> anyone knows the path to httpd.conf ??
[01:41] <tucemiux> pwd
[01:41] <tucemiux> ahh everything is in /etc/apache2
[01:42] <ChmEarl> tucemiux, /etc/apache2/ yes and see /etc/apache2/sites-available/default
[01:43] <qman__> ryoohki, the only way I know of is to manually edit the files, which are located in /etc/init/
[01:47] <clusty> anybody here with raids and also bothered to align partitions properly ?
[01:47] <clusty> curious if it's worth the hassle
[01:57] <ryoohki> qman__: thanks!  i was trying to avoid that but i could see vim was one of the ways
[02:02] <qman__> clusty, I didn't bother, but I don't think my drives have 4k sectors
[02:03] <qman__> of course my transfer speeds are limited by gigabit ethernet anyway, so I'm not super concerned about performance
[02:03] <clusty> qman__: i am getting strange speeds
[02:04] <clusty> on windoze i get from 30mb/s to 210
[02:04] <clusty> on a 3 disk raid 5
[02:05] <qman__> locally or over the network?
[02:05] <qman__> gigabit ethernet has a theoretical maximum of 125MB/s, but in reality is limited to closer to 100MB/s
[02:05] <clusty> qman__: locally
[02:05] <qman__> the fastest speeds I've ever sustained over SMB are about 80-85MB/s
[02:06] <clusty> qman__: also IO totally chokes it. as in a simple copy overloads the IO so much the PC is unresponsive
[02:06] <clusty> can barely wake up from screen saver
[02:06] <qman__> fakeRAID?
[02:07] <clusty> qman__: real raid
[02:07] <clusty> intel
[02:07] <clusty> barely any cpu used
[02:07] <clusty> but IO used to the brim
[02:08] <clusty> would have expected the io scheduler to perform in a decent manner
[02:08] <qman__> expecting windows to perform in a decent manner is a mistake
[02:08] <clusty> no no
[02:08] <clusty> that is linux
[02:08] <clusty> if windoze blows, i can understand
[02:08] <clusty> ...and expect
[02:08] <clusty> D:
[02:08] <qman__> what controller is it?
[02:09] <clusty> srcsas18e
[02:09] <qman__> misaligned blocks on 4k sectors could cause that, but there's likely something else going on here
[02:09] <clusty> i did not align anything
[02:09] <qman__> could cause slowdown, that is
[02:09] <qman__> and what kind of disks?
[02:09] <clusty> 7200 32mb cache disks
[02:09] <clusty> 1tb
[02:10] <clusty> middle of the road
[02:10] <qman__> WD, seagate, samsung? RAID class or consumer? WD Greens?
[02:10] <clusty> wd consumer black
[02:10] <clusty> one i managed to kill tLER
[02:10] <clusty> other 3 were to new :(
[02:10] <clusty> kill=enable
[02:11] <clusty> qman__: so you think alligning would help a lot?
[02:12] <qman__> it would improve performance certainly, but I don't think it's what's causing the whole machine to choke like that
[02:12] <ryoohki> qman__: do i comment out the "start on started" or change it to something like "start on manually" ???
[02:12] <qman__> but only if those disks use 4k sectors
[02:12] <qman__> I don't know if they do
[02:12] <clusty> btw, do i have to recreate partitions, or just moving them to the right boundary would be enough?
[02:13] <qman__> you could probably move them, but I don't know how or what utility to use
[02:13] <clusty> parted
[02:13] <clusty> can do it
[02:13] <clusty> so what do i need to align to?
[02:13] <clusty> 4KB multiple ?
[02:14] <qman__> you want your filesystem's 4k blocks to match up with the 4k sectors, so I guess
[02:14] <ryoohki> qman__: a moment of your time, please
[02:15] <clusty> qman__: the stripe size/width don't matter ?
[02:15] <qman__> ryoohki, I would comment out the "start on" line(s)
[02:15] <qman__> to disable a service
[02:15] <qman__> clusty, all the stripes/widths I know of would be multiples of 4k
[02:15] <qman__> 64k, 128k
[02:15] <ryoohki> qman__: thanks!
[02:16] <qman__> so they should be fine, unless the controller is reserving an amount of space before your filesystem that is not a multiple of 4k
[02:16] <clusty> qman__: they are indeed. 256kb x3
[02:16] <clusty> i can waste space
[02:17] <qman__> the thing is, most things are multiples of 4k, and most filesystems already use 4k blocks
[02:17] <clusty> i am not gonna cry over 4kB-1 B :D
[02:17] <qman__> the only problem arises when something reserves an amount of space that isn't a multiple of 4k
[02:17] <qman__> such as the MBR
[02:17] <clusty> do extended partitions eat a non multiple of 4 ?
[02:18] <qman__> so your system has to be intelligent enough to slide it to the 4k sectors
[02:18] <qman__> that, I don't know enough about
[02:18] <qman__> wikipedia should be some help there
[02:18] <clusty> thanks for the hint. i will look at that
[02:19] <qman__> but again, that shouldn't be system crippling
[02:19] <qman__> it'd cause notable slowdown
[02:19] <qman__> but it wouldn't be hanging your box
[02:19] <qman__> unless the controller is doing something stupid
[02:21] <qman__> absolute worst case scenario, each disk is doing three times the number of reads/writes they would otherwise need
[02:26] <qman__> I've got older systems with disks that can't even manage 30MB/s that run just fine
[02:34] <asanir22> hello, is it possible to choose kernel during Ubuntu server installation process?
[02:37] <asanir22> Is it possible to choose or change default kernel during ubuntu server installation process?
[02:38] <rasengan> Anyone have any good ideas on measuring PPTP traffic?
[02:38] <rasengan> (bandwidth)
[02:45] <clusty> qman__: running now a long iozone to see really what sort of bottlenecks are there
[02:47] <asanir22> anyone know how to choose kernel during "expert mode" installation?
[02:50] <asanir22> anyone know how to choose kernel during "expert mode" installation?
[02:52] <clusty> qman__: i have seen this also on another work PC
[02:52] <clusty> with a raid
[02:52] <clusty> that heavy DB access makes it so unresponsive it stops pinging for a short while
[02:55] <asanir22> anyone know how to manually choose kernel during "expert mode" installation? (sure it's possible)
[02:59] <clusty> asanir22: do you stutter ?
[03:01] <asanir22> clusty, what do you mean by stutter?
[03:05] <asanir22> nobody replies!
[03:20] <clusty> asanir22: that can mean 2 things: ppl don't know, or they don't like you. either way asking every 5 min won't change jacl schitt
[03:23] <asanir22> sorry clusty, i was not aware of discipline here
[03:25] <qman__> this channel moves a lot slower than #ubuntu, patience is important here
[03:26] <qman__> as for your question, I don't know if that's possible
[03:31] <robertpayne> there an easy way to see top like 10 processes by memory usage?
[03:38] <qman__> robertpayne, certainly, see `man top`
[03:40] <qman__> there's probably also a way to do it with ps
[03:46] <clusty> i prefer htop
[03:46] <clusty> it's a candy filled version of top
[04:23] <shebaloma> i have gest xp system running i was woundering how to enable 3d for the virt video card
[04:49] <clusty> shebaloma: virtual box ?
[05:49] <YankDownUnder> to the bloke asking about 10 top processes: I've just used a conky-cli customised rc that does basically that - and I've got it to show me some other important things via the server's console...
[05:58] <SpamapS> ps auxw O r | tail -n 10
[05:58] <SpamapS> shows top 10 (albeit, in ascending order)
[08:27] <g0rd0n> i have a little problem with ubuntu on my server... basically the server has a remote video redirection, which works fine with 8.04 lts, but then i upgraded to 10.04 and now, at some point during boot, ubuntu changes font or resolution or something, and all i see in my video redirection console is garbage... is there a way i can fix that?
[08:39] <blue-frog> trying to restrict login hours for the user "test". Added   ssh;*;test;Al0900-1000    to /etc/security/time.conf  and added    account requisite pam_time.so  in /etc/pam.d/common-account  (I also tried in /etc/pam.d/login). I even rebooted the computer afterwards. The user "test" can still ssh while the time restriction should occur. What am I missing?
[08:40] <qman__> g0rd0n, when booting, enter grub, and manually edit the boot line to include "modeset=0"
[08:40] <g0rd0n> ok, i guess i can set that in menu.lst as well
[08:41] <qman__> I've run into a lot of problems with the changes myself
[08:41] <g0rd0n> i tested debian 5 -> 6 and have even more problems there
[08:41] <qman__> that should get you booting to a normal terminal
[08:41] <g0rd0n> machine won't even boot
[08:42] <g0rd0n> but i think im gonna settle on 10.04 and do things with kvm
[08:54] <blue-frog> *;*;test;!Al0900-1000  works  but   ssh;*;test;!Al0900-1000  does not (meaning test can ssh).  An idea?
[09:11] <blue-frog> ssh* is the key
[09:15] <qman__> while I don't know anything about that system, I would assume that * means allow any type of login, where ssh would only allow ssh
[09:17] <g0rd0n> is it advisable to convert ext3 to ext4 on an ubuntu 10.04 system?
[09:19] <qman__> only if you think you'll benefit from the performance gains
[09:19] <qman__> there are still some bugs surfacing with ext4, some trivial, some rather nasty
[09:19] <blue-frog> qman__, that's because we have /etc/pam.d/sshd  so ssh* works as sshd does. this I understand now
[09:26] <g0rd0n> ok then ill wait :P
[09:27] <g0rd0n> i dont want a nasty server
[09:28] <qman__> it's considered stable, but unless you stand to benefit greatly, I would just stick with ext3
[09:29] <Pragat> After configure cloud computing in ubuntu. I am unable to access frontend server with https://ipaddress:8443 in our proxy network..... How i can change ssl port
[09:32] <chalcedony> i hate to look so stupid but i am stupid. my son set up mail on his server somewhere else, and he just told me that i have to copy it off there,     i don't know how?
[09:33] <chalcedony> i think his server is ubuntu, i know my box is and the one i want to copy it to is ubuntu 10.04
[10:06] <kuttan_> How can I have dom0/pvops kernel form 10.04 thanks
[10:06] <kuttan_> where can I donwload those debs ( dom0/pvops kernel )
[10:08] <twb> kuttan_: I believe Xen dom0 kernels are apt-gettable from the main archive.
[10:09] <xperia> hello to all. i have a question related to proxys. i need to have a transparent fast proxy with the possibility to skip/filter special html tags if a php page is called. Does exist something like that ?
[10:10] <kuttan_> twb thx for reply, but I can't see anything for lucid ( 10.04 )
[10:10] <twb> kuttan_: is universe enabled?
[10:10] <kuttan_> yes
[10:11] <twb> Hm.
[10:13] <kuttan_> twb linux-image-virtual , is domU ( guest ) only . am I right
[10:14] <twb> Last time I looked, all kernels were domU capable as at lucid
[10:14] <twb> And you're right, I can't see a dom0-enabled kernel package...
[10:15] <kuttan_> I shifted from centos/rhel to ubuntu .. missing dom0 :(
[10:15] <twb> I don't use xen myself, so I don't know if that's because it's no longer necessary, or because Ubuntu have dropped dom0 support
[10:16] <kuttan_> twb Yup looks like that
[10:18] <twb> http://osdir.com/ml/kernel-team/2010-01/msg00000.html (random google hit)
[10:19] <twb> I suppose you could use hardy as the dom0
[10:19] <kuttan_> thanks let me check hardy repo .
[10:19] <twb> Hardy has dom0 packages
[10:27] <kuttan_> okay
[10:51] <boki> is there an easy way to include some html on every page that apache spits out? (thats is htm, php, everything)
[10:54] <joschi> boki: libapache2-mod-layout http://www.musc.edu/webserver/mod_layout.html
[10:55] <boki> joschi, great!
[11:08] <boki> uhmm, whats the correct way to install apache mods?
[11:08] <alex88> apt-get install libapache2-mods
[11:09] <alex88> btw, i've problems with ssh, i've set the single ip to listen and syslog says "init: ssh main process (23654) terminated with status 255" because the network is not initialized, how can i posticipe the ssh start?
[11:10] <g0rd0n> am curious to see if my method of upgrading from 8.04 to 10.04 is correct: 1. modify /etc/apt/sources.list 2. aptitude update 3. aptitude install aptitude 4. aptitude safe-upgrade 5. aptitude dist-upgrade 6. reboot
[11:11] <alex88> g0rd0n: not really
[11:11] <alex88> g0rd0n: https://help.ubuntu.com/community/LucidUpgrades#Network%20Upgrade%20for%20Ubuntu%20Servers%20(Recommended)
[11:12] <g0rd0n> lol thank you
[11:13] <g0rd0n> uh thats even easier than my way, nice
[11:13] <alex88> yep :)
[11:13] <alex88> and for sure more compatible insted of editing sources.list
[11:15] <g0rd0n> yeh
[11:15] <g0rd0n> although i have a fresh minimal install, but it cant harm to do things properly :p
[11:18] <xampart> http://codepad.org/a5cFk17e should this configuration work concerning smtp? i want to accept smtp from those 3 ip's
[11:33] <boki> how can i figure out why a certain mod is not working?
[11:35] <g0rd0n> shit
[11:35] <psteyn> I'm using 2x Apache servers mounting their document root's from a NFS server.  I notice very high load on both apache servers, but low cpu usage...is this normal with NFS clients?
[11:35] <g0rd0n> now i wish i hadnt upgraded to grub2
[11:35] <g0rd0n> lol
[11:35] <psteyn> I'm using dedicated gigabit nics for the nfs shares/server
[11:44] <g0rd0n> can i format /boot on a running system? and if yes, how can i format it so that the first sector starts at 63? seems to be a requirement for grub2...
[11:49] <xampart> \j #grub
[11:49] <xampart> i got good help there
[11:54] <g0rd0n> xampart: thanks
[12:03] <boki> how can i figure out why the content that should display with mod_layout is not showing up? (apache gave no errors on startup)
[12:07] <Roxyhart0> hi, somebody know if there are someway to use rsync without promt the password? I can use password-file as my connection use ssh (the accounts for user are ldap/pam and login to their own folders) so i cant also use private key for that. any idea?
[12:10] <twb> I'm guessing it's because of the errors.
[12:18] <giovani> Roxyhart0: you'd need to explain in more detail what exactly you're rsyning, because you've mentioned useds and their own folders and that doesn't explain why a private key can't be used
[12:19] <Roxyhart0> becauseand i have 100 users and 100 computers, so i dont want create a private key for each user in each machine???
[12:20] <Roxyhart0> I am doing a scrip to backup clients machine from each user in the server each
[12:21] <Roxyhart0> hour with rsync and i dont know how dont promt the password..the users account are ldap accounts
[12:22] <Roxyhart0> and they shoudl copy the data in the folder that just each one have permision (each user have his pwn folder which permision just to him)
[12:23] <xampart> sounds nice
[12:27] <Roxyhart0> o i can do perfectly but it promt the password and i would like to avoid it as is a script running with cron...any idea?
[12:29] <alex88> boki: look at server error log, check if it's enabled via a2enmod
[12:29] <boki> alex88, it is, nothing in logs
[12:29] <alex88> what mod?
[12:30] <rawler_> does anyone here have experience setting up Ubuntu as a Kerberized SSH-server?
[12:30] <rawler_> that is, support full SSO, with Kerberos not only as PAM backend, but actual Kerberized authentication directly in SSH?
[12:31] <rawler_> i'm about to pull my hair out..
[12:31] <alex88> rawler_: http://www.pdc.kth.se/resources/software/login-1/linux/ssh-with-kerberos-gssapi-on-ubuntu like this?
[12:32] <alex88> rawler_: also https://help.ubuntu.com/community/SSH/OpenSSH/Advanced?action=show&redirect=AdvancedOpenSSH
[12:33] <alex88> rawler_: last page
[12:33] <alex88> 13:24 -!- paulws [~paulws@166.205.139.167] has joined #ubuntu-server
[12:33] <alex88> 13:25 -!- e-jat is now known as ejat
[12:33] <alex88> sorry
[12:33] <alex88> http://developer.novell.com/wiki/index.php/HOWTO:_Configure_Ubuntu_for_Active_Directory_Authentication
[12:34] <rawler_> alex88: is heimdal-client fully compatible with mit-server?
[12:34] <alex88> rawler_: dunno, try..i've just linked tutorials that seems to be fine...
[12:35] <rawler_> alex88: well, I've already found those through google..
[12:35] <alex88> rawler_: oh sorry.
[12:36] <rawler_> I haven't tried the heimdal client yet, since our previous CentOS hosting environment runs MIT-kerberos, and I want to avoid spurious compatibility-errors..
[12:47] <lenios> anyone with experience with incremental backup of mysql? does every 5 min sound doable with a full backup each day?
[12:48] <lenios> ftp backup, i'm wondering if servers can handle that
[12:50] <twb> http://mywiki.wooledge.org/FtpMustDie
[12:50] <Roxyhart0> lenios, injust test rsync and is great!!
[12:50] <lenios> no rsync
[12:51] <lenios> tar/zip + crypt + ftp
[12:52] <lenios> backup server is a remote ftp server
[12:53] <twb> Anyway, you can't make an incremental backup of mysql files
[12:53] <twb> They're not coherent when mysql is running -- you have to either stop mysql for the backup run, or tell mysql to generate a dump (basically a huge .sql file)
[12:53] <zash> (and the .sql file will have broken charset)
[12:54] <twb> Yeah, well, mysql is inherently brain-damaged
[12:54] <twb> AFAICT if sqlite is for rapid-deployment of toys and postgres is for production, there's not anywhere in between where mysql is a good idea
[12:54] <lenios> can't i make a sql query and put that in a file?
[12:55] <twb> lenios: sure you can, but it's not going to be useful
[12:55] <lenios> it's only for a few tables, not the entire database
[12:56] <twb> zash: we should convince all those PHP dweebs to store data in slapd instead of mysql
[12:56] <twb> "after all, it's an object database -- no need for an ORM!!1!"
[12:56] <lenios> by the way, i'm using java with hibernate
[12:58] <lenios> how would postgresql be better for backups?
[12:59] <alex88> hi twb
[13:00] <lenios> oh, postgre has a PITR
[13:41] <zash> twb: no! text files is most awesome!! ;)
[13:44] <g0rd0n> is there a way to make a non-graphical install? the video redirection of my server delivers garbage past the first install screen
[13:44] <zash> g0rd0n: there's a graphical server install? :O
[13:44] <zash> g0rd0n: do you want a text based, or a fully automated installer?
[13:47] <g0rd0n> zash: text-based... well you are right, it's not graphic... but somehow i still have problems
[13:47] <g0rd0n> odd...
[13:49] <g0rd0n> i see lots of vertical stripes... i tried with nomodeset but no luck with that eitehr
[13:51] <twb> zash: the lucid server install media FORCE framebuffer on systems that have video cards
[13:52] <twb> zash: you can opt-out of that on installed systems, but not on the install media (due to a bug).
[13:52] <zash> :(
[13:52] <twb> g0rd0n: you either need to type blind, or pull out the graphics card and do the install over the serial port, or preseed the entire install, or install hardy and upgrade to lucid afterward
[13:53] <g0rd0n> serial port... interesting, i will try to connect to the serial console and see if i can do there
[13:53] <patdk-wk> twb, what about a net-install?
[13:53] <twb> g0rd0n: oh, sorry, if you're installing over serial you'll need to roll your own install medium (e.g. PXE boot).
[13:53] <patdk-wk> I had problems with graphics mode once I installed, but not during the install, but I use net-install
[13:54] <twb> patdk-wk: all my installs are PXE-based.
[13:54] <patdk-wk> mine are pxe too
[13:54] <g0rd0n> i have no possibility of influencing PXE, i just can mount ISO files remotely and boot from them
[13:55] <twb> g0rd0n: yeah, I spoke to someone else with the same kind of funky remote access
[13:55] <twb> g0rd0n: I think you're totally fucked
[13:55] <g0rd0n> lol
[13:55] <patdk-wk> roll your own net-install iso? :)
[13:55] <twb> Oh, another way would be to boot whatever's on there, and write the boot media directly onto /dev/sda, then reboot off the hard disk.
[13:56] <twb> That's heaps of fun, because if mess up the install you can't reboot to get back to the installer, because you've just blown away /dev/sda
[13:57] <patdk-wk> twb, he can use any other iso image to boot to though, to fix it :)
[13:57] <twb> If only everything was running coreboot and we just had "target scsi" style AoE installs...
[13:57] <twb> patdk-wk: I guess
[13:58] <g0rd0n> patdk-wk: is there a howto for creating the net install iso? couldnt find an official one
[14:02] <patdk-wk> g0rd0n, that would be why I said, roll your own
[14:02] <patdk-wk> I dunno if these would help or not: https://help.ubuntu.com/community/Installation/MinimalCD
[14:06] <twb> Strictly, the *minimum* you need to install is the d-i kernel and ramdisk, and a bootloader to load them into memory
[14:08] <twb> http://archive.ubuntu.com/ubuntu/dists/lucid/main/installer-amd64/current/images/netboot/
[14:08] <twb> e.g. curl .../boot.img.gz | gunzip -c >/dev/sda
[14:09] <g0rd0n> i think im gonna go with 8.04 and upgrade it as i did before, avoiding to upgrade grub2 :P
[14:11] <prodigel> hi all. I have a simple set of iptables rules that should monitor and log into mysql - using ulogd - connections from and to the server. The problem is that the first ulogd mysql connection triggers a recursive set of connections that block my mysql server. How can I make such a rule to prevent ulogd localhost mysql connections from being logged?
[14:19] <twb> prodigel: so, don't log connections to mysql?
[14:20] <twb> e.g. -A LOG -i lo -j RETURN; -A LOG -j [whatever]
[14:20] <prodigel> twb,  I've tried something like this: http://pastebin.com/xFnD49RE with no luck.
[14:21] <prodigel> twb, those are all the iptables rules
[14:22] <twb> Sorry, ICBF helping further.  Try #netfilter.
[14:27] <prodigel> twb, is this the real meaning? http://www.urbandictionary.com/define.php?term=icbf
[14:29] <patdk-wk> prodigel, why don't you try something sane, like limiting the amount of ulogs per rule
[14:29] <prodigel> patdk-wk, do you think that would be useful for firewall monitoring?
[14:30] <patdk-wk> maybe you should think about this some more
[14:30] <patdk-wk> is it really interesting to know that someone connected to mysql 1000 times per second? vs 100times?
[14:31] <patdk-wk> even syslog stops logging dups
[14:33] <mathiaz> hggdh: hi!
[14:34] <mathiaz> hggdh: how is UEC testing going on 10.04.1?
[14:38] <hggdh> mathiaz: all-in-one done and gone, having some issues on topo2 (all distributed)
[14:48] <mathiaz> hggdh: related to 10.04.1 or hardware related?
[14:52] <hggdh> mathiaz: all ssh calls to instances fail on a ping (no response)
[14:52] <hggdh> mathiaz: sorry. before SSHing, I ping the instance. All pings are getting no response
[14:52] <mathiaz> kirkland: ^^ - could you help out on the 10.04.1 UEC testing front?
[14:53] <kirkland> mathiaz: hi there
[14:53] <kirkland> mathiaz: sure, will do
[14:53] <mathiaz> hggdh: and in the all-in-one setup it works correctly?
[14:53] <mathiaz> hggdh: can you ping the NC from the CC?
[14:53] <mathiaz> hggdh: rather
[14:54] <mathiaz> hggdh: can you ping the CC from the CLC?
[14:55] <hggdh> mathiaz: yes, the all-in-one worked
[14:56] <mathiaz> hggdh: ok - so it's probably a routing issue between the CLC and the CC
[14:56] <hggdh> mathiaz: and the CC is pingable from the CLC
[14:56] <hggdh> mathiaz: sounds like routing, yes
[14:56] <mathiaz> hggdh: from the CC can you ping the private IP of the instance?
[14:57] <hggdh> mathiaz: yes
[14:58] <mathiaz> hggdh: on the CC, is there an interface define for the publich IP of the instance?
[14:58] <mathiaz> hggdh: can you ping the public IP of the instance from the CC?
[15:01] <hggdh> mathiaz: for the CC interfaces: yes, see http://pastebin.ubuntu.com/479421/
[15:01] <mathiaz> hggdh: and a ping 10.55.55.101 from the CLC doesn't work?
[15:03] <hggdh> mathiaz: it does not. Interestingly, .101 is an active instance
[15:03] <mathiaz> hggdh: I'd check the routing table on the CLC
[15:03] <mathiaz> hggdh: I'd also start a tcpdump on the CC to see if there are any packets related to 10.55.55.101
[15:04] <hggdh> mathiaz: yes, I was going to get it now
[15:09] <g0rd0n> i see LVM is selected by default now in the installer, but is it really useful on a dedicated server?
[15:11] <_ruben> sure, why not?
[15:15] <twb> Define "dedicated"
[15:18] <phidah> I tried logging in to my server via SFTP, but the login credentials were denied. Is there anything I have to enable?
[15:21] <TeTeT> kirkland: hi there, can you enlighten me how the pre-seeding of nodes is done in eucalyptus? E.g. where is the user name and password for the nodes stored on the front end?
[15:21] <g0rd0n> twb: root server, two hard disks, software raid1
[15:22] <twb> Then I fail to see why you WOULDN'T put LVM on
[15:22] <twb> phidah: the failure will be explained in auth.log on the server.
[15:22] <twb> phidah: by design, detailed failure information is not disclosed to the client.
[15:23] <g0rd0n> heh i think i never used lvm
[15:23] <TeTeT> kirkland: I see /etc/eucalyptus/preseed/, but it's empty - maybe because it is a packaged install?
[15:23] <twb> g0rd0n: just wait until a cron job files /var/log or /var/tmp and you start losing mail from /var/spool/postfix
[15:23] <twb> s/files/fills/
[15:24] <phidah> ok since I tail -f'ed the auth.log I see the same host hammering the ssh trying to log in as root...
[15:24] <phidah> Do I just ignore that or...?
[15:25] <_ruben> use fail2ban for instance to block such attempts
[15:25] <twb> phidah: well, I would install firewall rules that -j TARPIT all packets from any host that makes more than three connection attempts in a sixty-second interval.
[15:25] <twb> I don't like fail2ban and friends because even when they correctly uses -m set, it's still relying on a userspace process instead of in-kernel counters.
[15:26] <phidah> twb: okay. do you have an application that does that for me correctly?
[15:26] <phidah> or should it be done manually?
[15:26] <twb> phidah: er, it's a sequence of iptables rules, not an application.
[15:27] <phidah> I know but wouldn't it be best to have some kind of automated procedure?
[15:27] <twb> What I'm proposing is a static set of rules
[15:27] <twb> The "procedure" would be to write an /etc/iptab that's sane, then leave it alone forever
[15:28] <phidah> What would the rule be like to block after say 5 attempts within 1 minute?
[15:28] <twb> The downside of the approach is that it assumes you already have a clue wrt. iptables best practices, which few people do.
[15:29] <twb> phidah: yes.
[15:29] <phidah> Well I tried a bit of iptables before but it scares the heck out of me
[15:30] <twb> phidah: in which use something like denyhosts/fail2ban/whatever; you're less likely to blow your foot off
[15:30] <twb> *in which case
[15:41] <Omahn> zul: ping
[15:41] <zul> Omahn: pong
[15:42] <Omahn> zul: Hello! I've just pushed a branch up to LP that provides upstart scripts for the nis package. ttx has previously pointed me in your direction as a suitable person to give it a check over.
[15:42] <Omahn> zul: Is it too late anyway for maverick given that I've missed the feature freeze?
[15:42] <zul> Omahn: cool ill have a look when i can
[15:43] <zul> Omahn: it shouldnt be
[15:43] <Omahn> zul: https://bugs.launchpad.net/ubuntu/+source/nis/+bug/569757
[15:44] <hggdh> GAAHHHHH
[15:44] <Omahn> zul: We've been using it here on Lucid in production for a couple of months now without any (known!) issues but I'm not sure how I should be coping with upgrades/etc in the post/pre etc scripts.
[15:44] <hggdh> mathiaz: typo in an euca command
[15:44]  * hggdh fumes
[15:44] <zul> Omahn:cool ill take a look
[15:44] <Omahn> zul: Thanks, much appreciated.
[15:46] <kirkland> TeTeT: sorry, i'm a bit overcommitted at the moment to handle that explanation
[15:46] <kirkland> TeTeT: there are no username/passwords transferred via UEC preseed
[15:46] <kirkland> TeTeT: rather the username is eucalyptus, and the necessary public key is installed
[15:46] <kirkland> mathiaz: okay, i'm off of the last phone call, can focus on some ISO testing now
[15:47] <TeTeT> kirkland: ok, sorry to bother you in stressful times, I just saw "d-i	passwd/username	string ubuntu" in the preseed
[15:48] <kirkland> TeTeT: huh?  where?
[15:48] <kirkland> TeTeT: i'm definitely not following your request, then
[15:48] <TeTeT> kirkland: https://pastebin.canonical.com/35930/ from /etc/eucalyptus/preseed/ from neem, training cloud front-end
[15:48] <TeTeT> kirkland: line 29, but please forget it if you need to do other work now, nevermind!
[15:49] <hggdh> TeTeT: this is setting the userId to SSH to
[15:50] <hggdh> TeTeT: 'eucalyptus' is the reserved username for the Eucalyptus pacakge run-time
[15:51] <TeTeT> hggdh: I see the eucalyptus ssh key and user in eucalyptus-common section\
[15:52] <phidah> Is there any groups that users per default should be member of in order to be able to do port forwarding?
[15:53] <hggdh> TeTeT: yes, this is correct. Eucalyptus exchanges internal data between the components
[15:53] <hggdh> TeTeT: I am sorry, I just got back here. Could you re-state your issue (so that we will be talking about the same thing)?
[15:54] <TeTeT> hggdh: I wonder how the user/password synchronisation between the front-end and nodes happens when using CD based install of UEC. Especially I wonder what happens if I change the password later on the front-end
[15:54] <TeTeT> hggdh: would new nodes still get the old password or get a new one?
[15:57] <hggdh> TeTeT: what user/password are you talking about? If it is the user in the instance... there is just one, 'ubuntu', and it does not have a password
[15:57] <hggdh> TeTeT: if it is the 'eucalyptus' userId, this is reserverd for Eucalyptus usage, nobody should ever log in as it
[15:58] <hggdh> TeTeT: in this case, it does not matter *what* would be the password, anyway, since Eucalyptus does SSH via public key
[16:00] <TeTeT> hggdh: I was refering to the user on the node controller, for example 'ubuntu'
[16:00] <TeTeT> or student
[16:08] <hggdh> TeTeT: this instance userId is fixed on 'ubuntu'; we do not provide a password to it, users are expected to SSH in with public-key authentication
[16:09] <hggdh> TeTeT: so, even if the user changes (i.e., *sets*) a password for 'ubuntu', it would not get to be used
[16:09] <hggdh> normally
[16:10] <TeTeT> hggdh: hmm, is /etc/eucalyptus/preseed for the instances? I thought it is used for the node controllers that connect to the front-end?
[16:12] <hggdh> TeTeT: yes indeed. There should not be an 'ubuntu' user there
[16:13] <hggdh> TeTeT: in this case the userId inherited is the 'eucalyptus' account. There is no password associated with it
[16:14] <TeTeT> hggdh: can you confirm that the node has the same user and password as the front end? I haven't done a CD based install in ages
[16:15] <hggdh> TeTeT: every NC installed inherits the 'eucalyptus' userId (and the public key into ~/.ssh/authorised_keys). There is *NO* password associated with this Id
[16:16] <hggdh> TeTeT: actually, the 'eucalyptus' userId is a system Id (uid=121); what is actually passed is the CC's public key
[16:16] <TeTeT> hggdh: so the admin user entered in the front-end install dialogue is never passed on to the node?
[16:20] <hggdh> TeTeT: well, now...
[16:21] <hggdh> TeTeT: good Q. Now I am unsure. I will do an install from CD, fully manual, and test
[16:22] <TeTeT> hggdh: ok, was just thinking to do the same
[16:24] <hggdh> TeTeT: any install has to have an admin user, so one would have to be created. I am not sure the 'ubuntu' would be used
[16:38] <sla> hi there, does anyone know if 10.04.1 iso's will be released today?
[16:42] <jpds> sla: It's expected to be today, yes.
[16:42] <sla> jpds: tnx, time unknown i gues? kind of waiting ;-)
[16:45] <kirkland> hggdh: ping
[16:45] <kirkland> hggdh: mathiaz says you might need some help testing 10.04.1 candidates?
[16:45] <kirkland> hggdh: i'm here to help
[16:45] <kirkland> hggdh: just tell me what you need
[16:46] <jpds> sla: Things like this are really: "When it's ready".
[16:46] <sla> jpds: ic, tnx!
[16:55] <sla> jpds: is this also a place where it will be announced or should i just press f5 once in a while on http://releases.ubuntu.com/10.04/ ?
[16:55] <jpds> sla: nl.releases.ubuntu.com
[16:56] <sla> jpds: ok, super ;-)
[17:01] <hggdh> kirkland: soorry, on a call up to now
[17:02] <hggdh> kirkland: I found the issue -- between the chair and keyboard :-( A change to the bloody test script was wrong, so all tests failed (I wrote 'iccp' instead of 'icmp'
[17:02] <kirkland> hggdh: doh
[17:03] <hggdh> kirkland: so life if good again. Just a bit balder
[17:03] <kirkland> hggdh: okay, do you need anything from me at this point?
[17:03] <hggdh> kirkland: well, since you so kindly offer... yes. There is r1231 on Eucalyptus, but I am unsure how to package it in
[17:03] <hggdh> :-)
[17:04] <kirkland> hggdh: let me look
[17:08] <kirkland> hggdh: okay, i'm working on it
[17:16] <hggdh> kirkland: later on I would like to know what I should do to package them, if you do not mind
[17:17] <kirkland> hggdh: sure;  i'm working on that now
[17:17] <kirkland> hggdh: i should have a script/doc for you by then
[17:27] <hallyn> so i've got a bzr branch that i had proposed for merge, and the relevant commits were in fact merged into lucid-proposed.  But the tree is still listed as pending merge.  Shoudl I cancel the merge request, or leave that until it gets from lucid-proposed into lucid-updates?
[17:28] <hallyn> oops, meant to ask on ubuntu-devel
[17:37] <alex88> hi guys..how can i start openssh server later then normal on boot?
[17:41] <hallyn> alex88: edit /etc/init/ssh.conf
[17:50] <alex88> hallyn: where?
[17:50] <alex88> i mean, what should i change?
[17:51] <soren> alex88: Why do you want it to start later?
[17:51] <alex88> because it runs before network, i'm on a vps, and i've set to listen to single ip, and the ip is set later
[17:52] <alex88> so it says error 255 and it not start
[17:52] <hallyn> alex88: hook whatever you want to wait on into the 'pre-start script' section
[17:52] <hallyn> i.e., 'while [ ! -f /var/run/network-started ]; do : ; done ]' or something
[17:52] <alex88> hallyn: insert a sleep command there?
[17:52] <hallyn> sure
[17:53] <alex88> thank you
[17:53] <soren> alex88: Even better:
[17:54] <soren> alex88: change /etc/init/ssh.conf to read: start on net-device-up INTERFACE=eth0
[17:54] <soren> So when eth0 is up, upstart will automatically start ssh.
[17:55] <alex88> soren: wait a sec, i'll post my syslog
[17:56] <alex88> http://pastebin.com/Ua2WVEwy for this vps i have venet0:0 and venet0:1 as public ip, venet0 has 127.0.0.1
[17:57] <soren> Yikes.
[17:57] <soren> Why not have 127.0.0.1 on lo?
[17:59] <alex88> vps... http://pastebin.com/rNWPpLYq this is ifconfig
[18:04] <kirkland> hggdh: okay, eucalyptus_2.0~bzr1231-0ubuntu1_source.changes uploaded to Maverick
[18:05] <alex88> soren: as you can see i have on both lo and venet'0
[18:07] <soren> alex88: and you think the time ssh is started is your biggest problem? :D
[18:12] <kirkland> hggdh: i'm testing locally here
[18:13] <Italian_Plumber> I love how  easy it is to install a GUI on a server
[18:14] <mdlueck> I am considering upgrading a 9.04 server to 9.10 / 10.04 maybe today... have one Q. Should I run the upgrade from the server console, or is via SSH fine? Will the upgrade process kill the SSH daemon and thus disconnect my session?
[18:15] <Italian_Plumber> running updates over SSH is best accomplished using the "screen" utility
[18:15] <Italian_Plumber> that way if you get disconnected the process continues
[18:15] <Italian_Plumber> there are probably other ways too
[18:15] <mdlueck> I have not heard of "screen" except on windows of course. ;-)
[18:16] <Italian_Plumber> screen is an awesome utility... you type "screen"
[18:16] <Italian_Plumber> then it will create the screen "session".
[18:16] <Italian_Plumber> you can type "CtrlA, D" to get out of the session, but it's still running.
[18:16] <Italian_Plumber> you type "screen -r" to get back to it
[18:16] <Italian_Plumber> that's a simple example.
[18:17] <Italian_Plumber> sudo apt-get install screen
[18:17] <mdlueck> I issue that on the remote server, or before I start the ssh session from my Ubuntu workstation?
[18:17] <Italian_Plumber> on the remote server
[18:17] <mdlueck> OK
[18:17] <mdlueck> I will check into that, thanks!
[18:18] <Italian_Plumber> otherwiste the screen is running on your workstation, so you would still have problems if you were disconnected.
[18:18] <Italian_Plumber> read up on screen...it has lots of options
[18:18] <mdlueck> I was thinking perhaps it was an alternative to SSH
[18:18] <Italian_Plumber> it's not...
[18:18] <Italian_Plumber> you would ssh to your remote machine, then type "screen"...
[18:19] <mdlueck> (me knods)
[18:20] <Italian_Plumber> you could also start the screen if you were logged into that remote machine's console, then unattach from the screen and log out.... then go to where you are now, ssh, and re-attach to that same screen
[18:20] <Italian_Plumber> you would join your regularly scheduled screen, already in progress.
[18:20] <mdlueck> cool
[18:21] <mdlueck> "saved sessions" for Linux term
[18:21] <Italian_Plumber> I use screen in conjunction with rtorrent
[18:21] <Italian_Plumber> I'm constantly unattaching and logging out, then logging in from a different location and re-attaching.
[18:22] <Italian_Plumber> you can also log  and save the output of the screen.
[18:23] <RoyK> screen is a nice tool :)
[18:24] <RoyK> ctrl+a c to create a new screen etc
[18:24] <RoyK> man screen
[18:24] <Italian_Plumber> RoyK, what other features do you use?
[18:24] <mathiaz> hggdh: kirkland: so 10.04.1 UEC testing is completed and successful?
[18:25] <kirkland> mathiaz: hggdh said that the problem he saw earlier was due to a typo on his part in his testing script
[18:25] <RoyK> Italian_Plumber: ctrl+a <esc> to enter copy and scroll mode - use vi commands (ctrl+u/d etc) to scroll
[18:26] <RoyK> or just man screen - there's probably more fun in there than I use
[18:27] <Italian_Plumber> oh I probably never discovered those features because I despise vi with a passion. :)
[18:27] <alex88> mdlueck: check also byobu....nicer thing...
[18:27] <RoyK> Italian_Plumber: might be possible to switch to other modes - my advice is to learn vi(m) and use it - it rocks!
[18:29] <Italian_Plumber> heh... I understand why I should know vi... I just go out of my way to make sure pico or nano is installed on any machine I use
[18:29] <Italian_Plumber> alex88: I've never heard of byobu... what does  that do?
[18:31] <Italian_Plumber> vi reminds me too much of FreeBSD. :)
[18:32] <Italian_Plumber> if Ubuntu is linux for human beings, FreeBSD is linux for people with too much time on their hands
[18:33] <kirkland> Italian_Plumber: http://launchpad.net/byobu
[18:33] <alex88> Italian_Plumber: in 10.04 it adds some themes, showing network, ram, disk info and also display opened terminals and activity
[18:34] <kirkland> Italian_Plumber: it's a text based window manager, allowing you attach/detach from sessions, and configure a set of notifications about system stats
[18:34] <Italian_Plumber> for machines with a GUI?
[18:35] <hggdh> mathiaz: yes, tested completed successfully
[18:35] <mathiaz> hggdh: great - thanks for the help!
[18:35] <pmatulis> anybody use NIS here?
[18:37] <alex88> Italian_Plumber: gui and not
[18:42] <alex88> is possible that i get 404 for each tracker? how can i check if my isp is blocking that?
[18:44] <RoyK> Italian_Plumber: menus belong in restaurants
[18:46] <alex88> RoyK: and in windows?
[18:48] <RoyK> alex88: yeah, and KDE/Gnome and Aqua and CDE and so on
[18:49] <RoyK> but neither of those belong on a server
[18:49] <alex88> oh sure.. :)
[18:49] <alex88> any answer to my question? :)
[18:51] <RoyK> what was that again?
[18:51] <alex88> < alex88> is possible that i get 404 for each tracker? how can i check if my isp is blocking that?
[18:51] <RoyK> url?
[18:52] <alex88> RoyK: can i pm
[18:52] <RoyK> brb
[18:53] <RoyK> b
[18:53] <alex88> RoyK: don't know if you've received..can i pm you?
[18:54] <RoyK> yes
[18:54] <ssureshot> I can't seem to find where the friendly name is set for my reports that are running and being set through php... the friendly name is Website.. I need to change this any ideas?
[19:06] <RoyK> ssureshot: friendly name??
[19:08] <ssureshot> From: Website (friendly name) -- email address is correct though
[19:10] <RoyK> ssureshot: if trying to send email from php, asking on ##php might be better - in short, you want to build an email envelope and then send it
[19:12] <ssureshot> RoyK: I'm kinda curious if this a php issue or an MTA issue... I have 2 servers one gento using sendmail (works like it should) and an ubuntu server wtih the same web apps with postfix.. that has the wrong name
[19:13] <RoyK> ssureshot: I usually start with installing postfix - I'm not really comfortable with exif (the default MTA)
[19:13] <RoyK> apt-get install postfix - try again
[19:13] <raubvogel> Does anyone know if syslog-ng 3 will make it to the repositories soon?
[19:14] <ssureshot> RoyK: postfix is installed and working, is there an alias I can use for the friendly name?
[19:14] <RoyK> ssureshot: seems your webapp needs configuring
[19:14] <RoyK> ssureshot: it probably sets that 'friendly name' and tries to send the email
[19:15] <patdk-wk> ya, postfix just passes input to output, it doesn't modify anything
[19:15] <patdk-wk> unless you attempt to configure masquerade settings or something (I have only done that using sendmail though)
[19:15] <ssureshot> royk: roger that.. I think I might need to research the mail command in php.ini then
[19:15] <RoyK> ssureshot: nope - the webapp
[19:15] <raubvogel> ssureshot:  which webapp are you using?
[19:15] <patdk-wk> php.ini only sets that stuff for windows :)
[19:16] <RoyK> ssureshot: the e-mail envelope is creating it
[19:16] <ssureshot> custom webapps
[19:16] <RoyK> well, create a decent email envelope in it
[19:16] <RoyK> it's not up to php to do that
[19:16] <RoyK> I guess you can do it in the php settings, but that's rather ugly
[19:17] <ssureshot> I will do some research on the envelope then... right on.. thank you both
[19:17] <raubvogel> Also, are you using postfix as a full MTA or just to send out emails?
[19:17] <patdk-wk> next question, why are my emails going to spam? :)
[19:18] <raubvogel> patdk-wk: because they like you? ;)
[19:18] <RoyK> patdk-wk: probably because you don't create a decent envelope for them :þ
[19:18] <Pupeno> How do I manually trigger the unnatended upgrades? They seem not to be happening.
[19:18] <ssureshot> raubvogul: postfix is an internal relay server..
[19:18] <RoyK> Pupeno: apt-get update && apt-get dist-upgrade
[19:19] <Pupeno> RoyK: that's not unnatended upgrades, that's the normal upgrades.
[19:19] <RoyK> I really don't know how to 'trigger' unattended updates
[19:19] <patdk-wk> put, apt-get upgrade && apt-get dist-upgrade && reboot, in cron? :)
[19:19] <jgcampbell300> hello, I have spent about a week trying to get openchange server installed on ubuntu and I have been reading today about groupware ... the question i am coming up with is this ... what exactly is the difrence in something like an exchange server and something like Citadel or "groupware" ... from a glance it seems to be doing the same thing
[19:20] <RoyK> patdk-wk: yeah, every hour
[19:21] <jgcampbell300> Pupeno: http://serverfault.com/questions/111201/how-to-get-automatic-upgrades-to-work-on-ubuntu-server .... found that on a search
[19:21] <RoyK> Pupeno: pkgsync can also be nice, but be careful
[19:22] <patdk-wk> dunno, I have never used it, my friend does though
[19:22] <raubvogel> ssureshot: so you mean you are using postfix as a mail relay server
[19:23] <mdlueck> Italian_Plumber: screen works slick-O, thanks for suggesting it! Backup done, on to the upgrade! :-)
[19:24] <Italian_Plumber> you're welcome... always glad to hlp
[19:25] <ssureshot> raubvogel: correct it's on the local machine that is my internal webserver also...
[19:25] <jcastro> hey zul
[19:25] <zul> jcastro: hola
[19:25] <jcastro> zul: turn on full text feeds in your wordpress so your blogs don't truncate on planet
[19:26] <zul> jcastro: yeah i thought I turned that on
[19:27] <jcastro> nice job on memcached!
[19:30] <zul> jcastro: thanks
[19:36] <kirkland> hggdh: hey, is auto-registration working for you in Maverick?
[19:36] <kirkland> hggdh: it didn't work for me... i had to manually register everything
[19:40] <SpamapS> kim0: are you around? can you join us for the server team meeting?
[19:42] <hggdh> kirkland: it works. Sometimes
[19:42] <hggdh> kirkland: sometimes it does not. I have not yet opened a bug on it, low prio for me...
[19:43] <hggdh> kirkland: I have found that leaving servers from a previous install running (and with the -publication active) can confuse auto-registration
[19:43] <kirkland> hggdh: so you've been manually registering services?
[19:43] <hggdh> kirkland: every other install or so, yes
[19:44] <kirkland> hggdh: that's annoying
[19:44] <hggdh> sounds like a race...
[19:45] <hggdh> kirkland: another thing -- the -publication services *usually* do not stop if you 'stop eucalyptus'
[19:45] <hggdh> kirkland: and usually do not restart if you 'start eucalyptus'
[19:46] <Italian_Plumber> alex88: does byobu install with apt-get?
[19:46] <hggdh> Italian_Plumber: yes
[19:46] <Italian_Plumber> E: Couldn't find package byobu
[19:47] <kirkland> hggdh: have you had any trouble bundling/registering images?
[19:47] <hggdh> kirkland: no, not there (unless the SC is down)
[19:47] <hggdh> Italian_Plumber: what Ubuntu version?
[19:47] <Italian_Plumber> oh.. could be it... hardy
[19:48] <hggdh> Italian_Plumber: yes, I think this is it... I do not think byobu made it to Hardy
[19:48] <Italian_Plumber> what version does it work on?
[19:48] <hggdh> kirkland: new issue?
[19:48] <kirkland> Italian_Plumber: it is available for Hardy in the PPA
[19:49] <Italian_Plumber> i'm currently working in a vm to test it out... I'd be glad to update to a newer version of ubuntu
[19:50] <kirkland> hggdh: yeah, possibly...
[19:51] <hggdh> dammit
[19:57] <hggdh> kirkland: does hang on bundle/register? If so, I have seen it on an SC down
[19:58] <kirkland> hggdh: doesn't hang;  python throws an error
[20:02] <hggdh> that's different indeed
[20:11] <mdlueck> Italian_Plumber: Upgrade to 9.10 went very smoothly. Starting the 10.04 upgrade now.
[20:11] <kirkland> hggdh: have you seen a CLC think that it's ip address is 169.254.169.254 ?
[20:12] <hggdh> kirkland: if this is the output from 'ifconfig', yes
[20:12] <hggdh> kirkland: not sure about the final byte, though
[20:54] <mdlueck> Frazzle razzle... 9.10 to 10.04 upgrade seemed to go well, via SSH and console, accepts my ID/pw, then goes to la la land... Suggestions please?
[21:20] <mdlueck> Very weird, I left the room for a few mins, and when I came back I was logged in. hhhmmm....
[21:21] <kirkland> hggdh: okay, i just uploaded eucalyptus_2.0~bzr1231-0ubuntu2_source.changes
[21:21] <kirkland> hggdh: i had to re-fix something i fixed before, but i think Daviey dropped the patch during the quilt migration
[21:22] <kirkland> hggdh: i'm going talk to upstream
[21:29] <smoser> hggdh, ping
[21:29] <smoser> maverick uec, have you seen functional euca-get-console-output ?
[21:29] <smoser> i get empty response, although the node controller's /var/lib/eucalyptus/instances/admin/i-48580863/console.log is popuated
[21:32] <hggdh> kirkland: k. Anything I should worry about?
[21:32] <kirkland> hggdh: well, i just re-fixed the eth0:169.254.169.254 error
[21:33] <kirkland> hggdh: this might be the source of some of the registration problems
[21:33] <hggdh> kirkland: ugh. I will download the diff, and rebuild locally, then
[21:33] <kirkland> hggdh: it's building now
[21:33] <kirkland> hggdh: it'll be built in launchpad in ~10 minutes
[21:33] <hggdh> smoser: I do not really remember -- I *think* yes
[21:34] <hggdh> kirkland: I will wait, then
[21:34] <kirkland> hggdh: yeah
[21:34] <hggdh> smoser: but I will be able to check in about one hour
[21:34] <kirkland> hggdh: here's the good news, though ....
[21:34]  * hggdh perks up
[21:34] <kirkland> hggdh: i was able to register all components, run an instance, and ssh to it :-)
[21:34] <kirkland> hggdh: that's the first time i have personally done that against euca 2.0
[21:34] <hggdh> yay!
[21:35] <kirkland> hggdh: gives me a bit of confidence
[21:35] <hggdh> so there's hope...
[21:36] <hggdh> smoser: what version of euca? this latest rev should have a fix for console output
[21:37] <smoser> whatever kirkland just loaded
[21:37] <mdlueck> In server 10.04, where has /etc/default/grub been moved to this time?
[21:42] <hggdh> smoser: this is a regression, then
[21:52] <sherr> Hello - I hope someone can help
[21:53] <sherr> I am using Server 10.04 as a KVM host (to Debian Lenny) and I want to make sure the guests are properly shutdown when the host is rebooted. But I can't get it working and the guests always end up killed. This really worries me for production use because FS corruption might happen.
[21:53] <sherr> From the command line, all the scripts work :
[21:53] <sherr> virsh shutdown <guest> - works
[21:53] <sherr> virsh list - works
[21:53] <sherr> But as a sys-v init script - or an upstart script - virsh doesn't seem to work i.e.
[21:53] <sherr> virsh list -> empty/nothing
[21:53] <sherr> virsh shutdown <guest> -> does nothing
[21:53] <sherr> They don't work in runlevel 6.
[21:54] <sherr> Surely people want to shut down guests on a reboot properly (e.g. UEC). How?
[21:54] <sherr> My upstart exec script and output is here : http://pastebin.com/n6rnkVbC
[21:54] <sherr> Can anyone help?
[22:03] <hggdh> kirkland: OK, starting to test 0ubuntu2
[22:33] <hggdh> kirkland: when you had registration issues were you running an all-in-one, or distributed?
[22:33] <kirkland> hggdh: all in 1
[22:35] <hggdh> kirkland: no prob here, at least right now, with regs
[22:36] <kirkland> hggdh: cool
[22:40] <delimiter> sherr: install acpid
[22:41] <sherr> delimiter: on the guest? acpid is installed.
[22:41] <delimiter> sherr: ya, cool
[22:42] <sherr> And the host has <acpi/> in the XML definition of the guest.
[22:42] <sherr> It works from the command line i.e. virsh list / virsh shutdown guest
[22:43] <sherr> Why does virsh not work from an init/upstart script (runlevel 6)?
[22:43] <delimiter> sherr: no idea, sorry
[22:43] <sherr> Thanks anyway.
[22:44] <delimiter> runlevel 6 = reboot
[22:44] <delimiter> that your intention?
[22:44] <sherr> yes
[22:44] <sherr> I want a host reboot to shut down guests cleanly, automatically.
[22:45] <delimiter> and killing the kvm procs doesn't trigger a clean shutdown within the guest?
[22:46] <RoAkSoAx> kirkland: where can i find the Cloud in your pocket ISO?
[22:46] <sherr> delimiter: Not tried that. But I suspect not. Let me try from the command line ....
[22:46] <kirkland> RoAkSoAx: http://blog.dustinkirkland.com/2010/06/cloud-in-your-pocket-uec-liveiso.html
[22:47] <RoAkSoAx> kirkland: thanks :)
[22:49] <sherr> delimiter: I did a kill on the KVM pid, guest died. On reboot, I see an EXT3 "recovery required" as usual, so it appears to be an unclean shutdown as far as I can see. As usual.
[22:56] <delimiter> sherr: that is really bogus, the system (host) should issue a shutdown to each running guest on reboot
[22:57] <alvin> sherr: It's being worked on: bug 350936
[22:58] <sherr> Yes, been banging my head on this all day. It doesn't work using "virsh" in an init script for level 6. Someone else said level 0 doesn;t work either.
[22:58] <alvin> It doesn't work at all yet, but there are some pointers in the bug report. Didn't try them yet.
[22:58] <sherr> Ah. I did do a search earlier ... let me look.
[22:58] <delimiter> well those aren't your normal "runlevels" either
[22:59] <alvin> there are no runlevels anymore as far as I know
[22:59] <delimiter> that is a very narrow view of the universe :)
[23:00] <alvin> lol, could be :-)
[23:01] <sherr> There are runlevel modes in upstart though : start on runlevel [016]
[23:01] <sherr> That's a big bug ... lots to read. I'll take a look further tomorrow and see if I can get anything working.
[23:02] <sherr> Like many, this is serious for me. I'm testing test VM's now, but production is a different matter ...
[23:03] <sherr> Thanks alvin and delimiter.
[23:21] <SpamapS> that one keeps coming up
[23:21] <SpamapS> I wonder if we're packaging a file that gets mucked with by maintainer scripts or something
[23:22] <ruben23> guys i have several errors on startu can i hols the screen or pasue the loading is so fast can barely identify the error on a ubuntu server
[23:27] <Crewsr3> I'm getting started with EC2 and set up an AMI and install postgre and now I need to connect to the postgre db.  I'm trying to tunnel to it but I'm getting an error.  Does the ubuntu AMI have ssh tunneling locked down and if so how to I open it up?
[23:29] <qman__> Crewsr3, by default, SSH tunneling is not locked down
[23:29] <qman__> the configuration is in /etc/ssh/sshd_config
[23:31] <Crewsr3> I've been looking this file over and set up everything to open this up but I'm still not able to get tunneled into
[23:38] <sbeattie> SpamapS: hunh, odd, it looks like some sort of debconf locking issue.
[23:40] <SpamapS> Crewsr3: try passing  '-v' to ssh so you get some debugging information (three -v's will give you a ton of debugging)
[23:41] <SpamapS> Crewsr3: though unless you've given the user you're using to connect a password, connecting over tcp/ip will be dependent on ident .. which probably won't work.
[23:42] <SpamapS> Crewsr3: unless you ssh as root, and forward ident back to your local machine, which you don't want to do on so many levels. ;)
[23:42] <SpamapS> Crewsr3: actually I take that back, ident will work if you run a local identd
[23:44] <sbeattie> SpamapS: hrm, also looking at dmesg, it appears that XFS is in use; do you have other example bugs, as I'm curious if they're using XFS as well.
[23:44] <sbeattie> SpamapS: re bug 619502 that is
[23:45] <SpamapS> sbeattie: https://bugs.launchpad.net/ubuntu/+source/mysql-dfsg-5.1 .. I see 6 with a 'bad inconsistent state' mentioned. All invalid or triaged.
[23:47] <sbeattie> oh bah, reading the dmesg *for comprehension this time* just shows that something is querying for filesystems, not that xfs is actually used anywhere.
[23:51] <ruben23>  guys i have several errors on startu can i hols the screen or pasue the loading is so fast can barely identify the error on a ubuntu server