[00:06] <nat2610> hey I have this box that I installed like 2 yrs ago, at that time I gues that was 8.4 ... at that time libbost was coming in 1.35 which is perfect for what I needed, now I want to test on the same box to upgrade to 1,40 but I checked and 1.35 is not packaged anymore some very old libs installed from
[00:06] <nat2610> oups .. I was editing my sentence when I hit enter...
[00:06] <nat2610> let me rephrase it
[00:07] <nat2610> hey I have this box that I installed like 2 yrs ago, at that time I gues that was 8.4 ... at that time libbost was coming in 1.35 which is perfect for what I needed, now I want to test on the same box to upgrade to 1,40 but I checked and 1.35 is not packaged anymore. If I want to upgrade to 1.40, apt want to remove 1.35 which make sense but will I be able to then put back 1.35 if something doesn't go well ?
[00:07] <nat2610> is there a way to save the current deb ?
[00:07] <nat2610> or something like that?
[01:26] <uvirtbot`> New bug: #513544 in libvirt (main) "libvirt 0.7.5 crashed on first virt-manager connection" [Undecided,New] https://launchpad.net/bugs/513544
[02:53] <error404notfound> I have 3 machines at home, i need to put redirect of say *.domain.com to my private IP say 192.168.1.1 instead of public, i don't want to edit /etc/hosts as i want it to be one time "*.domain.com" redirect,do i need a basic dns setup here or what?
[03:02] <uvirtbot`> New bug: #513562 in samba (main) "Windows 7 Pro machines trust relationship fails" [Undecided,New] https://launchpad.net/bugs/513562
[04:28] <bendj> Hi.  I'm setting up a 1st mai server, using ubuntu-server.  I'm planning on using exim, and integrating spamassassin.  Iiuc, I can either use spamassassin directly integrated into exim, or via amavisd-new.  Each community seems to recommend their own solution.  What do "server folks" in here use -- and Why?
[05:12] <unnotused> why would my virtual copy of ubuntu server have an ip address of 192.168.56.101 ?
[05:13] <unnotused> zomg, is anyone here ?
[05:15] <twb> no
[05:15] <unnotused> figures
[05:15] <unnotused> nothing but bots
[05:15] <twb> I imagine 192.168/16 is used because that's a private address range.
[05:15] <unnotused> nothing but bots/
[05:16] <unnotused> GRRR
[05:16] <unnotused> stupid chat
[05:16] <unnotused> idk
[05:16] <unnotused> why 56.101?
[05:17] <unnotused> idk
[05:17] <unnotused> i just wish i could get it to connect to the interwebs
[05:17] <unnotused> so i could apt-get
[05:17] <unnotused> and be all good
[05:20] <unnotused> i give up for now
[05:20] <unnotused> night
[05:22] <jmarsden> unnotused: The issue is more with your virtualization setup than with ubuntu-server... what are you using?
[05:22] <unnotused> sun virtualbox
[05:22] <jmarsden> And you set Networking to ... what when you created the VM?
[05:23] <jmarsden> I'm guessing NAT, and you really want Bridged
[05:23] <unnotused> no
[05:24] <unnotused> bridged to my wireless adapter emulated as w/e their default is
[05:24] <jmarsden> I've not tried bridging to wireless... can you bridge to your wired adapter instead?
[05:25] <unnotused> lol
[05:25] <unnotused> if i had a wired network
[05:25] <unnotused> maybe
[05:25] <unnotused> ive got both enabled
[05:25] <jmarsden> Well, IMO that's probably where the issue lies.
[05:26] <unnotused> prob
[05:26] <unnotused> one way to find out
[05:26] <jmarsden> Actually for apt-get access, NAT might be fine... did you try it?
[05:26] <unnotused> nope
[05:26] <unnotused> didnt figure it would work
[05:26] <jmarsden> A server you can't access from anywhere except itself is a fairly useless server, but to get outbound access, try NAT.
[05:27] <jmarsden> Or run a cable to your nearest switch :)
[05:27] <unnotused> kks
[05:27] <unnotused> ima try both
[05:27] <unnotused> i wanna get ebox working
[05:28] <unnotused> the college i work at just got a 1/4 million dollar blade center and it doesnt work b/c the SQL is fubar
[05:28] <unnotused> i thought "HEY virtbox exported ubuntu/mysql server with ebox admin so even my boss can use it"
[05:28] <unnotused> idk
[05:28] <unnotused>  .. . .
[05:29] <unnotused> its more of a pain then i expected
[05:29] <jmarsden> Why the virtualization?  Just run MySQL on your host OS, if that is all you need to export to the world?
[05:30] <unnotused> bladecenter needs the sql server virtualized anywa
[05:30] <unnotused> anyway
[05:30] <unnotused> the exchange is running on the card
[05:30] <unnotused> :P
[05:31] <unnotused> i installed ubuntu server, got mysql set up, got the databases/users running got apache configured and even got the obdc driver working . . .
[05:31] <unnotused> now i just need a way for my boss to be like (web browser to ip address, change network settings)
[05:31] <jmarsden> Wait... so you have a blade with only a wireless network connection???
[05:31] <unnotused> no
[05:31] <unnotused> im building the virtual image from home
[05:32] <unnotused> that would be
[05:32] <unnotused> sad at the verry least
[05:33] <unnotused> i almost have everything i need done
[05:33] <unnotused> i just need access to apt-get
[05:33] <unnotused> :P
[05:34] <jmarsden> Plug in the wire, bridge to eth0, and if your wired network has working DHCP, you should be all set when you restart the Ubuntu server VM.
[05:34] <unnotused> imma try it tomm :P be lazy and such
[05:34] <jmarsden> I have Ubuntu and Debian server VMs under VirtualBox here on my home desktop machine... so it definitely works for me :)
[05:34] <jmarsden> OK...
[05:34] <unnotused> kk
[05:35] <unnotused> its past my bed time
[05:35] <unnotused> lol
[05:35] <jmarsden> Only 9:35pm here in California, not my bedtime yet :)
[05:35] <unnotused> im starting to get tired and i can do it fairly quckly in the morning from the office
[05:35] <unnotused> i just wanted to know if it was a virtual ubuntuserver problem
[05:35] <unnotused> and its not
[05:35] <unnotused> so im good
[05:36] <jmarsden> It really doesn't sound like one to me.
[05:36] <unnotused> im hopeing ebox will work ok
[05:36] <unnotused> ive never used it
[05:36] <jmarsden> If it manages the things you need to tweak with it, it will work.  It's not always as featureful as you might like if you have used webmin
[05:38] <unnotused> just so long as my boss can log in and change the network settings
[05:38] <unnotused> he isnt the CLI type
[05:38] <unnotused> if it doesnt have a GUI its too old
[05:38] <unnotused> lol
[05:38] <jmarsden> Why not just leave it using DHCP so he never has to change the network settings??
[05:39] <unnotused> idk
[05:39] <unnotused> he asked to be able to change the network
[05:39] <unnotused> and he is my boss
[05:39] <unnotused> one of those deals
[05:39] <jmarsden> OK...
[05:39] <unnotused> (they dont trust me)
[05:39] <unnotused> <<< student worker
[05:40] <jmarsden> If they don't trust you, they shouldn't use a server image you create... :)
[05:40] <unnotused> <<<< student work that does his boss's job and gets paid min wage for it . . .  whos the only reason his boss is still working
[05:41] <unnotused> and jmarsden, the only reason they are letting me do it is b/c im the only one who knows how to set up mysql and they are too cheep to pay for MS SQL server
[05:41] <jmarsden> Maybe you should find another minimum wage job, wait for boss to be fired, and then apply for his job :)
[05:41] <unnotused> or wait till i get done with classes and move to california and get a job
[05:41] <unnotused> :D
[05:41] <unnotused> i live in WV
[05:41] <unnotused> the middle of nowhere
[05:43] <jmarsden> OK, well, I should let you sleep... have fun with Ubuntu and vbox in the morning :)
[05:43] <unnotused> lol
[05:43] <unnotused> thats
[05:43] <unnotused> not funny
[05:43] <unnotused> XD
[05:43] <unnotused> night
[05:43] <jmarsden> Goodnight
[06:36] <uvirtbot`> New bug: #513622 in squid (main) "package squid 2.7.STABLE7-1ubuntu3 failed to install/upgrade: ??? ???? post-installation ?? ?????? 1" [Undecided,New] https://launchpad.net/bugs/513622
[08:18] <jiboumans> morning
[08:18] <acalvo> how can I change the default language?
[08:29] <jmarsden> acalvo: At what level?  For the whole server, or for one user, or one shell session, or one command?
[08:29] <acalvo> whole server
[08:29] <acalvo> or shell session
[08:30] <acalvo> whatever is easier
[08:30] <jmarsden> Install the appropriate language pack, then edit /etc/default/locale to what you want, then log in.
[08:31] <jmarsden> For a single shell session, you can just do something like export LANG=de_DE.utf8   # as the first command in the session.
[08:32] <acalvo> thank you!
[08:32] <jmarsden> You're welcome.
[09:08] <faileas> I'm running ubuntu karmic server, and using a go6 ipv6 tunnel for ipv6 connectivity. I need to start a client for the ipv6 tunnel - this is at /usr/local/gw6c/bin/gw6c before apache starts, but i'm not clear on how to set the order of how scripts in /ect/init.d start
[09:18] <JimiDini> faileas: /etc/init.d does not define order. order is defined by /etc/rcX.d where X is a digit from 0 to 6, corresponding to run-level
[09:19] <faileas> JimiDini: and within it, there's scripts starting with two numbers, a letter (either k or s) and the script name...
[09:19] <JimiDini> k=kill, s=start
[09:20] <JimiDini> number defines order
[09:20] <faileas> ok
[09:20] <JimiDini> S20 will start before S30 for example
[09:21] <faileas> now i just need to work out why this app won't start unless i'm in its directory
[09:32] <twb> By now sysvinit should be deprecated :-/
[09:34] <faileas> lol
[09:34] <faileas> well i think my current issues arn't with that
[10:07] <freewillie> Heey
[10:08] <freewillie> Could someone help me out with my sound problem?
[10:09] <freewillie> I have installed ubuntu server, and i want to use it as a jukebox but i hear no sound
[10:15] <freewillie> Is there someone who can help me ?
[10:16] <faileas> twb: What else should i be using? ;p
[10:16] <twb> upstart
[10:16] <faileas> freewillie: got alsa installed?
[10:16] <twb> Or in my fantasy land, metainit
[10:16] <freewillie> faileas: yes
[10:17] <faileas> freewillie: go to alsamixer and see if your volume is turned up
[10:17] <freewillie> faileas: but not configured because i dont know how
[10:17] <faileas> (its not a joke. its happened to me before)
[10:17] <freewillie> faileas: how can i put the volume up
[10:17] <faileas> freewillie: in general you shouldn't really need to. try going to alsamixer (in cli)
[10:18] <faileas> (i'm assuming you're running cli. else it depends on your setup)
[10:18] <freewillie> i run from the commandline
[10:18] <freewillie> faileas: This is the error: alsamixer: function snd_ctl_open failed for default: No such file or directory
[10:19] <incorrect> strongswan or openswan?
[10:20] <faileas> freewillie: http://ubuntuforums.org/showthread.php?t=1094196 might be a server kernel issue
[10:20] <faileas> freewillie: mind installing a generic kernel, and seeing if that works?
[10:22] <freewillie> ok
[10:22] <freewillie> thanks
[10:22] <freewillie> faileas: I will try, but now i have to go
[10:46] <uvirtbot`> New bug: #513427 in samba (main) "'net usershare' returned error 255: net usershare add: cannot convert name "Everyone" to a SID. Invalid parameter." [Low,Incomplete] https://launchpad.net/bugs/513427
[10:46] <nakamuka> scp a.c host:welcome , and it should copy to   /myscps/project/welcome directory in to the server. instead of /home/username/welcome .
[10:47] <nakamuka> is it possible to change the prefix directory in the server ?
[10:47] <nakamuka> could someone help me ?
[10:47] <_ruben> incorrect: openswan here, but that was a decision made quite some time ago .. both have their pros and cons
[10:48] <_ruben> nakamuka: change the users homedir, or specify full path in your scp command
[10:48] <incorrect> _ruben, i went with openswan too
[10:49] <nakamuka> _ruben,  but are there any way to specifiy like in ftp we specificy vsftp.conf anon_root = prefixdir, ?
[10:50] <_ruben> nakamuka: doubt that, check the manual to be sure :)
[10:59] <twb> nakamuka: what are you actually trying to achieve?
[11:01] <nakamuka> twb, Ok, i need to copy some file to server through scp. by default if destination directory is relative path, then it is prefixed with the home directory to make absolute.  But for me i dont want home directory to be prefixed but some other directory lets say /myscps/projects1
[11:02] <twb> nakamuka: why?
[11:02] <nakamuka> twb, the present code scenario is like that. :)
[11:03] <nakamuka> twb, reason behind is the destination directory can be changed quickley just by chaning a small configuration file , ie. changing prefix.
[11:03] <jmarsden> nakamuka: ssh user@example.com "ln -s scp /myscps/projects1"  # then scp -p foo user@example.com:scp/  puts the files where you want them.
[11:04] <jmarsden> To change where they go, change the scp symlink .
[11:04] <twb> Or even just ssh fs 'ln -s /myscps/*'
[11:06] <jmarsden> twb: Maybe.  That could be a lot of symlinks to change for every user concerned, if the whole idea is to allow rapid switching of where stuff goes?
[11:06] <twb> jmarsden: I was assuming he was only doing it for his own silly user
[11:06] <twb> It certainly will make a mess of $HOME
[11:08] <nakamuka> twb, that is true, but  it could be nice if there is one option like we see in ftp,  In   vsftp.conf  anon_root parameter holds the prefix directory.
[11:08] <twb> nakamuka: file a feature request with upstream
[11:09] <nakamuka> twb, where do i have to file. link ?.. sorry to ask basic questions.!
[11:09] <twb> Try #openssh, if you're using the default implementation.
[11:57] <mobi-sheep> You too could throw in the aliases for SCP commands in the mix!
[11:59] <nakamuka> mobi-sheep,  means ?
[12:01] <mobi-sheep> nakamuka: That wouldn't work, I think. Because of specific parameters.  Scripts would though.
[12:02] <nakamuka> mobi-sheep, ok. any idea about ChrootDirectory in sshd_conf directory ?
[12:02] <mobi-sheep> nakamuka: I mean like this "scp-project2 <file>" would act like "scp <file> <hostname>@<ip>:<path>"
[12:02] <mobi-sheep> Assuming you're doing this for yourself, not many users.
[12:03] <mobi-sheep> nakamuka: I know nothing about that ChrootDirectory.
[13:13] <Italian_Plumber> good morning.  I just put a 100GB hard drive in my machine.  It shows 93GiB size, and 88GiB free.  I know that 100GB is actually 93GiB -- I totally get that.  And I know that the filesystem needs some room to do its thing.  But does it really need 5GiB out of 93?
[13:15] <Italian_Plumber> I'm farily sure that I used all availiable space for the partition.
[13:17] <Italian_Plumber> the drive is not the OS drive.
[13:21] <_ruben> Italian_Plumber:
[13:21] <_ruben>        -m reserved-blocks-percentage
[13:21] <_ruben>               Specify  the  percentage of the filesystem blocks reserved for the super-user.  This avoids fragmentation, and allows root-owned daemons, such as syslogd(8), to continue to function correctly after
[13:21] <_ruben>               non-privileged processes are prevented from writing to the filesystem.  The default percentage is 5%.
[13:21] <_ruben> 5% of 100G is 5G ;)
[13:23] <Italian_Plumber> ah.  well avoiding fragmentation will be good -- this drive will be written to and deleted from a lot.  Thanks!
[13:24] <zul> morning
[14:00] <zul> jiboumans: ping
[14:01] <jiboumans> zul: hi
[14:01] <zul> jiboumans: 1:1?
[14:01] <jiboumans> was just about to hit 'call' :)
[14:01] <zul> its my psychic powers
[14:16] <jiboumans> kirkland: ping?
[14:18] <MTecknology> You guys have any idea how to make a cron job that runs every hour except midnight?
[14:20] <MTecknology> * 1-23 * * * ; hrm.. I think that'll work :)
[14:22] <Pici> MTecknology: That will run every minute of every hour except 0:xx
[14:23] <Pici> MTecknology: 0 1-23 * * * would be what you're looking for
[14:23] <MTecknology> Pici: thanks, I spaced when I typed it out here
[14:27] <smoser> EtienneG, per -server mailing list
[14:27] <smoser> ii'm certainly not under the impression that -virtual is being phased out
[14:27] <EtienneG> smoser, ok, sorry for implying that then
[14:28] <smoser> the wierd thing about that snippit of the 'earth computing' thread is that -virtual *is* -generic-pae or -server
[14:28] <EtienneG> smoser, I see
[14:28] <smoser> its just that kernel, those modules, with some pruning of the module list
[14:28] <smoser> it makes no sense that there woudl be a performance difference
[14:28] <EtienneG> indeed
[14:28] <smoser> tha tis the case in karmic and lucid. i can't knowledgably speak for before that.
[14:29] <EtienneG> i am perplexed too
[14:38] <beniwtv> Hey all.... My newly installed server on RAID 1 wants to install LILO. How can I install GRUB instead? On 8.04.3. I read somewhere this is only grub-installl /dev/md0 but on the alternate installer in live mode I can't find this command :(
[14:39] <_ruben> beniwtv: is /boot on a lvm by any chance?
[14:40] <beniwtv> _ruben: no, only RAID1
[14:43] <beniwtv> _ruben: The problem with LILO is, it won't boot and it stalls on freeing initrd
[14:46] <_ruben> hmm .. wonder it thinks lilo is needed in that case, plain software raid is no reason to ditch grub
[14:48] <beniwtv> _ruben: Yeah, now I'm going to do this: boot in rescue mode from the alternate CD, activate LVM (if not active), and mount the /root partition (which is on LVM). Then I chroot, apt-get grub and install grub. Sounds possible right?
[14:49] <_ruben> i'd say so yeah
[14:50] <beniwtv> _ruben: I have someone there that will be able to follow some instructions, but it's difficult over the phone. I have ping to the server. Would is be possible to access it via ssh?
[14:52] <_ruben> the alternate cd probably wont have sshd running, but it can be installed manually (done similar stuff with the live desktop cd for recovery)
[14:52] <beniwtv> _ruben: how can it be installed? anna-install?
[14:54] <_ruben> not sure about the possibilities offered by the alternate cd, in my case i could just apt-get it
[14:54] <beniwtv> apt-get seems to be not there, though
[14:55] <_ruben> perhaps the network-console module could do the job
[14:56] <_ruben> anna/choose_modules network-console
[14:56] <_ruben> not sure if that works for recovery purposes though
[14:57] <beniwtv> probably I should do a ls /bin and see what's there ot use
[14:57] <beniwtv> s/ot/to
[15:11] <uvirtbot`> New bug: #513811 in bacula (main) "package bacula-director-mysql 3.0.2-3ubuntu2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/513811
[15:15] <incorrect> damn setting up L2TP/ipsec is a pain in the arse
[15:18] <beniwtv> how does someone use grub-installer manually? is it possible?
[15:22] <smoser> kirkland, mathiaz, ttx you have a euca instance up you can do something for me ?
[15:23] <smoser> python -c 'import boto.utils, pprint; pprint.pprint(boto.utils.get_instance_metadata())'
[15:26] <mathiaz> smoser: not for now sorry
[15:27] <ttx> smoser: ack
[15:28] <ttx> smoser: that seems to hang
[15:29] <ttx> smoser: you need an instance running ?
[15:29] <smoser> it shoudln't hang
[15:29] <smoser> yeah, in an instance do that.
[15:29] <smoser> sorry
[15:29] <ttx> ah
[15:30] <smoser> it will hang for a couple days if you didn't have a web server on 169.254.169.254
[15:30] <ttx> right, just a sec, then
[15:30] <ttx> smoser: good presentation yesterday, btw
[15:30] <ttx> raed the logs
[15:31] <smoser> why thank you.
[15:34] <ttx> smoser: http://pastebin.ubuntu.com/364633/
[15:34] <smoser>  'block-device-mapping': {'emi': 'sda1',
[15:34] <smoser>                           'ephemeral': 'sda2',
[15:34] <smoser>                           'root': '/dev/sda1',
[15:34] <smoser>                           'swap': 'sda3'}
[15:34] <smoser> gah
[15:34] <smoser> should i open a bug on that
[15:34] <smoser> ec2 has
[15:34] <smoser> 'ephemeral0': 'sda2'
[15:34] <smoser> (same other than '0' in ephemeral)
[15:36] <uvirtbot`> New bug: #513749 in dhcp3 (main) "dhcp3-server automatically starts ignoring startup rules on package update" [Low,Incomplete] https://launchpad.net/bugs/513749
[15:46] <ttx> smoser: yes, file one
[15:47] <smoser> hm... could you run 'ubuntu-bug eucalyptus' for me and pass me the link ?
[15:47] <smoser> that way we'll get the euc versions
[15:47] <smoser> apport hoosk
[15:49] <smoser> ttx,
[15:50] <ttx> 1.6.2~bzr1136-0ubuntu1
[15:50] <ttx> (don't have apport on that box, sorry)
[15:50] <ttx> (I know, I should)
[15:54] <ttx> smoser: let me know the bug number when filed
[15:54] <smoser> ttx, you can get it pretty easily: apt-get install apport-cli
[15:54] <smoser> :)
[15:55] <ttx> meh, kinda busy right now, trying to scrub 60 bugs in 4 min
[15:55] <smoser> fair.
[15:59]  * patdk-lap watchings the bugs scurry away from ttx, hoping to live
[15:59] <ahasenack> mathiaz: hey
[15:59] <ahasenack> mathiaz: when using puppet with ec2, how do you manage the certificates to bootstrap puppet in a script?
[15:59] <ahasenack> mathiaz: do you use autosign on the puppet master?
[15:59] <mathiaz> ahasenack: aha - good question
[16:00] <mathiaz> ahasenack: I'm about to investigate that issue
[16:00] <mathiaz> ahasenack: current plan is to allow autosigning
[16:00] <ahasenack> mathiaz: there are some alternatives, not all extremely scriptable
[16:00] <ahasenack> mathiaz: like pre-generating the certs on the master, or actually doing all this work in a script talking to two hosts
[16:01] <mathiaz> ahasenack: alternative is to provide the private key and publick key in the user data
[16:01] <ahasenack> mhm
[16:01] <mathiaz> ahasenack: IMO autosigning is the best option
[16:01] <mathiaz> ahasenack: what I need to check if wether the autosigning process can have hooks
[16:01] <ttx> mathiaz: I've been struggling with the pxe setup today :)
[16:02] <ttx> mathiaz: now I know that lighttpd cgi module doesn't pass PATH_INFO.
[16:02] <mathiaz> ahasenack: with the idea that before signing a cert request, it calls an external script to check wether the certname requested should be accepted
[16:02] <ttx> mathiaz: and I have patches for boot.py error reporting
[16:02] <smoser> ttx bug 513842
[16:02] <uvirtbot`> Launchpad bug 513842 in eucalyptus "block-device-mapping in metadata has 'ephemeral' not 'ephemeral0'" [Medium,New] https://launchpad.net/bugs/513842
[16:02] <ahasenack> mathiaz: I don't remember seeing that feature, but it might be deep inside the ruby things
[16:02] <ttx> since it fails quite badly when something happens
[16:02] <mathiaz> ttx: hm - I did my test with apache2 - I though that PATH_INFO was a standard CGI variable
[16:02] <ttx> mathiaz: it is in my book too
[16:03] <mathiaz> ahasenack: yeah -  I'm not sure if puppetca supports that
[16:03] <mathiaz> ahasenack: If not, I'll talk to upstream about implementing it
[16:03] <ahasenack> mathiaz: the idea about using user data sounds interesting
[16:03] <ttx> mathiaz: I'll switch to apache2, no big deal
[16:03] <ahasenack> mathiaz: I was hoping I could come up with a more general solution that didn't leverage on ec2, but it does sound nice
[16:03] <mathiaz> ahasenack: one of the hook is to verify if a node has already been allocated
[16:04] <mathiaz> ahasenack: ie if the puppet knows about the client and the requested name
[16:04] <mathiaz> ahasenack: if there isn't any hook available in the signing process, then I'd suggest auto signing
[16:05] <ahasenack> mathiaz: for ec2 that doesn't sound too bad, since the network setup will only allow machines you specify to reach the master
[16:05] <mathiaz> ahasenack: yop
[16:05] <ttx> mathiaz: pxe chaining is a little tricky to debug... especially running dnsmasq on a linksys router.
[16:05] <mathiaz> ahasenack: the next problem is about node allocation
[16:05] <ahasenack> mathiaz: I have thought about that yesterday
[16:06] <ahasenack> mathiaz: you mean how to tell which node will become what?
[16:06] <mathiaz> ttx: don't you see the tftp request in the log?
[16:06] <mathiaz> ahasenack: yes
[16:06] <ahasenack> mathiaz: I think I have a solution
[16:06] <mathiaz> ahasenack: I've got *plenty* of solutions ;)
[16:06] <ahasenack> mathiaz: if you export a shell var like FACTER_something in the node
[16:06] <ahasenack> mathiaz: you can use $something on the master as the contents of that var
[16:06] <mathiaz> ahasenack: right - that's an option
[16:06] <ttx> mathiaz: tftp is alright... it's the chaining that fails (get the gpxe option, point to http://)
[16:06] <ahasenack> mathiaz: so I was planing in using that in the node statement on the master
[16:06] <mathiaz> ahasenack: another on is to the hostname
[16:07] <mathiaz> ahasenack: actually the certname
[16:07] <mathiaz> ahasenack: webserver.test1.
[16:07] <ahasenack> mathiaz: what is the certname?
[16:07] <mathiaz> ahasenack: ^^ I'm a webserver in test1
[16:07] <ahasenack> isn't that just derived from the hostname?
[16:07] <mathiaz> ahasenack: not necessarly :)
[16:08] <mathiaz> ahasenack: you can set a specific certname when starting puppet
[16:08] <mathiaz> ahasenack: that helps in EC2 since ec2-init sets the hostname of the machine to what dhcp/metadata says
[16:08] <ahasenack> mathiaz: ok, but I suppose you can only have one certname in all your nodes
[16:08] <mathiaz> ttx: hm - what do you mean? what fails on the client?
[16:08] <ahasenack> mathiaz: I mean, it should be unique among the nodes
[16:09] <mathiaz> ahasenack: yes - hence the following trick: webserver.test1.%InstanceID
[16:09] <ttx> mathiaz: on a call
[16:09] <ahasenack> mathiaz: ok
[16:09] <mathiaz> ahasenack: and then you match on webserver.test1 in your puppet recipe
[16:09] <ttx> mathiaz: bbl
[16:09] <ahasenack> mathiaz: I was thinking about the node just saying "hey, I'm a node of type <foo>"
[16:09] <mathiaz> ttx: ok
[16:10] <mathiaz> ahasenack: right - that's what you're aiming for
[16:10] <ahasenack> mathiaz: then on the server I wouldn't need to include all the hostnames (or certnames) in the configuration
[16:10] <ahasenack> mathiaz: like, FACTER_role="ldap-slave"
[16:10] <mathiaz> ahasenack: yes - that's what a good way to do it
[16:10] <ahasenack> makes it easier if I can have several machines with the same role
[16:11] <mathiaz> ahasenack: yop - I'd try that - I know some puppet user are exactly doing that
[16:11] <ahasenack> mathiaz: you said "match on webserver.test1", I suppose that requires puppet with regexp support, right?
[16:11] <mathiaz> ahasenack: they don't have any nodes defined, just use facts
[16:11] <ahasenack> mathiaz: I read somewhere that  only newer versions have that regexp support (i.e., not the one from hardy )
[16:11] <ahasenack> mathiaz: interesting
[16:11] <mathiaz> ahasenack: yeah - probably something like that - I'm not sure if it actually works
[16:12] <ahasenack> the one from hardy can't even check for variable contents in the if/else statements
[16:12] <ahasenack> I had to resort to a case statement for that
[16:12] <mathiaz> ahasenack: the idea is to use the dns naming scheme to state the role of the machine
[16:12] <mathiaz> ahasenack: yeah - hardy is probably very old :)
[16:12] <mathiaz> ahasenack: you should try the one in lucid
[16:12] <mathiaz> ahasenack: :)
[16:12] <ahasenack> indeed
[16:13] <mathiaz> ahasenack: using something like: webserver.pool1.prod
[16:13] <mathiaz> ahasenack: to state that this is a webserver in pool1 in production
[16:13] <ahasenack> right
[16:13] <mathiaz> ahasenack: that being said you could also use facts
[16:13] <ahasenack> poor dns
[16:13] <mathiaz> ahasenack: but you need to get the complete role description into the node somehow
[16:13] <ahasenack> I try to avoid messing with dns when certificates are in place
[16:14] <ahasenack> mathiaz: right, if I can match on just "webserver.pool", that works just fine
[16:14] <ahasenack> otherwise I will have to list all poolN possibilities if I want to grow
[16:14] <mathiaz> ahasenack: well - the problem with using the hostname in ec2 is that there may be a chance that the hostname will be reused
[16:14] <mathiaz> ahasenack: overtime
[16:15] <ahasenack> mathiaz: what do you mean?
[16:15] <mathiaz> ahasenack: thus the idea of including the InstanceId in the certificate name
[16:15] <mathiaz> ahasenack: well if you start/stop a lot of instances, you may get the same IP address again
[16:15] <mathiaz> ahasenack: in which case you'd end up with the same hostname
[16:15] <ahasenack> hmm
[16:16] <mathiaz> ahasenack: so you can't get the same hostname in *running* instances
[16:16] <mathiaz> ahasenack: but you could get the same hostname over a long period of time
[16:16] <mathiaz> ahasenack: that's why you need to have a real unique ID for each machine
[16:16] <mathiaz> ahasenack: InstanceID can play that role
[16:17] <ahasenack> mathiaz: do you plan to have two types of AMIs perhaps? One for the master and the other being a sort of generic one for whatever you want, just enough to bootstrap puppet and then let it configure the rest?
[16:17] <mathiaz> smoser: what's a standard hostname in an ec2 machine?
[16:17] <ahasenack> ec2-foo-bar-bla.internal?
[16:17] <ryker> is there any way at all to get to an instance that is stuck in startup because of a typo in /etc/fstab for an unimportant mount point?
[16:18] <mathiaz> ahasenack: hm - I don't think we should two AMIs. My current plan is to document how to bootstrap a puppetmaster (via a puppet recipe)
[16:18] <smoser> $ hostname
[16:18] <smoser> domU-12-31-38-00-45-05
[16:18] <mathiaz> ahasenack: may be provide a package in >lucid
[16:18] <ahasenack> mathiaz: create a puppetmaster with puppet?
[16:18] <mathiaz> smoser: is the hostname unique in ec2 overtime?
[16:18] <mathiaz> ahasenack: yes :)
[16:18] <smoser> no
[16:18] <ahasenack> mathiaz: so it starts as a puppet node and then becomes a master?
[16:18] <smoser> external fqdn: ec2-72-44-32-161.compute-1.amazonaws.com
[16:18] <mathiaz> ahasenack: you can run puppet in a standalone mode
[16:19] <smoser> internal: domU-12-31-38-00-45-05.compute-1.internal
[16:19] <mathiaz> ahasenack: yes - you can ship complete puppet recipe and ask puppet to apply to the local system
[16:19] <mathiaz> ahasenack: that's how I would bootstrap a master
[16:19] <smoser> i guess i dont know for certain that the hostname is not globally unique.
[16:19] <mathiaz> smoser: ok - so to make the instance name unique, including it's IntanceID is enough?
[16:20] <mathiaz> smoser: what matters here is wether the hostname is unique *overtime*
[16:20] <mathiaz> smoser: over time
[16:20] <smoser> instance-id is enough
[16:21] <smoser> after i pated that internal, i checked. the mac on that machine is
[16:21] <smoser> 12:31:38:00:45:05
[16:21] <mathiaz> ahasenack: ^^ so using the instance-id is enough
[16:21] <smoser> so they're just using that.
[16:21] <ahasenack> mathiaz: if your master is not long-running, this uniqueness should not be too much of a problem, right?
[16:21] <mathiaz> ahasenack: and then with a hook in puppetca you can check wether the instance id does actually exist before signing the certificate
[16:21] <smoser> the obvious fact is that i-5e222c36 cannot be eternally globally unique
[16:21] <mathiaz> ahasenack: agreed
[16:21] <smoser> it is probably not cross-region unique
[16:22] <mathiaz> ahasenack: for test environement it's ok
[16:22] <ahasenack> mathiaz: hmm, we do something similar in landscape
[16:22] <mathiaz> ahasenack: as you can also purge certificates on the master
[16:22] <ahasenack> mathiaz: the client can auto-register when in cloud mode
[16:22] <mathiaz> ahasenack: basically the puppet master will *not* sign a request for a cert that has already been issued
[16:22] <ahasenack> mathiaz: because the server is able to verify it via the ec2 api
[16:22] <ahasenack> so
[16:23] <ahasenack> I was planning in just launching an instance with a user data script. One for a puppet master, one for puppet nodes
[16:23] <patdk-lap> ttx, how did you test lighttpd that you say it doesn't do path_info? cause mine do
[16:23] <ahasenack> then the certs gave me that headache
[16:24] <ahasenack> but in ec2 autosign sounds reasonable
[16:24] <mathiaz> ahasenack: yes - especially in test environement
[16:24] <mathiaz> ahasenack: I'd argue to use the autosign in production env as well
[16:25] <mathiaz> ahasenack: especially if you use things like AutoScaling
[16:25] <ahasenack> it's a policy decision, it's good to have the option
[16:25] <mathiaz> ahasenack: but you'd need a hook to verify that the instance actually exists
[16:25] <ahasenack> certainly when managing hundreds of machines you can't have manual signing
[16:25] <ttx> patdk-lap: enabled cgi mod, ran a python script, os.environ["PATH_INFO"] triggers KeyError
[16:25] <mathiaz> ttx: did you try to dump of os.environ?
[16:25] <patdk-lap> http://mx1.grsi.com/test.pl/test :)
[16:26] <mathiaz> ttx: that's usually how I check what's in there
[16:26] <ahasenack> mathiaz: when landscape starts an instance, it gives it an OTP via user data
[16:26] <patdk-lap> it shows path_info for me
[16:26] <ahasenack> mathiaz: so when the client later gets back to the server with that otp, we can verify it
[16:26] <ahasenack> (one time password)
[16:26] <mathiaz> ahasenack: yop - that standard keybased process - with a timestamp it's even better
[16:27] <mathiaz> ahasenack: and then you sign everything with a server key
[16:27] <mathiaz> ahasenack: so that you don't have to keep track of issued tokens
[16:27] <mathiaz> ahasenack: on the server
[16:27] <ahasenack> timestamp + replay detection for the duration of the timestamp, that's how kerberos detects these attacks, right?
[16:28] <mathiaz> ahasenack: yop - I think so
[16:28] <ttx> patdk-lap: I'll give it another try :)
[16:28] <ahasenack> mathiaz: ok, thanks for the discussion! :)
[16:28] <patdk-lap> sorry, lighttpd is my baby :)
[16:28] <patdk-lap> well, not mine, but I have been helping with it for awhile
[16:28] <mathiaz> ahasenack: you're welcome - let me know what you find out
[16:29] <mathiaz> ahasenack: I'm about to right an upstart job to include in ec2-init to enable puppet integration
[16:29] <mathiaz> ahasenack: so that you don't have to use a user script anymore
[16:29] <ahasenack> mathiaz: ec2-init fires other initscripts ("upstart jobs") then?
[16:30] <mathiaz> ahasenack: kind of
[16:30] <mathiaz> ahasenack: ec2-init will emit a cloud-config upstart event
[16:30] <ahasenack> mathiaz: or is it just another job and ec2-init depends on it to complete?
[16:30] <ahasenack> ok
[16:30] <mathiaz> ahasenack: and then the puppet upstart job starts on cloud-config
[16:31] <mathiaz> ahasenack: the key part is how to pass the role information to instance via user-data
[16:32] <ahasenack> mathiaz: so your job will take that information out of user-data and apply it to its local puppet client configuration, mangling, say, the certname?
[16:33] <mathiaz> ahasenack: hm I don't know yet
[16:33] <mathiaz> ahasenack: I'm leaning toward the following:
[16:34] <mathiaz> ahasenack: 1. you can specify a puppet certname in the userdata (ex: webserv.%i.%h.%d)
[16:34] <mathiaz> ahasenack: with %i substituted with InstanceID
[16:34] <ahasenack> ok
[16:34] <mathiaz> ahasenack: %h for the hostname, and %d for the domain name
[16:34] <mathiaz> ahasenack: if it's there
[16:35] <mathiaz> ahasenack: 2. pass facts in the user data
[16:35] <mathiaz> ahasenack: and these facts would then be availabe
[16:36] <mathiaz> ahasenack: in that case, the certname would default to the InstanceID
[16:36] <ahasenack> mathiaz: can you "take over" user-data like that? I suppose yes, since the instance will be launched by you
[16:36] <ahasenack> mathiaz: just wondering if other scripts elsewhere don't expect to have their own stuff in user-data
[16:36] <mathiaz> ahasenack: yes - we already have a specific format for doing that
[16:36] <ahasenack> mathiaz: the bit from eric hammond or has it evolved?
[16:36] <mathiaz> ahasenack: it has eveloved
[16:37] <ahasenack> mathiaz: #!/ in first line versus plain data?
[16:37] <ahasenack> mathiaz: ah, interesting
[16:37] <mathiaz> ahasenack: yes - it's based on that
[16:37] <mathiaz> ahasenack: IIRC if user-data starts with #cloud-config then you'll trigger the ec2-init simplified configuration
[16:37] <mathiaz> smoser: ^^?
[16:37] <mathiaz> ahasenack: and then one option of the simplified configuration will be to use puppet
[16:38] <mathiaz> ahasenack: see https://wiki.ubuntu.com/ServerLucidCloudConfig for a sample configuration file
[16:38] <ahasenack> mathiaz: where do these discussions take place? ubuntu-server? I don't remember seeing them, and they are very interesting
[16:38] <smoser> this is correct.
[16:38] <ahasenack> maybe just uds?
[16:38] <mathiaz> ahasenack: we talked about them at UDS
[16:38] <ahasenack> ah, ok
[16:39] <mathiaz> ahasenack: and then refined them in #ubuntu-server in december
[16:39] <mathiaz> ahasenack: and I also made an RFC on planet and ubuntu-server@
[16:40] <mathiaz> ahasenack: so the #cloud-config file format will be extend to support puppet specific options
[16:40] <mathiaz> ahasenack: such as puppet certname and puppet facts
[16:40] <mathiaz> ahasenack: to support both 1. and 2. use cases outlined above
[16:40] <squidly> with the UCE cloud how have I tell if my nodes are connecting to my main server?
[16:41] <mathiaz> ahasenack: this is part of https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-puppet-uec-ec2-integration
[16:41] <ahasenack> sounds very interesting
[16:51] <ttx> patdk-lap: alright, path_info works... something else is borking it
[16:51] <ttx> mathiaz: gpxe doesn't seem too happy with the pxelinux.0 it gets a 302 for
[16:51] <ttx> mathiaz: i'll spend more time on it tomorrow
[16:59] <ttx> mathiaz: PXE input/output error when going through boot.py/pxelinux.0 302, works when loading /pxelinux.0 directly
[17:03] <kirkland> smoser: monkeying with the uec image names again?
[17:03]  * kirkland goes update the wiki pages
[17:03] <smoser> no
[17:04] <kirkland> smoser: looks like -uec- was added in there
[17:04] <smoser> yeah.
[17:04] <smoser> that happened prior to alpha2
[17:04] <smoser> as i ahd removed the -uec-
[17:05] <smoser> and that caused pain for people as it name-collided with 'ubuntu-desktop'
[17:05] <smoser> so this is hopefully more final
[17:05] <smoser> sorry
[17:08] <kirkland> smoser: i have a euca instance running now
[17:08] <kirkland> smoser: what would you like
[17:09] <smoser> ttx got it for me. thanks though.
[17:12] <beniwtv> hey anyone has a problem where the recovery menu appears on shutdown?
[17:12] <beniwtv> (server edition, that is)
[17:49] <Elad> I just added an alias onto my server so that it forwards all root mail to my e-mail, and I am getting this e-mail every 20 minutes but don't see a cron for it under the root user .. /usr/share/sendmail/sendmail: 1248: /usr/sbin/sendmail-msp: not found
[17:54] <Elad> as some background, I had originally set up the server to use sendmail, but then switched to postfix - Can I just use apt to remove sendmail or do I risk breaking postfix in some strange way
[17:55] <lamont> see also /etc/cron*/*?
[17:56] <lamont> what does 'dpkg -S /usr/share/sendmail/sendmail' say?
[17:59] <RoyK> wtf. it's 2010 and people are still using sendmail???
[18:08] <mjeanson> kirkland: is it possible to have plymouth working ina kvm guest?
[18:20] <ryker> hate to ask a simple question like this, but what I've found online doesn't work for me.  How do I make a new hostname stick after a reboot?  Putting the name in /etc/hostname doesn't work.
[18:29] <Elad> lamont: sendmail-base: /usr/share/sendmail/sendmail
[18:31] <lamont> so something in that depends: sendmail, and fails to do so in the package, it would appear.
[18:31] <lamont> since sendmail-msp isn't in sendmail-base...
[18:35] <skot> Hi all
[18:35] <skot> I'm looking for a little help with a failed soft-raid array (ubuntu-9.10 -2.6.31-17 kernel- raid5 5-disks). I have got a replacement drive (A) but on adding it, another drive (B) is failing during the sync. It gets about 60% and then the new drive (B) gets marked faulty and the sync stops. Any suggestions?
[18:36] <bogeyd6> skot, is the drive faulty?
[18:37] <skot> yes. it seems to be. smartctl reports errors in the log.
[18:44] <skot> bogeyd6, the only thing I can think to do is try to make a byte-by-byte copy of the failing drive (B) to (A) (minus any corrupt data; if the disk will stay online long enough) and then try to recreate the array with (A) substituted in for (B). Can you think of another solution?
[18:45] <bogeyd6> I dont think so
[18:45] <bogeyd6> The drive has errors and you are trying to use it
[18:46] <Elad> lamont: think I should just be able to remove sendmail, and if it breaks something just put it back? I don't think anything relies on it . . .but could be wrong
[18:46] <uvirtbot`> New bug: #509734 in libnss-ldap (main) "initgroups() fails when using libnss-ldap (but not nscd)" [Medium,Triaged] https://launchpad.net/bugs/509734
[18:46] <incorrect> damn phpldapadmin is b0rk3d
[18:49] <skot> Yeah, unfortunately I have had a failure of 2 drive in a raid5 set. I don't what else to do but try to recover as many bytes as I can.
[18:50] <incorrect> skot, you sound kinda screwed
[18:51] <skot> incorrect, that is not so funny cause that is what I feel like.
[18:51] <skot> ;(
[18:51] <incorrect> backups!
[18:51] <incorrect> saying that i need to make backups
[18:53] <skot> I hear you, I have backups (mostly) but putting them together is no fun; and I will never be sure I have it all back the way it was...
[18:53] <incorrect> skot, these days i virtualise everything
[18:53] <skot> For the most part, that raid array was my backup of many other sources.
[18:53] <zul> yay samba has an apport hook
[18:54] <skot> incorrect, how do you virtualize disks? Somewhere there are going to be real drives. If you lose enough, you lose data.
[18:55] <incorrect> virtual disks, you snap shot those somewhere else
[18:56] <incorrect> ok it can take a bit of space
[18:56] <skot> incorrect, I should mention that this is my home system. I know is just a matter of money, I should have thrown N + 1 + 1 drives at the situation but I thought the odds on two disks going in the same week was low enough.
[18:56] <incorrect> this is my home system too
[18:57] <skot> incorrect, well, I got bit this time. Thanks for listening. I will go off and recover what I can.
[18:57] <incorrect> good luck
[19:00] <kirkland> mjeanson: sorry, I don't know anything about Plymouth
[19:03] <mjeanson> kirkland: I'm interrested in the boot messages and plymouth seems to be mandatory in lucid to display them
[19:05] <kirkland> mjeanson: why would plymouth not work in KVM?
[19:05] <kirkland> mjeanson: how can I tell if plymouth is working?
[19:06] <mjeanson> kirkland: never seen it working in any of my kvm, but you should get a graphical bootsplash and service startup messages
[19:06] <erichammond> mathiaz, smoser: The EC2 instance id is unique over all time.  Amazon will be changing the format/length of the instance ids before they run out.  I'm pretty sure instance ids are unique across regions, but have not confirmed this.  Host names are not unique over time.
[19:07] <smoser> hm.. i thought that instance ids would not be unique over regions
[19:10] <erichammond> smoser: I suspect that the id includes an id or range assigned to the authority giving it out so that even a single region does not have a single source.  I'm not positive about this.  There was an article which took apart instance ids in an attempt to figure out how many were being started each day.
[19:11] <erichammond> smoser, mathiaz: domU-12-31-38-00-45-05 is only one of a couple possible formats for internal hostnames for EC2 instances.  There is a different format based on the internal IP address.  Each instance only gets one of these, but it is somewhat arbitrary which one.
[19:12] <erichammond> EC2 also does not make any promises about hostname formats, internal or external.
[19:15] <mjeanson> kirkland: you're my kvm guru, I thought you may of ran into that but thanks anyway
[19:18] <patdk-lap> skot, make sure you try dd_rescue :)
[19:21] <incorrect> stupid installer created my extended partition the same size as a single parition
[19:21] <incorrect> urg
[19:22] <incorrect> what is the easiest way to scale a partition from the command prompt?
[19:28] <incorrect> ok that was easy
[19:31] <incorrect> how can i refresh /dev i am missing a disk
[19:31] <incorrect> err partition
[19:32] <kirkland> mjeanson: hmm, i just boot a lucid vm
[19:32] <kirkland> mjeanson: i didn't see any boot messages
[19:32] <kirkland> mjeanson: but it booted in about 4 seconds though
[19:32] <smoser> incorrect, try udevadm trigger
[19:33] <incorrect> didn't do it
[19:38] <incorrect> reboot ftw
[19:59] <RoyK> incorrect: as in reboot For The Windows-user?
[20:06] <squidly> how does a UCE system determine if there a enough resources for the vm?
[20:07] <lamont> Elad: if you have postfix installed, you don't need any of sendmail (and it shouldn't need any of itself either)
[20:53] <Italian_Plumber> I have this motherboard: http://tinyurl.com/ydtdjg6 running hardy server.  I have bought this PCI card: http://tinyurl.com/ydv29tc and this hard drive: http://tinyurl.com/yknn23p ... will I be able to put in the card and drive, and have Ubuntu recognize the device, or will it be likely that I'll need to install drivers?
[20:55] <patdk-lap> it will be fine
[20:58] <mathiaz> jjohansen: hi - is the cciss compiled in the -generic kernel?
[20:59] <mathiaz> jjohansen: I'm trying to install lucid on an HP server - these machines usually use the cciss module
[20:59] <MTecknology> This is irritating... I have a WinXP 64bit key but I can't use it in a VM beecause my BIOS won;t let me
[20:59] <mathiaz> jjohansen: but the installer is not able to detect any drive
[20:59] <patdk-lap> MTecknology, then you don't have a real xp key, but an oem key
[21:00] <MTecknology> patdk-lap: i have the real key on the pressed cd in my posession
[21:00] <patdk-lap> that doesn't matter
[21:01] <patdk-lap> where did the cd come from? bought at the store in a windows xp box?
[21:01] <patdk-lap> or did it come with a computer?
[21:01] <MTecknology> patdk-lap: yup
[21:01] <patdk-lap> with a computer, it's no good
[21:01] <MTecknology> store
[21:01] <jjohansen> mathiaz: cciss? I am not familiar with it but I don't find it as an option
[21:01] <MTecknology> you don't seem to know what you're talking about..
[21:01] <patdk-lap> I don't?
[21:03] <Italian_Plumber> sorry I had to get up from my desk.  patdk-lap: was your "it will be fine" directed at me?
[21:03] <mathiaz> jjohansen: http://cciss.sourceforge.net/
[21:04] <patdk-lap> Italian_Plumber, yep, you shouldn't have any issues at all
[21:04] <jjohansen> mathiaz: we have CONFIG_BLK_CPQ_CISS_DA=m
[21:04] <jjohansen> CONFIG_CISS_SCSI_TAPE=y
[21:04] <Italian_Plumber> ok... I"m going to to hold you to that. :)
[21:04] <patdk-lap> unless that card only supports sata 150, then make sure the drive is set to 150 only
[21:04] <jjohansen> but I don't see a cciss
[21:04] <Italian_Plumber> My controller card arrives tomorrow -- just in time for the weekend.
[21:04] <jjohansen> mathiaz: I'll poke around and see if I can figure out why
[21:05] <Italian_Plumber> 150, 300... I don't care.  Space is my biggest issue.
[21:05] <mathiaz> jjohansen: hmm - is it there on the karmic kernel?
[21:05] <patdk-lap> Italian_Plumber, ya card is only 150, so make sure you install the jumper on the drive to limit it to 150
[21:06] <Italian_Plumber> ok I'll check.  Thanks!
[21:06] <patdk-lap> cciss                  97481  3
[21:06] <patdk-lap> it's a module on my redhat system
[21:06] <jjohansen> mathiaz: karmic is the same thing
[21:07] <patdk-lap> cciss module exists in karmic desktop
[21:07] <mathiaz> jjohansen: ok - I'm still tracking down which module should be used for the controller
[21:07] <patdk-lap> it's in -generic
[21:08] <patdk-lap> /lib/modules/2.6.31-17-generic/kernel/drivers/block/cciss.ko
[21:08] <patdk-lap> dunno about the install cd though
[21:08] <mathiaz> patdk-lap: that's on karmic?
[21:08] <patdk-lap> yep
[21:09] <patdk-lap> want me to check hoary?
[21:09] <patdk-lap> it's in hoary also
[21:10] <patdk-lap> maybe it's just missing from the install cd initrd?
[21:10] <MTecknology> patdk-lap: what does the key I'm using have to do with wether or not I can run the 64bit os?
[21:11] <patdk-lap> MTecknology, microsoft is completely nuts about their stuff
[21:11] <patdk-lap> you have to have the cd and key that go together
[21:11] <patdk-lap> and some cd's you even have to have the right computer that goes with it
[21:11] <patdk-lap> you can't just use any win xp key with any win xp cd
[21:12] <patdk-lap> now, running a 64bit os in a vm is totally different, normally your computer must support VT extentions on the cpu to do that
[21:14] <MTecknology> patdk-lap: yup... now your on my page :P - proc supports that; bios doesn't allow me to enable that support
[21:14] <patdk-lap> what vm you using?
[21:14] <MTecknology> vbox
[21:14] <patdk-lap> bios upgrade? :)
[21:15] <MTecknology> none available
[21:15] <patdk-lap> your completely screwed, it's not a win xp 64 or key issue
[21:15] <MTecknology> coreboot - can't do that on this system
[21:15] <techsupport> i'm having trouble locating the download for ubuntu server 32 bit 9.10
[21:15] <MTecknology> I never said it was
[21:15] <patdk-lap> it's just your bois won't let you
[21:15] <MTecknology> which is what I already stated
[21:15] <MTecknology> :P
[21:15] <patdk-lap> you said your bios has issues with your win xp 64 key :)
[21:15] <patdk-lap> not that your bios has issues turning on vm support
[21:15] <patdk-lap> vt support
[21:16] <MTecknology> techsupport: updatedb; locate *.iso
[21:16] <MTecknology> patdk-lap: 14:59 < MTecknology> This is irritating... I have a WinXP 64bit key but I can't use it in a VM beecause my BIOS won;t let me
[21:16] <MTecknology> bios won't let me
[21:16] <techsupport> i didn't download it yet
[21:16] <techsupport> i'm trying to download it
[21:16] <techsupport> lol
[21:16] <techsupport> from ubuntu.com or something
[21:18] <MTecknology> techsupport: http://www.ubuntu.com/getubuntu/download
[21:18] <techsupport> it says desktop
[21:18] <techsupport> begin download desktop
[21:18] <MTecknology> Alternative download options, including Ubuntu installer for Windows
[21:19] <techsupport> and this one says 64 bit
[21:19] <techsupport> what the hell
[21:19] <techsupport> http://www.ubuntu.com/getubuntu/download-server
[21:20] <techsupport> where is the 32 bit server ?
[21:21] <MTecknology> techsupport: Alternative download options
[21:21] <MTecknology> techsupport: read the page
[21:25] <techsupport> sorry too much work lately
[21:27] <mathiaz> jjohansen: ok - so cciss.ko is available in the lucid generic kernel as well: /lib/modules/2.6.32-11-generic/kernel/drivers/block/cciss.ko
[21:32] <MTecknology> techsupport: I know the feeling
[21:51] <dthacker> Hello, I'm having problems coming up with the correct /etc/sudoers syntax for this scenario.   Users hans, bob, and fred should be be able to run any command that can be run by username "produser"
[21:54] <dthacker> Ah, I misread the record format.  This works     hans, bob, fred       ALL=(produser)   ALL
[22:03] <MTecknology> dthacker: you can make groups as well
[22:04] <MTecknology> dthacker: I think man sudoers explains it pretty nicely
[22:05] <dthacker> MTecknology: yes, one step at a time.  I'm still not getting the correct config.   I now get: "/usr/bin/ksh produser as root on my.local.host"
[22:05] <dthacker> argh
[22:06] <MTecknology> dthacker: I'd make a group called produser and add the users to that group; then give that group access
[22:06] <MTecknology> dthacker: not system group; sudoers group
[22:08] <dthacker> MTecknology: would that be "User_Alias     PRODUSERS= hans,bob,fred
[22:08] <hsn> how to disable mysql from autostarting on boot?
[22:12] <patdk-lap> correct way? something with up init or something
[22:12] <MTecknology> dthacker: I thought the syntax was closer to -  User_Alias ::= NAME '=' User_List
[22:12] <patdk-lap> personally I just do a chmod -x /etc/init.d/mysql
[22:12] <MTecknology> dthacker: could have changed though
[22:13] <dthacker> MTecknology: will re-read man pages
[22:13] <dthacker> didn'nt see any double colons
[22:13] <MTecknology> dthacker: I'd also make an alias 'Cmnd_Alias' then just link the two down below
[22:13] <patdk-lap> hmm, update-rc.d
[22:21] <jjohansen> mathiaz: yep, ccisss == cpq_ciss
[22:22] <patdk-lap> oh? cciss is gone in lucid? and replaced with cpq_ciss?
[22:36] <MTecknology> I'm gone for 1/2hr - windows is still at the "preparing to dload sp3"
[22:39]  * dthacker decides to build a test case for this at home on his server farmlet
[22:54] <hsn> patdk-lap: yes, update-rc.d did it
[22:55] <MTecknology> hsn: check out rcconf sometime; handy little tool
[23:16] <stickystyle> hsn: I second rcconf, I have it on all my servers.
[23:17] <glen1> you know when you buy a song off amazon, how if gives you a link that works for that time buy expires
[23:17] <glen1> what is that called?
[23:18] <hggdh> ripoff?
[23:19] <glen1> haha
[23:19] <hggdh> sorry. Long time I don't by a song off amazon, don't remember...
[23:20] <glen1> well if gives you a link but that link expires
[23:20] <glen1> so you can download it once but never again using the same url
[23:20] <glen1> how does it do that? does the actual location of the file move with each purchase?
[23:20] <hggdh> yes. I dimly remember you call contact them to reinstate it (if you have not yet downloaded the songs).
[23:21] <hggdh> I do not know. I would expect the location would be calculated based on the data on the HTTP headers
[23:21] <glen1> its not really to do with amazon, its just the concept. How do they give out a link that can expire
[23:22] <patdk-lap> gen1, on lighttpd it's call, secdownload
[23:22] <glen1> ohhh thats the money
[23:22] <patdk-lap> and there are many ways to do it
[23:22] <patdk-lap> entry in database that expires
[23:22] <patdk-lap> or many many other ways
[23:23] <glen1> i always wondered how, it interested me. thanks for pointing it out to me :D
[23:23] <glen1> so I guess it's used alot in ecommerce
[23:23] <patdk-lap> and other things
[23:24] <patdk-lap> like if you want to keep people from hotlinking files
[23:24] <patdk-lap> but still want to let people download it
[23:25] <patdk-lap> my company would use it, if they can get their act together, to give new employees their benifits and other documents
[23:25] <glen1> could you explain how it would be used in a company?
[23:25] <patdk-lap> heh?
[23:25] <patdk-lap> you don't use it, just cause you can
[23:25] <patdk-lap> you need a purpose for it
[23:26] <glen1> im looking through it, how does it authenticate the user?
[23:26] <patdk-lap> in my case, it would be so only that one user can download the info, without the problems of using authenication
[23:26] <glen1> I may try to run this on my own home server to learn
[23:27] <patdk-lap> that is up to you
[23:27] <glen1> are there any other things similiar to this?
[23:27] <patdk-lap> in this example, nothing is authenicated, once you get that one time use url
[23:27] <patdk-lap> getting that url normally requires somekind of authenication
[23:27] <patdk-lap> but that is up to you to handle
[23:28] <glen1> ahh cool thanks
[23:35] <Roxyhart0> Hi I just installed ubuntu 9.10 for my server, but i had some problems to install openldap...somebody have good experience or bad experience with this ubuntu version?
[23:40] <RoyK> Roxyhart0: if you have problems with openldap, please explain them and someone might know. I and probably several others are running 9.10 for servers
[23:42] <Roxyhart0> Thanks..9.10 is no comming with slapd.conf and I dont know how to configurate manualy my Domain Name
[23:58] <Roxyhart0> somebody know how to configurate manually openldap on ubuntu 9.10 (is doesn't have slapd.conf file)
[23:59] <Roxyhart0> configure (sorry)