[00:03] <Daviey> zul: So we need to add, log_host=,log_port= to the kernel command line for the default preseed.
[00:03] <zul> hallyn: running it now
[00:03] <zul> Daviey: why?
[00:04] <Daviey> zul: To add logging of the instal back to rsyslog.
[00:05] <zul> Daviey: okies
[00:06] <adam_g> roaksoax: ping
[00:06] <Daviey> zul / roaksoax: Does cobbler in precise work?
[00:07] <zul> i dont see why not....whats wrong?
[00:07] <adam_g> Daviey: ive just finished fixing packaging bugs enough to get all of orchestra installed, im about to test an all-precise setup
[00:08] <Daviey> adam_g: in the lab?
[00:08] <adam_g> zul: the last cobbler upload was broke
[00:08] <zul> adam_g: well crap..
[00:08] <adam_g> Daviey: no
[00:08] <Daviey> adam_g: Would it be a PITA for you to do that?
[00:09] <adam_g> zul: but it got fixed today bug #891527
[00:09] <Daviey> I'd quite like to develop on that tomorrow, based on your setup :)
[00:09] <adam_g> Daviey: none of the fixes are uploaded yet
[00:09] <zul> adam_g: ah yeah
[00:10] <adam_g> roaksoax: i just pushed changes to lp:orchestra with fixes to get it going with squid3
[00:10] <Daviey> adam_g: \o/
[00:11] <adam_g> Daviey: squid3 is blocked until fix for bug #891445 is merged + uploaded
[00:11] <adam_g> Daviey: but yeah, i'd like to get it going in the lab asap
[00:13] <Daviey> adam_g: Are you likely going to be able to get it online before you finish for the day?
[00:13] <Daviey> I'd like to crack on tomorrow where you leave off, you see.
[00:14] <adam_g> Daviey: on which? the new lab ?
[00:14] <Daviey> adam_g: legacy
[00:15] <adam_g> Daviey: hmm thats all still on oneiric, i'd need to try to get the current orchestra setup dist-upgrade'd and all
[00:15] <Daviey> adam_g: If it's not viable, don't worry.
[00:16] <adam_g> Daviey: ill see what i can get going
[00:16] <Daviey> adam_g: Great, thanks
[00:16] <roaksoax> adam_g: i fixed the last cobbler ... at least the upload I did in the morning should have
[00:17] <roaksoax> adam_g: and ok, i'll review your changes and release new orchestra
[00:17] <roaksoax> jamespage: still around?
[00:17] <adam_g> roaksoax: yeah, cobbler fixed
[00:21] <roaksoax> adam_g: cool
[00:21] <roaksoax> rbasak: ping
[00:22] <roaksoax> adam_g: btw.. have you tried to pxe boot a pandaboard lately?
[00:22] <adam_g> roaksoax: me? no
[00:22] <bkerensa> so much talk of pandaboards.... I need to get one
[00:28] <Daviey> roaksoax: have concerns?
[00:29] <roaksoax> Daviey: yeah can't pxe boot oneiric
[00:29] <Daviey> !
[00:29] <Daviey> roaksoax: Is that a last minute regression?
[00:30] <roaksoax> Daviey: idk, I haven't tried to pxe boot, just did a clean cobbler install
[00:30] <roaksoax> don't really think its cobbler/tftp
[00:30] <roaksoax> but gonna have to check that first
[00:31] <roaksoax> rebooting
[00:31] <Daviey> roaksoax: Great!
[00:32] <Daviey> roaksoax: Okay, regarding our conversation yesterday.. I was going to make the default preseed include http://pb.daviey.com/NIao/ , which is default (ie, machines that cobbler doesn't yet know about)
[00:32] <Daviey> it seemed you disagreed that the default preseed was the correct location?
[00:32] <zul> hallyn: fails at "test_CVE_2010_2237_2238 (__main__.LibvirtTestCVEs)"
[00:35] <roaksoax> Daviey: false alarm, seemed to be something with the tftp server
[00:37] <Daviey> Oh good. :)
[00:39]  * roaksoax looks
[00:40] <roaksoax> Daviey: right
[00:40] <roaksoax> Daviey: so, if you set a default preseed, every distro you import, which creates a profile, wwill default to that preseed
[00:41] <roaksoax> Daviey: so whwat I was saying is that instead of using a defualt preseed we should use a default profile
[00:41] <zul> unless you change the source :)
[00:41] <Daviey> roaksoax: Hmm, currently the default one is a Ubuntu preseed anyway, right?
[00:41] <Daviey> to do an /install/
[00:41] <zul> Daviey: really that preseed should be in orchestra
[00:42] <Daviey> zul: cobbler-enlist isn't directly part of orchestra, but part of the default iso.
[00:42] <roaksoax> Daviey: so for example. 1. machine pxe boots. 2. doesn't find 01-<mac-address> cause there's no cobbler system added. 3. launches PXE Menu 4. runs default profile (which points to default preseed) 5. add's itself to cobbler system ....
[00:43] <Daviey> roaksoax: Okay, do you know where that needs to be added as a patch, or do i need to investigate?
[00:43] <roaksoax> Daviey: let me check
[00:44] <twb> Grr, why does slapd build-conflicts against icu
[00:44] <Daviey> build-conflicts?
[00:45] <Daviey> roaksoax / zul / smoser: Would one of you mind reviewing and uploading bug 874981.. i made a mistake in my last comment, the descriptions are correct.  I am too tired to do so now.
[00:48] <roaksoax> Daviey: so we need to set the pxe_default_profile somewhere dont remember where exactly
[00:51] <Daviey> adam_g: have you touched cobbler-enlsit?  If so, can you push to bzr.. tomorrow i plan to add auto hostname default of hostname based on mac address.. and would rather not rebase.
[00:52] <Daviey> roaksoax: I know where that is..
[00:52] <Daviey> one moment
[00:52] <roaksoax> Daviey: yeah it is in a pxe template
[00:52] <roaksoax> Daviey: but I think we need to add a system named "default"
[00:53] <Daviey> roaksoax: currently we are doing:
[00:53] <Daviey>  # if no kickstart is specified to profile add, use this template
[00:53] <Daviey> -default_kickstart: /var/lib/cobbler/kickstarts/default.ks
[00:53] <Daviey> +default_kickstart: /var/lib/cobbler/kickstarts/ubuntu-server.preseed
[00:53] <roaksoax> Daviey: right and I think it should stay like that
[00:54] <roaksoax> Daviey: that's just a sample preseed that every imported distros (that creates a profile) should be pointing to
[00:54] <Daviey> roaksoax: Hmm, that isn't tftp:default file, is it?
[00:55] <adam_g> Daviey: the last change i made is still pending in MP
[00:55] <adam_g> https://code.launchpad.net/~gandelman-a/ubuntu/precise/cobbler-enlist/868492/+merge/80642
[00:55] <adam_g> before doing much else, i'd prefer to get that merged and come to consensus as to how the tools going to be used
[00:56] <Daviey> adam_g: I suck, will merge that tomorrow.
[00:56] <roaksoax> Daviey: but, when a machine tries to PXE boot, and there's no system added for that machine's MAC address. a PXE menu is provided. So we can default to a default profile
[00:57] <Daviey> roaksoax: Yeah, i don't think we want a menu, do we?
[00:57] <adam_g> Daviey: creating a hostname based of mac address perhaps shouldn't even be handled in cobbler-enlist, especially if we're going to be quering mac addr's outside cobbler-enlist, anyway. i think ti'd be better to generate the hostname elsewhere and pass it in as a parameter, perhaps via '-n / --name'  which already exists
[00:57] <roaksoax> Daviey: that's why I'm saying it should default to a default profiule that points to your preseed file
[00:58] <Daviey> adam_g: http://pad.daviey.com/orchestra-boot , is what i captured from last night.. but that could ALLLLLL change.. (input welcome)
[01:01] <roaksoax> Daviey: ok
[01:01] <roaksoax> Daviey: so I did this:
[01:01] <roaksoax> Daviey: 1. created a profile that points to *your* preseed file for orchestra-boot
[01:01] <Daviey> adam_g: Where do you think the hostname should be generated?  In Cobbler or on the node?
[01:01] <Daviey> using the first mac address seemed to make sense as a unique key, do you have another idea?
[01:01] <roaksoax> Daviey: 2. added a *default* system that uses the orchestra-boot-profile
[01:02] <Daviey> adam_g: Why does it make sense to move the mac address generation logic outside cobbler-enlist tool?
[01:02] <roaksoax> Daviey: 3. every time a system tries to PXE boot and does not find its 01-<NAC> in the tftp server, then it will use the *default* profile
[01:02] <roaksoax> Daviey: sudo cobbler system edit --name default --profile=orchestra-boot-profile && sudo cobbler sync
[01:02] <roaksoax> Daviey: is the hostname gonna be generated based on the MAC?
[01:03] <Daviey> roaksoax: What do we need to do to make that as part of the default install?
[01:03] <Daviey> roaksoax: That seemed to make sense, do you have other ideas?
[01:03] <adam_g> Daviey: because if we're going to be using another tool to gather details about the system, gathering the mac address will surely move outside of cobbler-enlist as well?
[01:03] <roaksoax> Daviey: no the mac usage for hostname generation is good, though, we cannot use it until *after* we already know the system within cobbler, which means *after* enlisting it
[01:04] <roaksoax> Daviey: and about the profile adding, we can just do that in postinst
[01:04] <Daviey> roaksoax: I was thinking BA:DD:CO:FF:EE -> "badcoffee" as the hostname
[01:04] <Daviey> adam_g: makes sense.
[01:04] <roaksoax> Daviey: yeah sounds reasonable
[01:05] <Daviey> roaksoax: Hmm, cobbler would not be generating the hostname - the node will do it and post home, as part of the enlistment
[01:05] <adam_g> Daviey: i think the more we can keep *out* of cobbler-enlist and passed in as input the better
[01:05] <roaksoax> Daviey: and I can make cobbler create the default profile that points to the orchestra-boot preseed, or I can make orchestra do that instead of cobbler
[01:05] <Daviey> adam_g: works for me
[01:05] <roaksoax> Daviey: better yet then, but I was thinking it in the way as it is with ec2 where it is the provider that assigns the hostname
[01:05] <roaksoax> Daviey: not the vm itself
[01:06] <Daviey> roaksoax: Where do you think it sits best?
[01:06] <roaksoax> Daviey: so I was thinking that maybe orchestra should assign a hostname
[01:06] <roaksoax> Daviey: presonally, I like it on the orchestra side cause we can do whatever we want there and just preseed it
[01:06] <roaksoax> Daviey: and we can easily modify
[01:06] <Daviey> roaksoax: Well the node knows the data, and cobbler/orchestra does not.. So it seems logical to generate it there and post it home.
[01:07] <roaksoax> Daviey: sure we can do it in either way
[01:07] <roaksoax> Daviey: as long as a system gets registered with hostname, mac, etc, etc and works with juju out of the box
[01:07] <roaksoax> it shouldn't really matter
[01:07] <Daviey> roaksoax: My thought was it's a preseed option "cobbler-enlist/auto-hostname true"
[01:08] <roaksoax> Daviey: sounds good
[01:08] <roaksoax> Daviey: though, in reality, administrators will wnat to name their systems
[01:08] <Daviey> (but you could also preseed cobbler-enlist/hostname string foobar)
[01:09] <Daviey> roaksoax: I'm not sure they'll care TBH.. I know most networks tend to have a naming scheme, but do people /really/ care?
[01:09] <roaksoax> Daviey: right, that's fine, but if you preseed, then how do we find out the hostnem for that particular system?>
[01:09] <Daviey> if it does the right thing out of the box, with cloud mentality and juju abstraction, it seems to be something we should worry about
[01:10] <Daviey> (note, people don't seem to care for setting hostnames in ec2)
[01:10] <twb> Daviey: because they forgot to set up /etc/hosts properly?
[01:10] <roaksoax> Daviey: right, but from the physical node point of view I believe it does care
[01:11] <roaksoax> Daviey: err they do care, because a naming system just come from the basic idea of inventory
[01:11] <roaksoax> Daviey: data center certification
[01:11] <roaksoax> Daviey: etc etc
[01:11] <roaksoax> Daviey: so I believe administrators tend to identify servers by their statically assigned hostname rather than a randomly generated one
[01:12] <dork> absolutely
[01:12] <roaksoax> i.e. all machines are named, tagged, under the inventory, etc etc
[01:13] <Daviey> roaksoax: Well yes, but what is the difference between calling servers after different types of plants and a random set of strings which is a mac address.
[01:13] <Daviey> We *could* default to dhcp assinged hostnames, but if unset - use the mac address?
[01:13] <roaksoax> Daviey: right, but the assignment of naming scheme comes way *before* the deployment
[01:13] <Daviey> This allows people to set the hostnames following a naming scheme if they want to, but fall back to mac address.
[01:14] <roaksoax> Daviey: you first design the data center, assign naming scheme, do the cable tagging, do the graphs, then you deploy the systems using the defined naming schemes
[01:14] <roaksoax> Daviey: that';s for both, machines and network devices
[01:14] <twb> Daviey: using something based on IP or mac is common, if you're an ISP
[01:14] <adam_g> hmph. cobbler on precise apparently cannot import oneiric or precise isos.
[01:14] <twb> like 4-3-2-1.stupid-domestic-customer.isp.net
[01:15] <roaksoax> twb: right, but that's for costumers, not for servers
[01:15] <dork> Daviey: as far as ISP like networks go the naming conventiion is super critical in resolving issues expeditiously
[01:16] <roaksoax> twb: servers have their own naming scheme
[01:16] <roaksoax> as well as routers/switches/etc
[01:16] <twb> roaksoax: but aren't they basically instances of an SOE
[01:16] <twb> roaksoax: they don't really have individual identity
[01:17] <Daviey> twb: heh
[01:18] <roaksoax> twb: in my experience, both servers and network devices are identified by a name in their design documents and that name is used as hostname
[01:18] <roaksoax> that's what I have seen
[01:18] <Daviey> So.. does defaulting to hostnames from dhcp, then falling back to macaddress based auto address make sense?
[01:18] <twb> I think in our prisons (where we force the SOE desktops to have names), the prison staff usually name them after their asset tracking number
[01:19] <twb> roaksoax: I thought we were talking EC2
[01:19] <Daviey> If people want to use their naming scheme, they can - by dhcpd'ing it
[01:19] <Daviey> twb: no, bare metal
[01:19] <twb> roaksoax: obviously for conventional servers, I agree with you
[01:19] <roaksoax> Daviey: right, dhcping it means that a cobbler system needs to know the hostname
[01:19] <roaksoax> twb: ;)
[01:20] <Daviey> roaksoax: depends if they are using the orchestra provided dhcpd or a legacy seperate one, right?
[01:20] <roaksoax> Daviey: right
[01:20] <Daviey> If they care that much about hostnames, they'd probably have a seperate dhcpd - right?
[01:21] <roaksoax> Daviey: but, IIRC we still need to know some information in cobbler (regardless of whether we are managing dhcp or not), since the hostname is passed to the kernel parameters
[01:21] <roaksoax> Daviey: we saw issues by not doing so
[01:21] <roaksoax> Daviey: so we fixed that, as it was a default behaviour
[01:23] <roaksoax> Daviey: and we also have to consider that juju addresses to the machines based on the hostname that is also the defined name of a system
[01:25] <twb> Just name each host in the /24 after the element number of the last octet
[01:25] <twb> e.g. unununium -> 1.2.3.111
[01:26] <Daviey> roaksoax: Erm. the minimal cobbler-enlist boot sends the hostname back to cobbler, this is either one THAT boot got back from dhcp or one generated from mac address
[01:26] <Daviey> if required, (which it should not be!), we can provide a null kernel param which is ignored on the enlistment boot
[01:26] <Daviey> twb: are you being serious? :)
[01:27] <twb> I tried that at work (as CNAMEs, not the canonical names)
[01:27] <adam_g> roaksoax: have you got a few min?
[01:27] <twb> But <boss> said it was not allowed because it was too hard to remember elements
[01:27] <roaksoax> Daviey: so if no hostname is obtained by dhcp, then one should be generated and sent back to cobbler
[01:27] <roaksoax> Daviey: then we should also provide a tool for adminsitrators to easily change the hostnames
[01:27] <twb> We only own a /24 so it would've fit neatly
[01:28] <roaksoax> Daviey: so they can provide a list of ip(or mac) with its matching hostname, and that should edit all the systems
[01:28] <roaksoax> adam_g: shoot
[01:28] <roaksoax> Daviey: but I thi nk is best if we discuss it further tomorrow as my brain doesn't wanna work anymore
[01:28] <roaksoax> :)
[01:28] <Daviey> roaksoax: it's 1:28 AM for me here.. my brain fell out of my ear some hours ago.
[01:29] <adam_g> roaksoax: im giving orchestra + cobbler a shot on precise, and importing oneiric and precise as distros seems to succeed, but nothing ends up in cobblers database. wondering if you knew of any recent changes that might affect that?
[01:29] <roaksoax> adam_g: what do you mean that nothing ends in cobbler databasE?
[01:30] <roaksoax> Daviey: he
[01:30] <roaksoax> heh
[01:30]  * Daviey EOF's the day.. Speak tomorrow chaps, have fun.
[01:30] <roaksoax> Daviey: have a good one
[01:30] <roaksoax> adam_g: as in no profile is availabel?
[01:31] <adam_g> roaksoax: as in, import returns 0 with no errors but 'cobbler distro list' shows nothing
[01:31] <adam_g> http://paste.ubuntu.com/741823/
[01:31] <roaksoax> adam_g: uuhmm interesting.. and yes... the change might be because we have a new cobbler in ubuntu that might be broken?
[01:31] <adam_g> lol
[01:32] <roaksoax> adam_g: cobbler import --name="oneiric-i386" --path=/mnt/oneiric --breed=ubuntu --os-version=oneiric<SPACE SHOULD GO HERE>--arch=i386
[01:32] <roaksoax> adam_g: TRy that again :)
[01:32] <roaksoax> adam_g: the log doesn't really import anything apparently
[01:32] <adam_g> roaksoax: bad paste, but same thing
[01:32] <roaksoax> adam_g: well I guess that new upstream release doesn't import :)
[01:32] <adam_g> orchestra-import-isos imports everything up until natty
[01:32] <adam_g> it does import, just not oneiric or precise
[01:33] <roaksoax> adam_g: ahhh then I know what the problem is
[01:33] <adam_g> roaksoax: see! thats why i asked you :)
[01:33] <roaksoax> adam_g: show /var/log/cobbler/cobbler.log
[01:33] <roaksoax> adam_g: or at least I presume what might be going wrong
[01:35] <adam_g> roaksoax: dah, ya.. oneiric and precise are missing from the list valid breeds
[01:35] <roaksoax> adam_g: yeah
[01:35] <roaksoax> adam_g: trying to fix that now
[01:35] <adam_g> roaksoax: cool thankya
[01:37] <adam_g> roaksoax: modules/manage_import_debian_ubuntu.py contains the two. im doing this on santol if you'd like to poke at it too
[01:40] <roaksoax> adam_g: modify /usr/share/pyshared/cobbler/codes.py and look for the line "ubuntu" and add the missing codes
[01:40] <roaksoax> adam_g: and stop/start cobbler
[01:40] <roaksoax> and should work
[01:41]  * Daviey raises his head.. adam_g / roaksoax: if you edit anything in-place on santol, can you document it please?  (as in, editing files directly)
[01:41] <roaksoax> adam_g: fix uploaded
[01:41] <roaksoax> Daviey: go sleep geeez
[01:42] <roaksoax> Daviey: hehe no worries the fix is coming :)
[01:42] <Daviey> \o/
[01:42]  * roaksoax yay!! pandaboard operational!!!
[01:44] <adam_g> Daviey: im importing a bunch of debs manually to install while bug fixes are pending upload. id like to tear down and rebuild from scratch with everythig from the archive once its sorted
[01:44] <roaksoax> alright
[01:44] <roaksoax> im off
[01:44] <roaksoax> have a good one
[01:44] <roaksoax> adam_g: if you find any more bugs just pm me or file them and assign them to me
[01:45] <adam_g> roaksoax: thanks
[01:45] <adam_g> roaksoax: did orchestra get uploaded with those changes?
[01:46] <roaksoax> adam_g: not yet
[01:46] <roaksoax> will do early morning tomorrow
[01:47] <adam_g> thanks andres
[01:47] <hallyn> zul: is that the only fail you get?  with a fresh precise install?
[01:48] <zul> not fresh install but yes thats the only fail i get...oh nm
[01:48] <zul> gimme a sec
[01:49] <hallyn> ok
[01:49] <zul> http://paste.ubuntu.com/741833/
[01:51] <hallyn> that looks more like mine
[01:52] <hallyn> feh i'll look again in the morning.  i'm fed up :)
[01:54] <hallyn> zul: lemme know if you have any ideas
[02:26] <adam_g> Daviey: looks like the log_host, log_port already get set for all profiles imported via orchestra-import-isos, pointing to the orchestra server
[03:01] <roaksoax> adam_g yes they do
[03:01] <adam_g> roaksoax: any reasons off the top of your head why clients get a 404 when trying to grab their preseed?
[03:02] <adam_g> roaksoax: the default kickstart (ubuntu.preseed) gets assigned for all profiles and nodes and the file is there
[03:04] <roaksoax> adam_g uhmmm i eill.have.to check is tjat in the lab?
[03:05] <adam_g> roaksoax: that is, 404 from http://$orchestra_host/cblr/svc/op/ks/system/$client_host
[03:05] <adam_g> roaksoax: its in the lab, yeah but just confirmed the same here on VMs
[03:07] <roaksoax> adam_g: let me check
[03:07] <roaksoax> adam_g: what lab machine?
[03:07] <adam_g> roaksoax: santol is the orchestra server
[03:07] <adam_g> roaksoax: sapodilla is the last box booted with a 404 (its still sitting in the installer)
[03:08] <roaksoax> adam_g: there seems to be an error in the template
[03:08] <roaksoax> adam_g: is the logging server installed?
[03:10] <adam_g> roaksoax: i haven't installed it, only the prov. server
[03:10] <roaksoax> adam_g: good catch then
[03:10] <adam_g> roaksoax: which template?
[03:11] <roaksoax> adam_g: trying to figure it out
[03:13] <roaksoax> adam_g: the preseed that is using is orchestra.preseed
[03:14] <roaksoax> adam_g: and since the logging server the module is failing to open the orchestra-ca key
[03:14] <roaksoax> adam_g: I will fix that tomorrow
[03:14] <roaksoax> adam_g: but you should be able to deploy now
[03:14] <adam_g> roaksoax: what did you do? i'm doing the same here in VMs
[03:15] <roaksoax> adam_g: orchestra.preseed coment with ##$SNIPPET('orchestra_rsyslog_obtain_keys')
[03:15] <roaksoax> adam_g: use two ## to comment
[03:18] <adam_g> roaksoax: great, thanks
[03:19] <roaksoax> adam_g: i will fix that in orchestra tomorrow
[03:21] <RoAkSoAx> wtg
[03:21] <RoAkSoAx> wtf
[03:23] <roaksoax> adam_g: btw we will have to check that the changes you made to orchestra's squid conf won't break package installation
[03:25] <adam_g> roaksoax: installing right now with those changes in place
[03:25] <roaksoax> adam_g: cool
[03:25] <roaksoax> adam_g: anyways im off
[03:25] <roaksoax> just let me know if you find any other issue
[03:26] <roaksoax> ttl
[03:26] <adam_g> roaksoax: most of them are now handled internally automatically by squid3, and no longre needed, so it should be okay
[03:26] <adam_g> cya
[04:06] <twb> Ugh, anybody remember how to get a shell in an openvz guest
[04:06] <twb> I thought "vzctl enter" but can't see that
[04:06] <twb> vzctl exec 101 /bin/bash works but only as a dumb term
[04:09] <twb> UGH, "vzctl enter" is valid, it just isn't mentioned in --help
[04:10] <koolhead17> hi all
[04:33] <Zanzacar> I am trying to follow this tutorial http://ve3emb.wordpress.com/2010/10/03/d-link-dwa-125-on-linux-ubuntu-10-04/ but I can uncompress the .bz2 file does anyone have any recommendations?
[04:34] <Zanzacar> I tried using tar xvjf but that was making everything into one large file and not a directory of files which I thought was odd
[04:39] <twb> Zanzacar: are you on lucid?
[04:41] <Zanzacar> twb: No I am on 11.10
[04:41] <twb> Maybe follow a guide for that, then
[04:42] <twb> The driver's probably pre-installed in current releases
[07:23] <ipl31> anyone have any experience with Orchestra and multi NIC servers?
[08:38] <_ruben> !feisty
[08:42] <koolhead17> _ruben, wa00
[08:42] <koolhead17> lynxman, hola
[09:01] <_ruben> koolhead17: was looking for the eol date, not running it myself or anything ;)
[09:02] <koolhead17> _ruben,  I was kind of surprised :)
[09:51] <lynxman> koolhead17: hello :)
[09:53] <koolhead17> lynxman, i can just laugh at my foolishness!! Am yet to  go home, slept in office last nyt. :D
[09:56] <koolhead17> but am happy to report few bugs and ask few questions as well :D
[12:09] <tero> how can I change eth2 to eth0 ?
[12:12] <patdk-lap> edit udev
[12:13] <patdk-lap> /etc/udev/rules.d/70-persistent-net.rules
[12:21] <tero> uuu thanx! :)
[12:21] <KHendrik> Good morning
[12:22] <tero> ok my home server is finnaly done :)
[12:23] <KHendrik> RoyK, I have one question for you I think I now understand why you prefer setting up an array on the whole drive instead of just a partition but why do you want to seperate Data and System and where do you draw the line for that
[12:24] <KHendrik> tero, nice what's your setup?
[12:30] <tero> KHendrik it is actualy funny
[12:30] <tero> I used and old laptop :D
[12:31] <tero> my friend give me an hp dv6000 laptop with broken screen
[12:31] <tero> and i mod it to be used as a server
[12:32] <tero> i even have 2 network cards (one pc-card) so it is used as a softare router
[12:33] <KHendrik> tero yeah thats kindof funny
[12:36] <Daviey> rbasak: Do you know what happend with bug 858878?  It seemed zul uploaded one of your patches, but it's not clear to me if the other was also included/
[12:36] <Daviey> https://launchpad.net/ubuntu/precise/+source/cobbler/2.2.2-0ubuntu1
[12:37] <zul> it got applied upstream
[12:37] <Daviey> zul: fancy updating the bug then? :)
[12:37] <zul> sure lemme have breakfast first
[12:37] <zul> just woke up
[12:37] <Daviey> ah, cool :)
[13:26] <zul> jdstrand: dont want to be a pain in the ass but can you review python-keystoneclient please
[13:28] <jdstrand> zul: this is going to require a security team audit, and I can assign it to myself, but it won't be completed today
[13:29] <zul> jdstrand: ok....why is it going to need a security team audit
[13:29] <hallyn> jdstrand, zul, qa_regression_testing is in fact bugged by some recent change in precise
[13:29] <hallyn> I redo the steps to create a testuser by hand, and then I can't connect to qemu:///session though I can to qemu:///system
[13:29] <jdstrand> zul: it is a deamon running on the network used for authentication
[13:29] <zul> jdstrand: afaik its just a client
[13:30] <hallyn> (I'll keep looking, just mentioning it in case it rings a bell)
[13:30] <jdstrand> that is a security sensitive piece of software
[13:31] <zul> jdstrand: ok
[13:31] <jdstrand> http://docs.openstack.org/diablo/openstack-identity/admin/content/what-is.html - there is a lot going on there
[13:31] <hallyn> hm, nm.  i did that wrong.
[13:32] <zul> jdstrand: blah ok ;)
[13:33] <tero> em... I have an philosophical question :) what is the "corret" first username when you install ubuntu server since ubuntu does not have root?
[13:33] <tero> or what do you guys have?
[13:33] <jdstrand> hallyn: I recommend you use 'sudo ./test-libvirt.py setup-all'. you might logout and back in, but that shouldn't be necessary
[13:39] <ersi> tero: there is no 'correct' first username
[13:39] <ersi> Or well, it's the username of the admin
[13:44] <hallyn> AHA
[13:45] <hallyn> qemu without kvm is not working
[14:05] <pmatulis> tero: on a server i typically create user 'ubuntu' during the install, he will have superuser rights via the sudo command (by automatically being made a member of the 'admin' group)
[14:20] <raubvogel> Which user does subversion expects to be?
[14:20] <raubvogel> s/expects/expect/ I am getting coffee
[14:26] <zul> jdstrand: er...we might be getting keystone and python-keystoneclient mixed up, kestone is the one getting the MIR, python-keystoneclient is a new binary for universe (although it will be getting a MIR eventually)
[14:29] <hallyn> zul, jdstrand, Daviey: question on qemu-kvm.  The new behavior (as of precise) is that if you want un-accelerated qemu, you have to add 'accel=tcg' to the machine specification.  It has its advantages...   do we want t oleave it like that, or patch around that?
[14:29] <hallyn> (it is the reason why qa-regression-testing of libvirt has been failing)
[14:30] <hallyn> if we patch around it, we'll be doing that forever
[14:30] <zul> what is the disadvantages/advantages of it
[14:31] <jamespage> roaksoax: around? I have a powernap question
[14:31] <hallyn> zul, advantage is that the people who ignore warnings about kvm not being configured right and then complain that kvm is slow, now get a harder (clearer) failure
[14:31] <hallyn> zul, second advantage is simply that we don't have to patch in perpetuity :)
[14:31] <kirkland> jamespage: perhaps i can help, if roaksoax is not around yet
[14:31] <zul> i like that advantage :)
[14:31] <hallyn> disadvantage, I don't know how many tools are going to be breaking as a result
[14:32] <jamespage> kirkland: pm
[14:34] <jdstrand> hallyn: it should be easy enough to update qrt-- detect in setup if we have accel capabilities, then add the necessary arguments. I can't speak to libvirt's handling of it though, but I'm guessing the package you're merging can handle it
[14:35] <hallyn> jdstrand, so you're ok with sticking with the new qemu-kvm behavior?
[14:35] <hallyn> I'll do an update for qa-regression-testing then
[14:37] <jdstrand> hallyn: I am, yes. in general people should be using libvirt with qemu-kvm in Ubuntu. that said, things like testdrive could break
[14:37] <hallyn> jdstrand, ok, thanks
[14:37] <jdstrand> hallyn: you may want your tech lead to weigh in though (I am but one opinion :)
[14:38] <hallyn> jdstrand, Daviey is mia :)
[14:38] <jdstrand> well, I guess he can weigh after the fact :P
[14:38] <jdstrand> kirkland: see backscroll. could testdrive break there?
[14:38] <jdstrand> kirkland: and hello! :)
[14:39] <jdstrand> kirkland: and anything other than testdrive that might break
[14:39] <hallyn> jdstrand, testdrive would only break for ppl using it without kvm - which would be painfully slow so i have doubts ppl do it :)
[14:46] <kirkland> jdstrand: hmm, what's changing in kvm?
[14:47] <g0bl1n> AWS with Ubuntu Server. I created a new user (in a secure shell), and created a new key in the AWS console. Can't login with that user. Any hint ?
[14:47] <kirkland> hallyn: testdrive support virtualbox, for people without kvm
[14:48] <hallyn> kirkland, have you tried out testdrive on uptodate precise?
[14:48] <jdstrand> kirkland: 08:29 < hallyn> zul, jdstrand, Daviey: question on qemu-kvm.  The new behavior (as of precise) is that if you want un-accelerated qemu, you have to add 'accel=tcg' to the machine specification.  It has its advantages...   do we want t oleave it like that, or patch around that?
[14:50] <kirkland> hallyn: i have not
[14:50] <jcook_5xdata> How can I tell if may server has been hack and sending email? i received this http://pastebin.com/SRX7fumt look at the header info. I am running failtoban
[14:50] <hallyn> kirkland, ok, tbh i have and it worked fine :)
[14:50] <kirkland> jdstrand: i think I'm okay okay with that change in behavior
[14:50] <smoser> jdstrand, so if you do not have kvm support in hardware and do not add that flag, what happens ?
[14:51] <kirkland> jdstrand: very, very few people, if anyone, uses unaccelerated qemu for i386/amd64 ubuntu
[14:51] <jdstrand> heh, people keep asking me-- I don't know :P
[14:51] <hallyn> qemu refuses to run
[14:51] <jdstrand> hallyn: ^
[14:51] <hallyn> smoser, ^
[14:51] <smoser> i think thats a bug.
[14:51] <smoser> you will break people.
[14:51] <jdstrand> well, upstream chose to do this
[14:51] <hallyn> smoser, the argument from qemu-kvm ppl is that most of the ppl that will break is the ones who are complaining that they ahve kvm support but it's slow
[14:53] <smoser> so is there a way to say "just do it" that does not lose hardware acceleration if possible ?
[14:53] <smoser> ie, does 'accel=tcg' turn it off if it could have used it?
[14:53] <hallyn> turn what off?
[14:53] <smoser> hardware acceleration
[14:53] <hallyn> -machine pc,accel=tcg will let it work
[14:53] <hallyn> -machine pc,accel=kvm,tcg
[14:54] <smoser> well at least it is possible.
[14:54] <smoser> it will break things.
[14:54] <hallyn> except that doesn't work for me, hm
[14:54] <smoser> and really, the number of people who are comlaining about "it is slow" surely is becomming less and less.
[14:54] <hallyn> ah,
[14:54] <hallyn> -machine pc,accel=kvm:tcg
[14:56] <hallyn> anyway, a patch to default to tcg if kvm is not available shouldn't be too bad, but it's an unwelcome delta
[14:57] <hallyn> comment on bug 892050 if you are so inclined :)
[14:57] <smoser> i just know that i have explicitly used 'kvm' knowing there was no hardware acceleration
[14:58] <smoser> they also do this in devstack
[14:58] <roaksoax> jamespage: i'm here now
[14:58] <jamespage> roaksoax: kirkland got me sorted - I needed to quickly disabled CPU power management with powernap for the test lab installs
[14:59] <roaksoax> jamespage: ok cool
[14:59] <kirkland> roaksoax: yeah, the cpu offlining thing seems to be causing more trouble
[14:59] <roaksoax> kirkland: yeah
[14:59] <roaksoax> kirkland: i think i will just ship it disabled by default
[14:59] <roaksoax> and SRU that
[14:59] <kirkland> roaksoax: not a bad idea
[15:00] <roaksoax> kirkland: but I wanna figure out why
[15:00] <kirkland> roaksoax: yeah, we ran into the same problem with the 64-way system we used for the openstack demo
[15:00] <kirkland> roaksoax: you'll need to work with the kernel team on that
[15:00] <roaksoax> kirkland: maybe offlining to 1 single CPU is the cause, and rather we should offline to ~4 if its a 16 CPU core?
[15:01] <roaksoax> kirkland: cause we offline CPU's and reduce the frequency
[15:01] <roaksoax> kirkland: and I'm changing back the default values to 300 for absent seconds and 30 for grace period
[15:01] <kirkland> roaksoax: okay
[15:02] <kirkland> roaksoax: i liked 30/6
[15:02] <kirkland> roaksoax: but whatever :-)
[15:03] <roaksoax> kirkland: yeah but doing so means that powernap is likely to send the server to powersave mode when important things are happening
[15:03] <roaksoax> kirkland: i.e. jamespage problem on which cloud init was running and 30 secs after it entered to powersave
[15:03] <kirkland> roaksoax: if so, then you're not monitoring the right things
[15:03] <roaksoax> kirkland: and boom it exploded :)
[15:03] <kirkland> roaksoax: then you need to monitor that
[15:03] <kirkland> roaksoax: and make sure that things don't explode
[15:04] <kirkland> roaksoax: personally, i like the agressive timings, especially on my laptop
[15:04] <roaksoax> kirkland: yeah I guess we should handle that in profiles
[15:05] <roaksoax> kirkland: so a juju powernap profile to deal with that
[15:08] <matrix3000> damn, i would love to figure out how to do cloud processing
[15:08] <matrix3000> one of these days
[15:13] <hallyn> jdstrand, smoser, well feh, i'll have to either patch libvirt or qemu, at any rate.  patching libvirt is more useful in that that fix will be needed upstream eventually.  But patching qemu will mean i don't halso have to fix 100 other tools built on qemu.
[15:16] <jdstrand> hallyn: another thing to think about-- for the lts, it might make sense to have this delta, and then remove it in 12.10. it will allow smoothe upgrades for lucid to precise and let any bugs in libvirt/qemu chake out
[15:16] <jdstrand> shake
[15:18] <zul> heads up...im going to be starting to switch the init scripts for swift to upstart
[15:21] <hallyn> jdstrand, that sounds reasonable
[15:21] <hallyn> Daviey, ^ i think i will go with jdstrand's suggestion.  patch qemu-kvm to keep tcg as a default until 12.10
[15:28] <Daviey> hallyn: sounds good
[15:29] <hallyn> kewl
[15:29] <hallyn> thanks
[15:29] <Daviey> hallyn: I don't think the inverse gives us much gain, does it?
[15:29] <hallyn> what do you mean?
[15:36] <Daviey> hallyn: if qemu is patched, we don't get an extra benefit.. do we?
[15:36] <hallyn> well, "benefit" of tools not breaking during LTS
[15:43] <mtaylor> Daviey: what's the tool you wrote to inject machine info into orchestra/cobbler?
[15:43] <mtaylor> Daviey: the discovery thing?
[15:46] <Daviey> mtaylor: cobbler-enlist
[15:46] <RoyK> KHendrik|afk: for a number of reasons, you can't boot from a RAID-5, nor can you install grub on a disk without a partition table, you don't want to touch your data in case of a reinstall, you generally want the data Somewhere Else
[15:47] <Daviey> mtaylor: I've lost the seed information to do it hands free, but it's based on:
[15:47] <Daviey> mtaylor: http://pb.daviey.com/NIao/
[15:48] <Daviey> it can be used standalone aswell.
[15:48] <mtaylor> Daviey: awesome. thanks. I'm going to pimp it a little bit at LinuxCon Brazil today
[15:50] <ikonia> mtaylor: where is linuxcon brazil
[15:51] <ikonia> I'm in Brazail now and would be interested in seeing it
[15:51] <hallyn> all right, dirty deed is done.   qemu-kvm pushed
[15:51] <mtaylor> ikonia: Sao Paulo - expo center norte - I'm speaking in an hour, so I doubt you'd be able to make it here in time with traffic
[15:52] <Daviey> mtaylor: interesting, is it being filmed?
[15:55] <mtaylor> Daviey: not that I know of
[16:06] <hallyn> ahs3, answer came in - xml files are under same copyright as the rest of netcf.  so i think the package i posted yesterday is ...  dare i say ... done :)
[16:13] <ahs3> hallyn: okey dokey.  could you please add a copy of the emails somewhere to the docs part of the package, just so we've got it written down somewhere?
[16:14] <hallyn> docs part?
[16:14] <hallyn> as in debian/README?
[16:15] <ahs3> right.  that, or somewhere in the copyright info, or in /usr/share/docs/netcf somewhere...just don't want to lose the discussion
[16:17] <hallyn> ahs3, so http://paste.ubuntu.com/742378/  as debian/README.debian is ok?
[16:18] <ahs3> hallyn: yup, that'll do
[16:19] <hallyn> awesome, thx
[16:22] <hallyn> ahs3, http://people.canonical.com/~serge/netcf-0.1.9-package-v3.tar.gz
[16:24] <RoyK> http://blog.laptopmag.com/usb-stick-contains-dual-core-computer-turns-any-screen-into-an-android-station
[16:25] <ahs3> hallyn: thx.  i'll take a look
[16:25] <kpettit> What's a good simple monitoring tool?  Nagios was the last one I've used but I hated dealing with all the config files for only a dozen or so sites/services.  Any recommendations?
[16:27] <RoyK> kpettit: imho nagios/icinga (a fork) is still good for that, and with templated configs, it doesn't take too long to set it up
[16:27]  * RoyK uses icinga for that
[16:27] <kpettit> ok, cool  I'll take a look.  Some templates would help
[16:28] <kpettit> is zabbix any good?
[16:30] <RoyK> kpettit: never tried it...
[16:30]  * RoyK tests
[16:30] <kpettit> me neither.  Better the devil you know I guess
[16:38] <jcastro> SpamapS: when are you planning to attend scale, should I plan for being there thursday night?
[16:51] <RoyK> damn - oneiric VMs won't shutdown properly - that is - they shut down ok, but never turn themselves off
[17:00] <SpamapS> jcastro: All 3 days, since its in my backyard. :)
[17:00] <patdk-wk> royk, apci issues?
[17:00] <patdk-wk> I have that issue randomly with windows vm's on qemu/kvm
[17:00] <patdk-wk> seems like the longer uptime, the more often it won't shutdown
[17:02] <SpamapS> jcastro: I think it will be easier for you if you come out Thursday night.
[17:02] <RoyK> patdk-wk: acpi, perhaps...
[17:03] <RoyK> patdk-wk: I've seen the same on physical hardware as well
[17:03] <patdk-wk> hardly see it on physical hardware, most of the time it's my fault there :)
[17:36] <matrix3000> RoyK?
[17:36] <matrix3000> RoyK: I never had that issue, and I am running about 4 of them 11.10 servers and 11.10 desktop installs
[17:37] <matrix3000> RoyK: on ESXi and Vmware Workstation
[17:55] <Guest16563> hello, how to set which map can user access at ftp?
[17:55] <Guest16563> with vsftpd
[17:56] <Guest16563> any1?
[17:56] <Guest16563> hello
[17:56] <irv> no luck :(
[17:56] <Guest16563> lol :D
[17:57] <Guest16563> geeks don't have time for newbies, right? :D
[17:57] <irv> whoops
[17:57] <tero> hi guys. Any suggetions on a cheap vps that runs ubuntu server?
[18:01] <WeissWaschbaer> my cron is not working "1 16 * * * /root/backup.sh &>> /var/log/backup.log"
[18:11] <jMCg> WeissWaschbaer: what'd you mean it's not working?
[18:14] <WeissWaschbaer> jMCg: the log file is still empty after 2 days, work when i use the same command manually
[18:14] <jMCg> WeissWaschbaer: you are aware of the fact that cron executes things in a very restricted environment?
[18:24] <Skaag> anyone remembers what the sshd package was called back in 9.10?
[18:26] <Myrtti> Skaag: openssh-server? it is pulled by dependencies if you just install ssh IIRC
[18:27] <Skaag> cool, trying
[18:27] <Myrtti> but then of course 9.10 has been EOL for quite some time
[18:42] <zul> lynxman: swift upstart got in thanks..
[18:43] <lynxman> zul: \o/
[19:05] <WeissWaschbaer> jMCg: no im not
[19:12] <aljosa> i've upgraded to 11.10 and postgres is now 9.1, 8.4 data seems still available. is there a safe (or maybe documented) way to use pg_upgradecluster
[19:12] <aljosa> to upgrade 8.4 data for 9.1?
[19:17] <jMCg> WeissWaschbaer: read up on that in cron's man pages. Also read up on stream redirection, I think you're doing that wrong.
[19:30] <adam_g> roaksoax: ping
[19:34]  * negronjl out to lunch
[19:36] <boxybrown> hey guys, do any services update /etc/hosts at startup?
[19:37] <boxybrown> I have some machines running ubuntu-desktop, and they aren't returning hostname -f correctly
[19:37] <boxybrown> i'm starting to suspect its the Avahi crap causing this, because my headless ubuntu servers dont have this problem
[19:39] <roaksoax> adam_g: pong
[19:41] <adam_g> roaksoax: thoughts on bug #891950? theres some options in there that are likely going to be need to be tweaked after import-isos is run. should we point people to the web interface? or come up with some way for people to easily update cobbler profiles/systems based on a config?
[19:42] <boxybrown> anybody?
[19:42] <roaksoax> adam_g: I don't think that it should be tweaked after import-isos is run that's why we provide defaults
[19:43] <roaksoax> adam_g: in case an administrator wants to change the behaviour, then it should edit the file
[19:43] <roaksoax> adam_g: and put the options they want
[19:44] <roaksoax> adam_g: setting the defaults in orchestra-import-isos is just a failsafe in case the config file hasn't been tweaked
[19:44] <roaksoax> manually
[19:44] <adam_g> roaksoax: the defaults we use dont' come from that config file, they are hard-coded in the script, which is run immeditely after the pkg is installed. unless someone drops a custom config file there before installing, theres no way to edit and update the settings after-the-fact
[19:44] <roaksoax> adam_g: right but that's the thing the intention of that config file is to make awareness that new options can be placed
[19:44] <roaksoax> adam_g: it is not to set the defaults
[19:44] <roaksoax> adam_g: if someone wnat's to change from default behaviour then they should edit the file
[19:45] <roaksoax> adam_g: though, even though no option is set in the file, then the orchestra-import-isos should still provide defaults as failsafe
[19:45] <roaksoax> and that's what we are doing
[19:45] <adam_g> roaksoax: its broken workflow
[19:45] <roaksoax> adam_g: i don't see it as a broken workflow
[19:46] <adam_g> if i edit the config file, there is no way of getting those changes reflected in cobblers profiles unless i remove all current profiles and re-import with the new settings in the config file
[19:46] <adam_g> theres no wa of updating profiles based on the config file
[19:46] <roaksoax> adam_g: well, then the config file is not broken, the update process is the one broken
[19:46] <adam_g> roaksoax: yes, its a broken workflow
[19:46] <adam_g> of course the config files not broken, its never being used
[19:47] <roaksoax> adam_g: right, but not caused by the config file
[19:47] <adam_g> so my question
[19:47] <adam_g> if i've installed orchestra nd the import script has run for the first time
[19:48] <adam_g> and all of my distros are imported and profiles configured, but i want to change the default kernel parameters for a given profile.. do i go to the web interface/cobbler CLI to do that, or do i eidt the config and re-run the import script with an '--update' flag..
[19:48] <adam_g> currently i either need to edit cobbler manually, or remove all the initially imported distros, edit config, and re-run import
[19:49] <roaksoax> adam_g: well that's up to the administrator, either modify manually all the profiles or edit the config and update the profiules
[19:49] <roaksoax> adam_g: which doesn't mean re-running the import
[19:50] <roaksoax> adam_g: cause it would make no sense to import everything again
[19:50] <adam_g> exactly
[19:50] <roaksoax> so what might be broken there is that there's no update process
[19:50] <roaksoax> so either provide a new binary
[19:50] <adam_g> my point, exactly
[19:50] <roaksoax> that only updates profile options
[19:50] <roaksoax> or make orchestra-import-isos update the profiles *without*
[19:50] <roaksoax> importing anything
[19:51] <adam_g> ya
[19:52] <roaksoax> adam_g: so I think it would be best to use a orchestra-import-isos --update
[19:53] <adam_g> roaksoax: +1, i'd favor an --update-profiles flag or something, but i'd rather being doing that in python directly via cobbler api rather than wrapping the cobbler CLI in shell
[19:53] <adam_g> Daviey: ^
[19:53] <roaksoax> adam_g: or orchestra-import-isos --update-profiles
[19:53] <roaksoax> yeah
[19:53] <roaksoax> exaclty
[19:54] <roaksoax> adam_g: the problem using the API is that we would need authentication
[19:54] <roaksoax> adam_g: the idea of orchestra-import-isos is to run on the orchestra system itself
[19:54] <roaksoax> adam_g: with simple sudo access
[19:54] <adam_g> roaksoax: well, i meant using the same local API that the cobbler cli uses
[19:55] <roaksoax> adam_g: I personally don't see the point of doing so because you would be "re-writting" this that the cobbler command already does
[19:56] <zul> adam_g: +1 from me
[19:56] <adam_g> roaksoax: i could be wrong but i thought the CLI basically does local xmlrpc to the API server on 127.0.0.1, which is authenticated
[19:56] <roaksoax> adam_g: right, so you sudo sudo cobbler system add etc etc
[19:56] <roaksoax> adam_g: if you do that using the local API, you haven to create the methods to add
[19:56] <roaksoax> delete
[19:56] <roaksoax> edit
[19:56] <roaksoax> etc etc
[19:57] <roaksoax> which means rewriting cobbler commands
[19:57] <roaksoax> on a simple script
[19:57] <roaksoax> that doesn't need it
[19:57]  * Daviey catchs up
[19:57] <adam_g> roaksoax: those methods are all implemented on the other end, hence xmlrpc
[19:59] <roaksoax> adam_g: right, but if you want to rewrite it in python
[19:59] <roaksoax> adam_g: instead of doing sudo cobbler profile edit --etc etc etc
[19:59] <roaksoax> adam_g: then you would do:
[20:00] <roaksoax> self.url_cobbler_api = utils.local_get_cobbler_api_url() self.url_cobbler_xmlrpc = utils.local_get_cobbler_xmlrpc_url()
[20:00] <Daviey> adam_g: Do you just want to check viability.  It sounds seem cleaner to do it via the api, and it's something we could push upstream aswell.
[20:00] <roaksoax> and etc etc
[20:00] <Daviey> There doesn't seem to be a need for it to be orchestra centric, with my understanding?
[20:00] <roaksoax> Daviey: it is a method that imports orchestra iso and makes some configurations for juju
[20:00] <Daviey> ah
[20:01] <Daviey> roaksoax: Do you think it makes sense for adam_g to sniff viability of doing it via the API?
[20:02] <Daviey> If it starts to smell more complicated than it needs to be, we revert to the shell wrapper?
[20:02] <adam_g> Daviey: there's going to be instances where i want to modify the orchestra-specific configuration bits of cobbler system/profile configuration. most of it is stored in a config file at /etc/orchestra/import-isos but there is no way to update cobbler based on the config file (after the initial import has run)  my question is basically how do we enable that? push people to cobbler's webui/cli, or create/expand our tooling to do it orchestra-specific
[20:03] <Daviey> adam_g: Is it something which needs addressing early in the cycle?
[20:03] <adam_g> it sounds like we need an upgrade path, so yeah.. shell or python? looking at dump of 'cobbler profile dumpvars --name=oneiric-x86_64', i'd much rather acccess that via python than awk/sed/grep
[20:03] <boxybrown> guys, making hostname -f work properly?
[20:03] <boxybrown> any tips?
[20:04] <boxybrown> it works fine out of the box on my ubuntu-server installs that don't have any desktop
[20:04] <Daviey> Yeha, I think we should try to avoid hacky awk/sed/grep scripts.
[20:04] <roaksoax> adam_g: the intention of orchetra-import-isos is not to handle specific profiles
[20:05] <roaksoax> adam_g: so if you wanna do something with a specific profile/distro/system, then its best to use cobbler cli itself
[20:05] <roaksoax> adam_g: my point being is
[20:05] <adam_g> roaksoax: i know, thats why im wondering if we 1, create new tooling for this or 2, just push people to cobbler webui/cli
[20:05] <roaksoax> adam_g: 2
[20:05] <adam_g> roaksoax: since you assigned the bug to youself, i was wondering what you're thoughts were
[20:06] <roaksoax> adam_g: the import_isos config file is for whenever we import the isos and we assign some values we want into *all* the profiles
[20:06] <Daviey> what is the bug #?
[20:07] <roaksoax> adam_g: so as agreed the idea is to use --update-profiles which will update the profile options in case the config file is updated
[20:07] <hallyn> zul, with the new fixed qemu-kvm, precise with current libvirt just gives me two failures:
[20:07] <hallyn> Test virt-install --location ... FAIL
[20:07] <hallyn> Test CVE-2010-2237 and CVE-2010-2238 ... FAIL
[20:07] <hallyn> zul, when you get a chance can you re-test on your precise image?
[20:07] <roaksoax> adam_g: i.e. changed the management classes for juju from ubuntu-juju-available and ubuntu-juju-acquired
[20:07] <adam_g> Daviey: last night, after installing orhcestra and registering a number of systems, i realized serial port needed to be redirected via kopts. theres a config file that has the defaults, i would have liked to modify that and run an update script rather than sifting through cobbler dumpvars myself
[20:07] <Daviey> ahh, i understand the requirement a bit better now.
[20:08] <roaksoax> adam_g: so that should update only the profiles
[20:08] <roaksoax> adam_g: if you wanna update *spepcific* profiles, then you need to do it manually
[20:08] <Daviey> roaksoax: for this to be acheieved in an automagic way, what would you propose?
[20:08] <roaksoax> adam_g:and for that you use cobbler web/cli
[20:08] <adam_g> Daviey: bug #891950
[20:08] <zul> hallyn: yeah
[20:09] <Daviey> ta
[20:09] <roaksoax> Daviey: again there's no automatic way to do things, if you want to add a kopts to a specific profile, then you do it manually, if you want to add it to *all* profiles, then you edit the config and we do orchestra-import-isos --update-profiles
[20:09] <hallyn> zul, awesome
[20:09] <adam_g> roaksoax: so, just a quick loop over all profiles and resets the various config parameters based on whats in the file
[20:09] <adam_g> ?
[20:09] <roaksoax> adam_g: yeah
[20:09] <roaksoax> adam_g: there's no need to create new tools when we do have the tools for data
[20:10] <roaksoax> for that*
[20:11] <roaksoax> adam_g: and maybe extend it in a way on which you can specify a list of profiles you want updated, rather than all of them
[20:11] <roaksoax> adam_g: but if it's just for 1 or 2 it makes no sense to have new toolset
[20:11] <roaksoax> when cobbler has its own tool set
[20:11] <adam_g> roaksoax: what we have now does a decent job of installing and bootstrapping the initial environment, but i predict we'll need more
[20:12] <roaksoax> adam_g: right, but that's administrator specific
[20:12] <roaksoax> adam_g: we only need to care about installation and providing an initial environment
[20:12] <roaksoax> adam_g: if administrators have their own hw requirenments and need to modify specific stuff for each profile/system, they have cobbler for that
[20:12] <adam_g> we'll see about that :)
[20:13] <roaksoax> adam_g: I think that orchestra-import-isos + config file, should be used *only* for initial configuration
[20:14] <roaksoax> adam_g: if you want to provide further configuration or costumization, it makes no sense to provide our own orchestra toolset
[20:14] <roaksoax> when there's already a cobbler toolset
[20:14] <adam_g> zul: thanks for the squid3 upload, you're the man-bomb
[20:14] <roaksoax> adam_g: so you want to update a specific profile with new settings in import_isos, we can do that with orchestra-import-isos --update-profiles profile1,profile2
[20:15] <roaksoax> adam_g: but if you want to make changes to profile1 and then make different changes in profile2
[20:15] <roaksoax> then makes no sense to use import_isos, does it?
[20:15] <roaksoax> because there wont be a import_isos config file for each of the profiles
[20:15] <roaksoax> but it will be just 1
[20:15] <roaksoax> that will and should work for all
[20:15] <roaksoax> if you wanna change behaviour of all, then modify import_isos and update the changes
[20:16] <roaksoax> if you want to mkae change A to profile1, then use cobbler cli/web
[20:16] <adam_g> i'd argue it doesn't make sense to use import_isos to modify existing profiles, but im just being difficult. it just occured to me, that if we disable automatically running import-isos during package installation (which we agreed at UDS), it would give people a chacne to customize the config before the initial import
[20:16] <roaksoax> if oyu want to make change B to profile2, use cobbler cli/web
[20:17] <roaksoax> adam_g: we agreed at UDS to set a debconf question to decided whether to import isos automatically or not
[20:17] <roaksoax> adam_g: but in any case, if you wanna provide debconf questions for each option and setup the environment on installation we can do that aswell
[20:18] <adam_g> roaksoax: a simple run now: yes or no? should suffice, i'd think
[20:18] <roaksoax> adam_g: yeah
[20:18] <roaksoax> adam_g: that simple question is all that we need
[20:18] <roaksoax> adam_g: but even so, i think we would still need to provide the --update-profiles
[20:18] <adam_g> ya
[20:18] <roaksoax> adam_g: to handle the case that administrator selected Yes, but later, they wanna change something in all profiles and "reconfigure"
[20:19] <roaksoax> adam_g: alright then, I guess we are in agreement
[20:19] <arrrghhh> hey all.  can anyone help me setup guest access to my ubuntu server?
[20:19] <arrrghhh> i just want the user to be able to ssh in, and run make.
[20:20] <arrrghhh> i guess a chroot jail?  i'm a noob to this, and don't want this guy to be able to blow up the server.  he's trustworthy, but i want to cover my bases if possible.
[20:21] <roaksoax> adam_g: bug #892328
[20:22] <adam_g> roaksoax: sweet
[20:22] <roaksoax> awesome then
[20:25] <arrrghhh> ok, chroot might be a little overkill
[20:25] <arrrghhh> it appears if i create a new user, by default they only have write access to their home and /tmp
[20:25] <arrrghhh> everything else is just read
[20:27] <greppy> arrrghhh: yes, that is normal behavior for most linux/unix boxes.
[20:42] <jodlajodla> hello
[20:43] <jodlajodla> can anyone help me with some settings for vftpd?
[20:44] <jodlajodla> hello, i have some questions for vftpd -> how to limit users to only one directory and give them permissions for uploading file in this directory?
[20:56] <virusuy> jodlajodla: you should use vftpd + chroot
[20:56] <virusuy> take a look at this how-to http://www.cyberciti.biz/tips/vsftp-chroot-users-limit-to-only-their-home-directory.html
[21:01] <arrrghhh> can anyone help me troubleshoot this additional user access?
[21:02] <arrrghhh> trying to setup ssh for him
[21:07] <arrrghhh> where do i look?
[21:07] <arrrghhh> he ran -vvv, it doesn't seem like the negotiation is happening
[21:07] <arrrghhh> i have nothing in auth.log, syslog or messages
[21:10] <boxybrown> guys
[21:10] <boxybrown> is there any way to set the fqdn
[21:10] <boxybrown> when network-manager is installed
[21:40] <virusuy> arrrghhh: but he can reach the server ?
[21:40] <virusuy> or not even ask for password ?
[21:41] <arrrghhh> virusuy, i got him in
[21:41] <arrrghhh> i enabled password auth
[21:41] <arrrghhh> but i'd like to fix it
[21:41] <arrrghhh> so he uses keys
[21:42] <virusuy> arrrghhh: allright then,
[21:42] <virusuy> did you copy his pub key ?
[21:42] <arrrghhh> do you know where i can look to see why his key was failing?
[21:42] <arrrghhh> yes
[21:42] <arrrghhh> to .ssh/authorized_keys2
[21:42] <arrrghhh> i don't know why there's 2...
[21:42] <arrrghhh> but whatever.
[21:42] <arrrghhh> i also removed his username@box on the end of the key
[21:42] <arrrghhh> so it's just ssh-rsa and the key - is that OK?
[21:42] <arrrghhh> (that's how my other key was)
[21:43] <virusuy> uhmmm
[21:43] <virusuy> what about if he use ssh-copy-id command ?
[21:44] <virusuy> that command helps you to install her publig key in your server
[21:44] <arrrghhh> ah right
[21:44] <arrrghhh> i'll have him try
[21:45] <virusuy> first, take a look at ssh-copy-id's manpage
[21:45] <virusuy> its quite simple
[21:45] <arrrghhh> er
[21:45] <arrrghhh> should i do it, or should he?
[21:45] <virusuy> he
[21:45] <arrrghhh> i'm assuming he should
[21:45] <virusuy> indeed
[21:45] <arrrghhh> does he have access to do that tho?
[21:45] <virusuy> he'll install her public key in your server, right ?
[21:45] <arrrghhh> i guess i can chmod the authorized_keys2 file
[21:46] <arrrghhh> i assume so.
[21:46]  * arrrghhh is a ssh-keys noob
[21:46] <virusuy> if he can log in with user and password
[21:46] <arrrghhh> i feel like every time i do this i have to relearn it lol
[21:46] <virusuy> then, he can execute ssh-copy-id
[21:46] <arrrghhh> ok
[21:46] <virusuy> arrrghhh: isn't easy at first , but then it's really easy
[21:46] <arrrghhh> lol
[21:47] <arrrghhh> i've... nvm
[21:54] <virusuy> arrrghhh: did it work ?
[21:54] <arrrghhh> sorry haven't tried it yet
[21:56] <virusuy> arrrghhh: aahh ok :-)
[22:03] <arrrghhh> virusuy, this seems too simple
[22:03] <arrrghhh> ssh-copy-id <username>@<host>
[22:03] <arrrghhh> that's it?
[22:03] <arrrghhh> lol
[22:03] <virusuy> yes, you also can use -i to indicate wich key will copy
[22:03] <arrrghhh> right
[22:23] <arrrghhh> virusuy, he's in
[22:24] <virusuy> arrrghhh: :-)
[22:24] <arrrghhh> i have re-disabled password auth, and still good
[22:24] <arrrghhh> thanks!
[22:24] <virusuy> arrrghhh: you're welcome !
[22:55] <matrix3000> anyone here using 11.10 and orchestra that could give me some tips
[22:55] <matrix3000> http://cloud.ubuntu.com/2011/10/getting-started-with-ubuntu-orchestra-servers-in-concert/
[22:56] <matrix3000> i was reading that but had questions on how exactly to setup my own cloud
[22:56] <matrix3000> and how it works
[23:03] <adam_g> matrix3000: this might be of more help: http://cloud.ubuntu.com/2011/10/ubuntu-cloud-deployment-with-orchestra-and-juju/
[23:39] <irvee> what would be a good VM server to install on my ubuntu server that would allow me to control/view the vms from a win 7 machine?
[23:43] <irvee> 10.04
[23:53] <Daviey> zul: Can you update status of bug 858878, please - thanks.