[00:04] <pilif12p> Which option do i use to dual boot?
[00:04] <kirkland> mathiaz: yo
[00:05] <kirkland> mathiaz: wanna test a libvirt-bin upstart script for me in your env where you're having node troubles?
[00:22] <pilif12p> Hello again. I'm confused as how to dualboot
[01:01] <tos__> how do i set the mac address to use when adding a network alias such as eth0:0 75.1.75.1
[01:02] <JohnA> help gui
[01:06] <jla> jdstrand: just got back from supper. an overly complex method for setting up a fw.
[01:14] <JohnA> jdstrand: I really need to learn to type.
[01:17] <jdstrand> kirkland`: re libvirt/upstart-- I do not
[01:25]  * jdstrand wonders how 'ufw allow OpenSSH ; ufw enable' is so complex. *shrug*
[01:25]  * jdstrand wanders off
[01:30] <JohnA> a
[01:33] <kirkland`> jdstrand: cool, i have one, working well for me
[01:33] <kirkland`> jdstrand: are you interested in helping test it?  :-) :-) :-)
[01:40] <erichammond> smoser, et al: Do you remember where the document is describing the release policy/schedule for official Ubuntu AMIs?
[01:40] <smoser> https://wiki.ubuntu.com/UEC/Images/RefreshPolicy
[01:41] <smoser> its on my plate to get that officially radified and agreed to by QA and tech board
[01:41] <smoser> at the moment i'm leaning towards simplyfying the conditions for update to be something like:
[01:41] <smoser> a. security or major issue
[01:41] <smoser> b. monthly unless not necessary
[01:42] <erichammond> I'd like to make a request for updates to be released for Karmic (and optionally Hardy).
[01:42] <smoser> rather than some other more complex to predict update policy
[01:42] <smoser> you mean *now* ? or in general.
[01:42] <erichammond> The apt-get upgrade on Karmic has an ugly manual grub config prompt which can't be avoided with DEBIAN_FRONTEND=noninteractive
[01:42] <smoser> the policy does say it covers karmic and hardy
[01:43] <smoser> is there a bug open?
[01:43] <erichammond> smoser: now as in "some day soon"
[01:43] <erichammond> Would you like me to create a bug in launchpad?
[01:43] <erichammond> will do
[01:45] <erichammond> I like your monthly/major issue proposal
[01:46] <erichammond> Ubuntu is very schedule driven in other ways so that folks can do their planning around it.
[01:52] <erichammond> Hm, what package should I use as an argument to ubuntu-bug when I'm submitting a report against the AMI building software/process?
[02:04] <zul> smoser: still around?
[02:04] <smoser> here
[02:04] <zul> python-boto got rejected because the changelog still said karmic ill fix it and re-upload it
[02:05] <smoser> bugger. sorry.
[02:05] <zul> no problem ill fix it and re-upload it
[02:11] <erichammond> smoser: bug 503649 for your enjoyment
[02:11] <smoser> oh, erichammond i actually was wondering about the grub update bug :-(
[02:11] <erichammond> Well, the info there could be turned into such a bug if it really is a bug.
[02:12] <erichammond> I don't like the behavior, but didn't know if it was intended.
[02:14] <zul> smoser: fixed
[02:21] <tos__> hi i am trying to make use of all my ips.... i have ubuntu-server installed with 1 NIC.. and im trying to get each ip on that NIC, but my router only allowd 1-ip per MAC, how can i make Virtual NIC's with different mac's
[02:22] <tos__> the eth0:0-1-2-3-4 doesn't work same mac for each
[02:22] <tos__> aliasing
[03:59] <tos__> !vzeth
[04:57] <majuk> Hey guys, I have a group Admins and a group Inst. They both have rw to /srv/instr but only Admin has rw to /srv/admin. The problem I have is users in Admin are dumping folders into /srv/instr and getting their group, thus making instr group unable to see them.
[04:57] <majuk> I've found the set guid for the folder, but that only modifies moved/created files, not directories
[04:58] <majuk> set guid bit*
[05:01] <tos__> ok so i tried to install openVZ to assist with creating virtual ethernet interface, However im using Ubuntu 9.10, and have never recompiled a kernel or anything like that... does anyone know of another way to accomplish this, My router only assigns 1 ip per MAC address and I cant change the router so I need a virtual solution... if there is an easy way out there for ubuntu with a HOW-to or not too much config'ing ??
[05:16] <qman__> majuk, I have a similar situation, though I'm unaware of an enforcing solution
[05:17] <qman__> I use a create mask in samba and a daily cron job to work around the problem
[05:18] <majuk> qman__, Yeesh
[05:19] <qman__> yeah
[05:19] <qman__> I didn't say it was a good solution
[05:20] <qman__> if there's a better way, I'm certainly interested
[05:21] <majuk> I think I might have found one.
[05:21] <majuk> qman__, Guy in #gentoo said it is determined by the program creating the folder. So there is probably a setting in Samba
[05:21] <majuk> Just a guess. I looked through the man doc, but I'll look again.
[05:22] <qman__> yeah, with samba you can force group and create mask
[05:22] <qman__> but it only helps with files created via samba, stuff put in other ways can be a problem
[05:22] <majuk> I know the create mask, but I only found a force user option
[05:22] <majuk> Yea, my users don't. Windows kiddies.
[05:22] <majuk> :D
[05:22] <qman__> ah
[05:22] <qman__> then samba should support everything you need
[05:22] <majuk> Well then I'll take my pick-axe back to the man pages.
[05:23] <qman__> force group = groupname
[05:23] <qman__> create mask = 0664
[05:23] <qman__> directory mask = 0775
[05:23] <qman__> that's what I use
[05:23] <majuk> Yea I have my everyone locked out for these folders. Hence the problem with my 2 groups not being able ot see one another's files.
[05:23] <majuk> 0660
[07:57] <JohnA> quit
[08:34] <xiaomai> i've mounted an nfs partition rw, but when i attempt to write to that partition (as root), i get permission denied errors.  how can that be?
[08:51] <alvin> xiaomai: There are different possibilities here. NFS3 of NFS4?
[08:56] <xiaomai> alvin: i think it's nfs3.  i've chmod 777 on the root directory and now i get this weird 4294967294 uid/gid on the files i create
[08:57] <alvin> xiaomai: That's the reason. What line is in /etc/exports on your NFS server?
[08:58] <xiaomai> alvin: i've got "/mntpoint    myip/range(rw,sync,no_subtree_check)"
[08:58] <xiaomai> er, i guess mntpoint is a directory (/data/webserver (it's just part of the root filesystem))
[08:59] <alvin> xiaomai: add 'no_root_squash' to your options
[08:59] <alvin> then do $ sudo exportfs -av en remount the share
[09:00] <Jaredster> has anyone here had troubles with CUPS after upgrading to karmic?
[09:00] <Jaredster> I cant get it to share anymore over the network
[09:00] <Jaredster> I've tried everything I can think of.
[09:01] <Jaredster> it worked fine on jaunty
[09:01] <xiaomai> alvin: beautiful, that did the trick, thanks!
[09:01] <alvin> xiaomai: You're welcome
[09:01] <xiaomai> alvin: i saw that option in the manpage, but it didn't seem like i needed it--guess i did
[09:04] <jiboumans> morning
[09:07] <Jaredster> ipp doesnt work in 9.10 apparently?
[09:08] <twb> Jaredster: that's not a question.
[09:09] <Jaredster> thank you
[09:10] <_ruben> hmm .. is it "normal" for the `du` command to not work properly on fuse mounts?
[09:10] <_ruben> it shows 0 bytes for everything
[09:10] <_ruben> ls does show proper filesizes
[09:11] <twb> _ruben: --apparent-size ?
[09:12] <_ruben> twb: thanks .. that seems to do the trick ..
[09:12]  * _ruben hides in shame for not even looking at the manpage of du
[09:13] <twb> I am guessing that you're looking at files that don't have any blocks on a physical disk or something
[09:13] <twb> Or maybe your FUSE implementation just doesn't implement those ioctls
[09:13] <twb> s/implementation/filesystem/, I mean. e.g. sshfs
[09:15] <_ruben> they're fuse mount of vmfs volumes (vmware luns)
[09:18] <twb> Yecch
[09:18] <twb> How does that relate to vmware's hgfs?
[09:18] <_ruben> not at all
[09:19] <_ruben> hgfs is a special fs to enable drag'n'drog for virtual machines on hosted products ... vmfs is used to store disk images on
[09:20] <twb> Oh, so kinda like using LVM and giving a VM direct access to an LV (instead of a file on a filesystem) for its block device.
[09:20]  * twb reads wikipedia
[09:22] <Jaredster> I'm trying to configure cups right now to share a printer over the network
[09:22] <Jaredster> and no other computer on the network can see it
[09:22] <Jaredster> it worked fine with 9.04, but when I upgraded to 9.10 it broke.
[09:24] <Jaredster> has anyone else dealt with this?
[09:24] <twb> I stick to LTS, sorry.
[09:24] <_ruben> twb: not really, basicly vmfs is just another cluster fs like ocfs2 .. though with some features like thick and thin provisioning
[09:24] <twb> _ruben: okey dokey
[09:31] <Jaredster> what do you do if you delete an init script and need to restore it?
[09:34] <Jaredster> nevermind, got it
[09:39] <twb> Jaredster: sudo dpkg-divert --rename?
[09:39] <twb> Oh, "if you deleteD"
[09:39] <Jaredster> yeah
[09:39] <Jaredster> I purged it and reinstalled
[09:40] <Jaredster> I was going to do it anyway
[09:58] <jiboumans> ttx: ping?
[09:58] <ttx> jiboumans: o/
[09:58] <jiboumans> ready?
[09:58] <ttx> jiboumans: fire
[09:58] <jiboumans> awesome :)
[11:18] <ivoks> ttx: thank you for reminder :)
[12:39]  * kirkland going for a run; will try to be back for start of meeting
[12:46] <zul> morning
[14:00] <ttx> kirkland: for some mysterious upstart reason, my "start on (started ssh and started avahi-daemon)" publication scripts are borken in the current euca. I'm investigating
[14:04] <ttx> kirkland: for some reason when ssh/avahi-daemon restart when picking up eth0, the publication task is *not* restarted.
[14:15] <mathiaz> ttx: are the upstart jobs marked respawn?
[14:15] <mathiaz> ttx: if the avahi-daemon dies, avahi-publish aren't restartd
[14:16] <ttx> mathiaz: that's the lead i've been following... but marking them respawn doesn't help. Furthermore, shouldn't it get restarted when ssh and avahi-daemon are "started" again ?
[14:17] <ttx> it gets properly killed by the "stop on stopping ssh or stopping avahi-daemon"
[14:17] <mathiaz> ttx: I'm not sure that a new interface coming up generates a started sshd/avahi-daemon upstart event
[14:17] <ttx> but doesn't get restarted when ssh and avahi-daemon are started again
[14:17] <ttx> hmm
[14:18] <mathiaz> ttx: hm... I don't know then
[14:18] <ttx> so they would stay started
[14:18] <mathiaz> ttx: I'd ask keybuk about it
[14:18] <ttx> and avahi-publish would get SIGTERMed when avahi-daemon dies
[14:18] <ttx> I will
[14:20] <jdstrand> kirkland: re upstart/libvirt> I can test it, sure
[14:21] <kirkland> jdstrand: i was having trouble building lucid's libvirt locally
[14:21] <kirkland> jdstrand: was failing some tests
[14:23] <jdstrand> kirkland: the test suite is flaky in spots. did you build in a deep directory?
[14:23] <kirkland> jdstrand: /local/source/libvirt/libvirt-0.7.2$
[14:24] <kirkland> jdstrand: http://paste.ubuntu.com/352342/
[14:25] <kirkland> jdstrand: interestingly, it built in my ppa
[14:25] <kirkland> jdstrand: but not on my local system
[14:25] <jdstrand> kirkland: are you using pbuilder?
[14:25] <kirkland> jdstrand: not for this
[14:26] <kirkland> jdstrand: is that a requirement for the test to work?
[14:26] <jdstrand> I use sbuild, and it built fine
[14:26] <jdstrand> kirkland: all the vmx2xml stuff dies cause it can't find something
[14:27] <jdstrand> kirkland: I'd look at the build log in LP and make sure you have everything installed
[14:27] <kirkland> jdstrand: okay
[14:27] <jdstrand> kirkland: beyond that, soren was working on enabling the test suite-- so I'd ping him on problems there
[14:28] <soren> hm?
[14:28] <jdstrand> soren: libvirt testsuite not working for kirkland. seems a dep isn't installed, but he'll see
[14:28] <kirkland> jdstrand: also, there's a regression in debian/control build deps
[14:28] <soren> kirkland: Which test fails?
[14:28] <kirkland> jdstrand: seems that open-iscsi snuck back in there (rather than open-iscsi-utils)
[14:29] <kirkland> soren: http://paste.ubuntu.com/352342/
[14:29] <jdstrand> kirkland: oops. can you fix?
[14:29] <kirkland> jdstrand: yeah
[14:29] <kirkland> jdstrand: i have it fixed with the libvirt stuff i'm working on
[14:31] <soren> kirkland: *shrug* No clue.
[14:32] <kirkland> soren: k
[15:07] <garymc> Hi guys anyone know how i can resett my my phpmyadmin user and password, i seem to have forgotten them
[15:23] <jdstrand> soren: btw, I don't mind doing the libvirt merge before it hits testing (in fact, I already looked at it earlier this week ;) but since it was such a big change, I wanted to see what bugs came out of it. Also, kirkland is working on libvirt/upstart and I have some other things I'm working on atm, so saying 'when it hits testing' more or less gave an accurate timeframe ;)
[15:23] <kirkland> jdstrand: cool
[15:23] <kirkland> jdstrand: yeah, i hope to upload the upstart change today
[15:23] <jdstrand> kirkland: nice
[15:24] <kirkland> jdstrand: i sent it to my PPA, while i figure out why it's not building for me locally
[15:24] <kirkland> jdstrand: side benefit of getting others testing it
[15:46] <smoser> kirkland, ttx you kwno what networking device the instances in EUC get ?
[15:46] <Davidf88> anyone know about the enterprse clud stuff?
[15:47] <kirkland> smoser: you mean eth*?
[15:47] <smoser> ie, what -net nic,model=XXXXXX
[15:47] <Davidf88> Guys, hoping someone can help
[15:48] <Davidf88> I have a 64 bit cluster and 64 bit node
[15:48] <Davidf88> however having an issue with running instances
[15:48] <Davidf88> they start, run then terminate straight away
[15:48] <Davidf88> anyone got any ideas?
[15:52] <smoser> Davidf88, need more info, this is karmic ? instance and uec?
[15:52] <smoser> one thing to try is using a larger --instance-type
[15:53] <Davidf88> smoser: its karmic yes
[15:54] <Davidf88> yeah when using the larger --instance-type
[15:54] <Davidf88> it just does the same thing
[15:54] <Davidf88> but prolongs it
[15:58] <smoser> kirkland, if you have an instance running, you can just dump its libvirt xml for me
[15:58] <smoser> or even just ps -ww for the cmdline of kvm
[15:59] <Davidf88> smoser: I am also gettng this
[15:59] <Davidf88> in MANAGED-NOVLAN mode, priv interface 'eth0' must be a bridge, tunneling disabled
[15:59] <Davidf88> in cc.log
[16:00] <smoser> Davidf88, i'm sorry, the instance-type was really the only suggestion i had. i've had ones that were to small misteriously dying
[16:01] <Davidf88> ok smoser somehow now have it running :s bit of a cluster-f**k to be honest
[16:02] <mathiaz> nijaba`: I need to refine the script
[16:02] <mathiaz> nijaba`: and make sure it gives out the correct packages
[16:02] <nijaba`> mathiaz: have you looked at the python rewrite that mvo has done
[16:02] <nijaba`> ?
[16:03] <mathiaz> nijaba`: which one?
[16:03] <nijaba`> mathiaz: ubutnu-maintenance-check
[16:03] <nijaba`> mathiaz: he has rewrote it in python and verified the results
[16:04] <mathiaz> nijaba`: does this grab the list of packages published by the security team?
[16:04] <nijaba`> mathiaz: nope, it use seeds/germinate
[16:05] <nijaba`> mathiaz: https://code.launchpad.net/~mvo/ubuntu-maintenance-check/python-port
[16:05] <mathiaz> nijaba`: ah ok. I'll have a look at it then
[16:07] <mathiaz> nijaba`: hm - would the result of the script include things like basic libraries?
[16:07] <mathiaz> nijaba`: such as libopenssl for example?
[16:07] <nijaba`> mathiaz: every package that is in main because of a server related seed
[16:08] <mathiaz> nijaba`: IIUC libopenssl is maintained for 5 years, but not relevant to the server tema
[16:08] <mathiaz> nijaba`: right - getting the list of -server related packages is a bit more difficult
[16:08] <mathiaz> nijaba`: as we're looking for packages that are *only* pulled in by a server seed
[16:09] <mathiaz> nijaba`: plus a couple of others
[16:09] <nijaba`> mathiaz: well, what product do we have that is maintained for 5y which is not server releated?
[16:09] <mathiaz> nijaba`: the perspective here is from the server team - and get the list of packages related to it
[16:09] <mathiaz> nijaba`: example: libopenssl is maintained for 5 years
[16:10] <nijaba`> mathiaz: or do you exclude packages maintained bu the foundation team, but which are used by the server?
[16:10] <mathiaz> nijaba`: however I don't think that ubuntu-server should be a bug contact
[16:10] <mathiaz> nijaba`: seems like a fondation thing to me
[16:10] <mathiaz> nijaba`: right
[16:10] <mathiaz> nijaba`: I'd like to exclude packages that are maintained by other team
[16:10] <nijaba`> mathiaz: ok, so you do not want all server seeded pacakes, only the ones assigned to the server team, which is not tacked in seeds
[16:11] <mathiaz> nijaba`: so we need to have a more fine-grained selection
[16:11] <nijaba`> mathiaz: yep
[16:11] <nijaba`> mathiaz: why do you add a ` to my handle?
[16:11] <mathiaz> nijaba`: the script I wrote selects packages that: 1. are *directly* seeded in a server seed
[16:11] <nijaba`> mathiaz: ok, makes sense
[16:12] <mathiaz> 2. are *only* pulled in by a server seed
[16:12] <mathiaz> nijaba`: because that's what irssi completes to
[16:12] <mathiaz> nijaba`: at least that how I see your nickname
[16:13] <nijaba`> hmm... weird.  should not...
[16:18] <nijaba> should be fixed...
[16:26] <rj175> Hello, I now have a instance running how do I ssh to it?!
[16:27] <nijaba> rj175: ssh -i key ubuntu@ipaddress?
[16:28] <rj175> nijaba: ssh: connect to host 192.168.1.35 port 22: No route to host
[16:28] <nijaba> rj175: well, looks like you have a routing pb, then, or you are not specifying the public address
[16:29] <rj175> nijaba: how can I debug this issue?
[16:30] <nijaba> rj175: are you on uec or ec2?
[16:30] <rj175> in using ubuntu 9.10 with eucalyptus
[16:31] <nijaba> rj175: ok. I that case, what is the public ip address range you specified during install?
[16:32] <rj175> 192.168.1.35-39
[16:32] <nijaba> and what is your station ip address?
[16:32] <rj175> u mean the cluster and node ips?
[16:33] <rj175> nijaba: the cluster (the web front) is 192.168.1.32 and the node is 192.168.1.33
[16:33] <nijaba> nope, the machine from which you are trying to ssh to your instance
[16:33] <rj175> 192.168.1.82
[16:34] <nijaba> rj175: can you ping .35 from .82?
[16:34] <sub> Anyone know if HP's support pack available in any of the standard repos?
[16:34] <rj175> nijaba: no
[16:34] <nijaba> rj175: then your instance is not fully up, it seems.
[16:35] <rj175> nijaba: the instance says its running
[16:35] <nijaba> rj175: what does euca-describe-instance report?
[16:35] <rj175> INSTANCE        i-38FC0669      emi-E01C1076    192.168.1.35    172.19.1.2      running         mykey   0       m1.large        2010-01-06T16:20:12.528Z        cluster1        e
[16:36] <rj175> ki-F69E10F7     eri-0B151165
[16:36] <nijaba> rj175: hmm...  then you should have a look at the boot messages from that instance.  sonds like it got stuck somewhere
[16:36] <rj175> nijaba: where can i find the boot message?
[16:38] <nijaba> rj175: euca-get-console-output
[16:38] <ttx> smoser: ping
[16:38] <smoser> here
[16:39] <rj175> nijaba: all i get is:
[16:39] <rj175> i-38FC0669
[16:39] <rj175> 2010-01-06T16:39:33.132Z
[16:45] <nijaba> rj175: you should really have something from the console.   Can you see what it says with --debug?
[16:46] <rj175> it says the exact same
[16:46] <rj175> let me try stopping it and running it again
[16:49] <rj175> nijaba: just the same :()
[16:49] <nijaba> rj175: :/
[16:51] <rj175> nijaba: the is one thing in the cc.log, in MANAGED-NOVLAN mode, priv interface 'eth0' must be a bridge, tunneling disabled
[17:01] <Doonz> Ok here's my issue. server 1 can log into server 2 through ssh without password authentication. but server 2 cannot log into server 1 without having to enter a password. Im trying to set up rsync from server 2 to server 1 without being prompted for password
[17:13] <Doonz> Ok here's my issue. server 1 can log into server 2 through ssh without password authentication. but server 2 cannot log into server 1 without having to enter a password. Im trying to set up rsync from server 2 to server 1 without being prompted for password
[17:14] <alex_joni> a server cannot log into another server
[17:14] <alex_joni> a users logs in to another server
[17:14] <alex_joni> you can setup users to not require a password using shared keys, examine your ~/.ssh/ folder
[17:27] <jiboumans> zul++ # burner of charts
[17:31] <zul> not when im eating
[17:34] <Doonz> alex_joni: yes
[17:34] <Doonz> user1 exists on both servers
[17:34] <Doonz> user1 runs the rsync job
[17:35] <Doonz> user1 on server1 can just type ssh server2 and log into terminal. user1 on server2 runs the rsync command. but when user1 tries to ssh into server1 he is prompted for a password
[17:46] <mathiaz> kirkland: jdstrand: where is the libvirt upstart job you worked on yesterday?
[17:47] <jdstrand> mathiaz: I have not worked on it (presumably in kirkland's ppa)
[17:59] <Doonz> user1 runs the rsync job
[17:59] <Doonz> user1 on server1 can just type ssh server2 and log into terminal. user1 on server2 runs the rsync command. but when user1 tries to ssh into server1 he is prompted for a password
[18:07] <kirkland> mathiaz: http://paste.ubuntu.com/352435/
[18:10] <nod> hi - i've just done a vanilla install of 9.10 cloud with 1 cluster head and one node.  my store is empty of images, yet the docs seem to say i should have some default images there.
[18:10] <nod> i don't mind bundling my own (and intend to) but I'm just curious if I've not missed something and therefore my installation isn't complete.
[18:10] <nod> my node is being detected and the zone listed
[18:10] <mathiaz> kirkland: great. I'll give it a try in daily UEC testing
[18:14] <TeTeT> nod: there should be images in the image store, but not deployed in the cloud
[18:19] <pmatulis> anyone have any experience with cyrus2dovecot (migration from cyrus to dovecot)?
[18:41] <Doonz> user1 runs the rsync job
[18:41] <Doonz> user1 on server1 can just type ssh server2 and log into terminal. user1 on server2 runs the rsync command. but when user1 tries to ssh into server1 he is prompted for a password
[19:01] <zooko_sg> What's your favorite way to install Ubuntu as a guest in a VirtualBox?
[19:04] <kirkland> mathiaz: hey
[19:04] <kirkland> mathiaz: i have my pxe boot/install working
[19:04] <mathiaz> kirkland: howdy sunland!
[19:04] <kirkland> mathiaz: can you point me to your preseed files you're using for UEC?
[19:04] <kirkland> ;-)
[19:05] <zooko_sg> I guess I'll just dl the .iso and do an install...
[19:05] <mathiaz> kirkland: sure - let me generate one
[19:05] <kirkland> mathiaz: gimme a nc to start with
[19:07] <smoser> kirkland, when you do get an instance up and running, please do a 'ps -aww | grep kvm' for me
[19:07] <smoser> i just want to see what kvm is invoked as
[19:07] <kirkland> smoser: will do
[19:07] <smoser> which, i think, will tell me what network driver is being used.
[19:08] <mathiaz> kirkland: http://people.canonical.com/~mathiaz/uec-node.preseed
[19:08] <mathiaz> kirkland: you should have a quick look at it
[19:08] <mathiaz> kirkland: as you may wanna change some information
[19:09] <mathiaz> kirkland: the file is documented
[19:10] <kirkland> mathiaz: of course ;-)
[19:10] <mathiaz> kirkland: the key part is: d-i anna/choose_modules string eucalyptus-udeb
[19:10] <kirkland> mathiaz: so where does this go in my tftpboot dir?
[19:10] <mathiaz> kirkland: that line will kick the UEC installer component
[19:10] <kirkland> mathiaz: or does it go with my http deb dir?
[19:10] <mathiaz> kirkland: hm - multiple option
[19:11] <mathiaz> kirkland: hm - multiple options
[19:11] <mathiaz> kirkland: easiest is http deb dir
[19:11] <mathiaz> kirkland: then you point the install to preseed=http://server/preseed via the command line
[19:11] <mathiaz> kirkland: the most elegant is to stick into the ramdisk sent to the system
[19:12] <mathiaz> kirkland: but that means regenerating the ramdisk
[19:12] <kirkland> mathiaz: hrm
[19:12] <mathiaz> kirkland: IIRC the ramdisk is fetched via tftp
[19:12] <kirkland> mathiaz: okay, so this is perhaps where the cgi-bin comes in?
[19:12] <kirkland> mathiaz: b/c i'd like the install to be automatic
[19:12] <kirkland> mathiaz: in case i'm not at the console of the installing machine
[19:12] <mathiaz> kirkland: right - both methods will be automatic
[19:13] <mathiaz> kirkland: so if you use the http based method, you'll have to bootstrap a couple of debconf question via the kernel command line
[19:13] <kirkland> mathiaz: oh?  so the name "preseed" is special?  a netboot always looks for that?
[19:13] <mathiaz> kirkland: oh no. I made that up
[19:14] <mathiaz> kirkland: on the kernel command line, you'll add: preseed=http://my-server/preseed
[19:14] <mathiaz> kirkland: that means the installer will download the preseed as soon as it can (ie after networking is up)
[19:14] <kirkland> mathiaz: okay -- how do i do that automatically?
[19:14] <mathiaz> kirkland: that's through the dhcp option IIRC
[19:14]  * mathiaz thinks
[19:15] <kirkland> mathiaz: something in the install/netboot dir?
[19:15] <kirkland> mathiaz: i was bind/loop mounting it
[19:15] <kirkland> mathiaz: but i can create a rw copy
[19:15] <kirkland> mathiaz: and edit that into there
[19:16] <Doonz> user1 on server1 can just type ssh server2 and log into terminal. user1 on server2 runs the rsync command. but when user1 tries to ssh into server1 he is prompted for a password
[19:16] <Doonz> i have added the id_rsa.pub to the servers authorized keys
[19:16] <mathiaz> kirkland: it's set in the pxelinux.cfg file
[19:16] <mathiaz> kirkland: http://archive.ubuntu.com/ubuntu/dists/lucid/main/installer-amd64/current/images/netboot/ubuntu-installer/amd64/
[19:17] <kirkland> mathiaz: okay, coo
[19:17] <mathiaz> kirkland: ^^ this should be the root of your tftpboot directory
[19:17] <mathiaz> kirkland: the dhcp server points to pxelinux.0
[19:17] <mathiaz> kirkland: then pxelinux.0 download its configuration from pxelinux.cfg/
[19:18] <mathiaz> kirkland: it tries a list of well-known filenames
[19:18] <mathiaz> kirkland: starting with the mac-address, then the IP address in HEX, and finally default
[19:18] <mathiaz> kirkland: you'd have to look up the configuration format of pxelinux
[19:19] <mathiaz> kirkland: IIRC you can specify which kernel, initrd should be downloaded
[19:19] <mathiaz> kirkland: as well as the kernel command line
[19:19] <kirkland> mathiaz: awesome
[19:19] <kirkland> mathiaz: just what i need
[19:19] <mathiaz> kirkland: the tricky part is to manage the pxelinux.cfg/ directory
[19:20] <mathiaz> kirkland: what you'd usually do is that a the default file make the system boot from the local harddrive
[19:20] <kirkland> mathiaz: right
[19:20] <mathiaz> kirkland: if you want to (re)install the system, you just add a specific filename with the correct installation information
[19:21] <mathiaz> kirkland: and then you need a script that will automatically remove the specific filename when the installation has completed
[19:21] <mathiaz> kirkland: so that upon reboot the system doesn't reinstall
[19:21] <mathiaz> kirkland: one way to do that is to write a cgi script that gets the IP address of the installed system
[19:22] <sub> ^ that's how I've done it
[19:22] <mathiaz> kirkland: and deletes the specific filename (using the HEX version of the IP address)
[19:22] <sub> python CGI script to remove a symlink for the requesting IP address
[19:22] <mathiaz> kirkland: and you just need to call the cgi script using a late_command in the preseed file
[19:23] <mathiaz> sub: right - using symlinks helps as well
[19:23] <mathiaz> kirkland: ^^ you can setup different installation profile
[19:23] <mathiaz> kirkland: and then all you need to do is to create a symlink to the installation profile whenever you want to reinstall a node
[19:23] <mathiaz> kirkland: and then reboot the node
[19:24] <mathiaz> kirkland: another way to detect when an installation has been done is to scan syslog
[19:24] <mathiaz> kirkland: you'll see request for tftp
[19:24] <mathiaz> kirkland: when the pxelinux configuration file has been requested via tftp by the client, you can also erase it
[19:25] <mathiaz> kirkland: that way you don't have to call for the cgi script at the end of the installation
[19:26] <kirkland> mathiaz: cool
[19:26] <kirkland> mathiaz: this is good stuff
[19:26] <kirkland> mathiaz: its working well so far
[19:26]  * kirkland is done burning USB sticks
[19:26] <mathiaz> kirkland: yeah - it's a lot of pieces to be glued so far
[19:27] <mathiaz> kirkland: and then you need to think about how you can automatically reboot a system
[19:27] <mathiaz> kirkland: remotely - either via powercycling it
[19:27] <mathiaz> kirkland: or wake-on-lan
[19:27] <kirkland> mathiaz: we totally need a eucalyptus-auto-install package
[19:27] <kirkland> mathiaz: or some such
[19:27] <kirkland> mathiaz: that does everything i've done so far
[19:27] <mathiaz> kirkland: well - we need a system-installation package with eucalyptus profiles
[19:27] <kirkland> tftp, mounts, etc.
[19:28] <mathiaz> kirkland: make sure that BIOS of the system are set to always boot first from the network
[19:29] <mathiaz> kirkland: you wanna control everything from the network
[19:29] <mathiaz> kirkland: as you can use pxelinux to boot from a local harddrive
[19:29] <kirkland> mathiaz: right
[19:29] <mathiaz> kirkland: (chainloading IIRC)
[19:34] <smoser> kirkland, so i got the answer for mac
[19:34] <smoser> err, for network adapter type
[19:34] <smoser> its e1000
[19:34] <smoser> aubre got http://paste.ubuntu.com/352480/ for me.
[19:34] <smoser> 2 things interesting
[19:35] <aubre> :)
[19:35] <smoser> a.) i think ideal situation is that is virtio ... i'm just guessing that that should be the fastest network inteface for kvm
[19:35] <aubre> I like how the mac starts with d0:0d
[19:36] <mathiaz> smoser: right - virtio would be nice
[19:36] <smoser> b.) i have the strongest case to get CONFIG_VIRTIO_NET=y and thus get us networking for kvm guests built into kernel rather than a module
[19:36] <mathiaz> smoser: however it break compability with existing EC2 images
[19:36] <smoser> (i dont think we're going to convince anyone to build in e1000 at this point)
[19:36] <smoser> mathiaz, you're sure of that ?
[19:37] <smoser> ec2 probably doesn't use e1000 nick, probably xen virtual. id have to check though
[19:37] <mathiaz> smoser: well - existing EC2 images don't necessarly support virtio net devices
[19:37] <kirkland> smoser: yes, definitely virtio would be nice
[19:37] <kirkland> smoser: i *think* we're finally to a point where most guests should support virtio
[19:37] <mathiaz> smoser: I think there is the same problem with virtio block devices
[19:37] <smoser> remember that hte kernel is different for ec2 -> uec . so thats unavoidable.
[19:37] <smoser> well, i think that virtio block was somethign else
[19:37] <mathiaz> smoser: agreed.
[19:37] <smoser> the last point i wanted to make was
[19:38] <mathiaz> smoser: right - it was also a naming issue
[19:38] <smoser> ./tools/gen_kvm_libvirt_xml was modified in November of this year to turn e1000 as the device model
[19:38] <smoser> previously it was not set
[19:38] <mathiaz> smoser: virtio block device show up as /dev/vda, where as EC2 images expect it to be sd*
[19:39] <kirkland> mathiaz: i think we could reasonably hack that with symlinks
[19:39] <smoser> http://paste.ubuntu.com/352486/
[19:39] <smoser> mathiaz, your'e right on that.
[19:39] <mathiaz> kirkland: good point
[19:40] <kirkland> mathiaz: smoser: i'm all for using virtio disk and network the default in UEC
[19:40] <smoser> kirkland, yes, it could be "fixed" but you'd have to change the image (udev in it to do that)
[19:40] <kirkland> mathiaz: smoser: our VMs would *smoke*
[19:40] <smoser> so with changing to virtio for block device you'd have to change the AMI/EMI
[19:40] <kirkland> mathiaz: okay, one more silly question ... ks=http:/path/to/a/preseed file on the kernel command line...
[19:40] <mathiaz> kirkland: smoser: right - ^^ that was the issue against virtio block devices
[19:41] <kirkland> mathiaz: that should work right?
[19:41] <smoser> what does a virtio network device show up as ?
[19:41] <smoser> mathiaz, right, you are correct.
[19:41] <kirkland> mathiaz: are preseed/kickstart files interchangable in ubuntu?
[19:41] <kirkland> smoser: just eth0
[19:41] <smoser> yeah, so we might be able to get away with that
[19:41] <mathiaz> smoser: that's the good point: eth*
[19:41] <smoser> at very least it would be nice to allow someone to change that
[19:41] <kirkland> smoser: thankfully, that's no different
[19:41] <kirkland> 00:03.0 Ethernet controller: Qumranet, Inc. Unknown device 1000
[19:41] <mathiaz> kirkland: hm - I don't think so
[19:41] <kirkland> eth1      Link encap:Ethernet  HWaddr 00:16:36:32:4a:25
[19:42] <kirkland> mathiaz: oh?
[19:42] <kirkland> mathiaz: what's the kernel cmdline for a preseed file then?
[19:42] <mathiaz> kirkland: hm - file=http://
[19:42] <smoser> since it has recently changed (from whatever kvm default is) to e1000
[19:42] <mathiaz> kirkland: ?
[19:42] <smoser> we might be able to argue that it should have been changed to virtio
[19:42] <mathiaz> kirkland: I think it's file=
[19:43] <mathiaz> kirkland: https://help.ubuntu.com/9.10/installation-guide/i386/appendix-preseed.html
[19:43] <kirkland> smoser: who changed it?
[19:43] <smoser> see patebin for log
[19:43] <kirkland> mathiaz: cheers, mate
[19:43] <smoser> http://paste.ubuntu.com/352486/
[19:43] <smoser> oh, silly  me. root did :)
[19:43] <mathiaz> kirkland:   preseed/url=http://host/path/to/preseed.cfg
[19:43] <kirkland> smoser: nice
[19:44] <kirkland> smoser: is that you, or eucalyptus?
[19:44] <mathiaz> kirkland: ah - it's url= actually
[19:44] <smoser> that log was from bazaar.launchpad.net/~eucalyptus-maintainers/eucalyptus/1.6/
[19:44] <mathiaz> kirkland: file= will point the installer to a local file on the ISO
[19:44] <smoser> http://paste.ubuntu.com/352487/
[19:44] <kirkland> smoser: those guys kill me with that
[19:45] <smoser> well, you kniow it was someone who was root
[19:45] <smoser> so that limits it
[19:45] <mathiaz> smoser: hm - I can't think of any reason to *not* use virtio network devices by default on UEC
[19:45] <smoser> unless they used euca-root-wrap :)
[19:46] <kirkland> mathiaz: clients are the only limitation
[19:46] <kirkland> mathiaz: the client must support virtio
[19:46] <kirkland> mathiaz: which is true for Ubuntu >= hardy
[19:46] <mathiaz> kirkland: right - and that's an issue at the kernel level
[19:46] <kirkland> mathiaz: i don't know what that threshold is for debian, fedora, centos, etc.
[19:46] <mathiaz> kirkland: like smoser mentionned the EC2 kernel != UEC kernel
[19:46] <kirkland> mathiaz: but like i said, i think it's a safe assumption in 2010
[19:47] <garymc> Hi anyone know how i make my server work through a static ip address?
[19:47] <garymc> I bought 5 static ip address and im assigning a ip to each server
[19:47] <mathiaz> kirkland: since we provide official UEC kernels, we can say that they will support virtio
[19:48] <kirkland> mathiaz: sure
[19:48] <garymc> anyone?
[19:48] <mathiaz> garymc: "man interfaces" outlines how to configure a static ip address
[19:49] <garymc> where is that? im a newbie
[19:49] <mathiaz> garymc: http://doc.ubuntu.com/ubuntu/serverguide/C/network-configuration.html
[19:52] <garymc> mathiaz i think maybe this is something differnt
[19:52] <garymc> My Internet provider has gave me 5 ip address to the one broadband connection. Now before this I could access the server fine, and i would have to tunnel to the other servers
[19:53] <garymc> but now I assign an static ip to each server the servers are not connecting to the router. My phone company say i have to get my server to either recognise this or make sure they have dhcp enableed which i think they do?
[19:56] <mathiaz> garymc: you need to connect all of your servers directly to the router
[19:56] <mathiaz> garymc: and set them up with their respective ip address
[19:57] <garymc> they are connected directly to the router
[19:57] <guntbert> garymc: and the router must be configured appropiately
[19:57] <garymc> yes it is supposed to be, would i need to set anything else?
[19:57] <garymc> like in the server?
[19:58] <mathiaz> garymc: you need to make sure that the server is configured with the static ip address
[19:59] <mathiaz> garymc: if you run the command: ifconfig
[19:59] <mathiaz> garymc: it will give you the list of configure network interfaces
[19:59] <garymc> ok, im working remotley
[19:59] <mathiaz> garymc: and what IP address they're listening to
[19:59] <garymc> eth0 and eth1
[19:59] <garymc> 192.168.1.29
[19:59] <garymc> the router is handing that out
[20:01] <nod> hi - i asked a question earlier then ran off to lunch about 5 mins later.
[20:02] <nod> i see Tetec responded but is gone now
[20:02] <nod> hi - i've just done a vanilla install of 9.10 cloud with 1 cluster head and one node.  my store is empty of images, yet the docs seem to say i should have some default images there.
[20:02] <nod> i don't mind bundling my own (and intend to) but I'm just curious if I've not missed something and therefore my installation isn't complete.
[20:02] <nod> there's currently nothing listed in the image store
[20:03] <mathiaz> nod: is the cluster head allowed to reach the internet?
[20:03] <nod> mathiaz: ya
[20:04] <mathiaz> nod: you should check if the image-store-proxy is running on the system
[20:04] <mathiaz> nod: and look into its log files: /var/log/image-store-proxy/
[20:04] <nod> excellent.  will do
[20:04] <nod> thx for the pointer
[20:05] <nod> it says it's started up and listening on port 52780
[20:05] <nod> no errors.  and - it's currently running now
[20:07] <nod> whoa...  bizarre!  i just restarted it and _now_ there's images in the store :|
[20:17] <blackxored> hello, anyone knows about a way to reset a user's password in a AD domain through the shell?
[20:30] <garymc> ok i had to reset server or ifdown then ifupo for it to sort its ip address
[20:31] <garymc> anyone know the command to change my root password?
[20:32] <neonfreon> passwd
[20:32] <guntbert> !root | garymc
[20:32] <garymc> what?
[20:33] <garymc> I changed it before so i could give someone access i just want to change it back
[20:33] <neonfreon> passwd -l
[20:34] <garymc> l
[20:34] <garymc> ?
[20:34] <garymc> ok
[20:34] <garymc> no still dont get it
[20:35] <garymc> so dont i do something like, passwd change letmein
[20:36] <neonfreon> no
[20:36] <neonfreon> you want no passwd to work
[20:36] <neonfreon> so that sudo is the only way
[20:37] <neonfreon> read the man page on passwd
[20:37] <neonfreon> see what the -l command does
[20:40] <guntbert> garymc: you *should* disable the root password again (like neonfreon said)
[20:51] <ChrisRut> I am running Ubuntu Server 8.04 (hardy), and when I login remotely i can't see the colors when using ls, however after I type "bash" into the shell I see colors when using ls, I've already checked my .bashrc and all the colors are enabled by default (colors=auto), and I've checked "echo $SHELL
[20:51] <ChrisRut> " and which bash, and they both report /bin/bash
[20:51] <ChrisRut> hoever I don't see colors until I enter "bash" into shell
[20:52] <ChrisRut> why might this be?
[20:58] <soren> mathiaz: You know, a team ppa for those auto builds would be good. The first package just failed, and I just realised that as long as I'm the only one getting those e-mails, I'm also the poor sod who has to deal with it :)
[20:59] <mathiaz> soren: :)
[21:00] <mathiaz> soren: that being said, having team doesn't mean that the build failures will automatically be dealt with by *others* ;)
[21:01] <soren> mathiaz: No, but then we can all join in on procrastinating. It's a joint effort. :)
[21:37] <Doonz> Hi guys, What im trying to do is set up passwordless transfers using rsync between my two servers. each server has the same user on it with the same password. this will be the user calling for the rsync transfer on server2. The user we will call user1. from server 1 user1 can initiate a ssh session with server2 without being prompted to enter a password. Now from server2 user1 cannot initiate a ssh session with server1 without enter
[21:44] <ChrisRut> Doonz,
[21:44] <ChrisRut> you need to setup password-less authorization keys
[21:44] <Doonz> Chris wich i did. i did it from server 1 to server 2 first
[21:45] <Doonz> then redid the steps from server2 to server1
[21:46] <ChrisRut> Doonz: are both servers using the same "user"
[21:46] <Doonz> yes with same passwrd
[21:46] <Doonz> server 1 to server2 works
[21:46] <Doonz> server2 to server1 doesnt
[21:47] <ChrisRut> I think you need to generate a new key for server2 and then copy that over to server1, you can't use the same key on both (could be wrong thou)
[21:49] <Doonz> yeah ChrisRut thats what i did
[21:49] <Doonz> :/
[21:50] <ChrisRut> Doonz: hmm, not sure then, what (if any) error messages are you getting? can you try using verbose to get more info?
[21:54] <Doonz> ~/.ssh$ ssh -o PreferredAuthentications=publickey private.com
[21:54] <Doonz> Permission denied (publickey,password)
[21:54] <Doonz> thats server 2 to server 1
[21:55] <Doonz> when i run that command on server1 to server2 it logs me in
[21:56] <kisielk> Doonz: check the permissions on your files in .ssh on server1
[21:56] <Doonz> identical on each server
[22:01] <ChrisRut> Doonz: can you ssh from 2 to 1 without using a key (just password)?
[22:02] <kisielk> is the key in your authorized_keys?
[22:21] <ivoks> nice
[22:21] <ivoks> i've tested pacemaker with ip and service failover and it works great
[22:21] <ivoks> easy to set up, easy to configure and runs quite nice
[22:22] <ivoks> when i think of it, i don't know how to set up the same thing with RHCS :)
[22:28] <jiboumans> ivoks: that's pretty good news :)
[22:29] <ivoks> yeah...
[22:29] <ivoks> i need to do more testing for final conclusion
[22:29] <ivoks> but, at the moment, it looks awesome
[22:30] <jiboumans> ivoks: glad to hear it. keep us appraised of progress, we're all quite interested
[22:30] <ivoks> sure
[22:32] <dug_> I upgraded my server to karmic, works fine. Removed the 2nd hard drive tho (empty), and now it won't boot, just a flashing cursor. Holding shift doesn't load grub, I checked https://help.ubuntu.com/community/Grub2
[22:33] <dug_> I ran grub-install from an 8.10 cd, no change.  I'm burning a 9.10 cd now, will that install grub2?
[22:38] <dug_> oh i see the instructions here now at the bottom of the page https://help.ubuntu.com/community/Grub2#Reinstalling%20from%20LiveCD
[23:00] <Doonz> fix it home dir was chomd to 777
[23:29]  * kirkland hugs mathiaz
[23:29] <kirkland> my pxe UEC install setup totally kicks ass now