[02:16] does anyone know of a tool to check an rss feed and update a twitter account with the most recent item? [02:19] Oscar: yeah, indian students [02:20] twb, ? [02:20] You pay them $1/hr and they will just sit there watching the feeds and copy-and-paste them into twitter [02:20] ha ha. [03:34] as per http://www.netfilter.org/downloads.html#svn, I can't run "svn co https://svn.netfilter.org/netfilter/trunk/iptables [03:35] " [03:35] ok? [03:36] is there anyone who could explain http://lartc.org/howto/lartc.rpdb.multiple-links.html to me ? [03:37] seems pretty well spelled out to me, what don't you understand? [03:37] linocisco: didn't I already tell you? [03:38] twb, I am checking possible solution [03:53] is there anyone who could explain this "http://lartc.org/howto/lartc.rpdb.multiple-links.html" to me? [03:56] please don't repeat yourself [03:56] what is it you don't understand? [03:57] qman___, are you replying me? [04:12] how to ip route add? [04:12] how to ip route add according to split access in http://lartc.org/howto/lartc.rpdb.multiple-links.html ? === smw_ is now known as smw__ === smw__ is now known as smw [04:41] how to ip route add according to split access in http://lartc.org/howto/lartc.rpdb.multiple-links.html ? [04:47] how to ip route add according to split access in http://lartc.org/howto/lartc.rpdb.multiple-links.html ? [05:48] How can I disable eth0 or eth1 permanently down? [05:48] I tried to comment # in interfaces file. but it is not working. [06:48] hi [06:48] who could help me? [07:52] Question, I'm configuring my BIND server and it states to enter subdomains that I currently do not have set up, i.e: ns1.mydomain.com. Do I need to create these subdomains? [07:55] Anyone? [07:56] buehler? [07:56] http://pastebin.ubuntu.com/855049/ === smb` is now known as smb [08:00] Question, I'm configuring my BIND server and it states to enter subdomains that I currently do not have set up, i.e: ns1.mydomain.com. Do I need to create these subdomains? [08:01] <_ruben> What is "it" in this case? [08:06] http://pastebin.ubuntu.com/855089/ [08:07] hi [08:07] i got error on nslookup [08:07] parse of /etc/resolv.conf failed [08:22] parse of /etc/resolv.conf failed [08:24] linocisco: pastebin your resolv.conf [08:24] lautriv, content of resolv.conf = nameserver x.x.x.x x.x.x.y [08:24] <_ruben> each nameserver needs its own line [08:25] Question, I'm configuring my BIND server and it states to enter subdomains that I currently do not have set up, i.e: ns1.mydomain.com. Do I need to create these subdomains? [08:25] _ruben, I tried with two lines .line 1. nameserver x.x.x.x line 2. nameserver x.x.x.y. but not working [08:25] linocisco: you need to read the documentation that comes with ubuntu, for example, type "man resolv.conf" in a terminal. [08:26] Please, anyone? [08:26] linocisco: possible EOL issue [08:26] greppy, I am doing only simple and right [08:26] linocisco: also https://help.ubuntu.com/ and https://help.ubuntu.com/11.10/index.html [08:26] humungulous, what EOL ?? [08:27] what did you edit resolv.conf with [08:28] linocisco: that's the point, you are running into a simple issue, that should be answered in the documentation. Every time you run into an issue, asking in here should not be your first action. [08:28] Can anyone please help me? [08:29] hex20dec1: what is telling you to create a subdomain? [08:29] greppy, yes. I read online and only when I got problem, I ask here . because I know I can be blamed anytime [08:29] I'm going by this guide, with webmin. http://rimuhosting.com/support/bindviawebmin.jsp [08:29] ok . thanks anyway [08:29] greppy [08:31] hex20dec1: 1) webmin isn't really supported on ubuntu 2) ns1 is usually the hostname of your frist dns server, if you look at the SOA records for most domains you will see ns1, ns2, ns3, ns4, etc. [08:34] greppy: Thing is, I'm really new to this, I'm just trying to make my server work correctly, but it seems really difficult. I'm having a big confusion with this whole domains because my hosting and domains are on different places. [08:38] hex20dec1: a better option would be to set things up manually, that way you know how it is actually setup and have a clue when you do need to troubleshoot something. [08:38] hex20dec1: what version of ubuntu are you using? [08:39] greppy: You're right, I want to set it up manually, but it just really difficult, so I thought I would ease the process by doing it via webmin. I'm using 10.04 (I think) [08:39] greppy: let me just make sure. [08:39] Yes. [08:40] hex20dec1: start with this: https://help.ubuntu.com/10.04/serverguide/C/dns.html [08:40] there is a PDF version as well if you want to snag that. [08:40] Thank you. [08:41] By the way, did you get my private IM? [08:41] yeah, and no :) [08:45] greppy: Anything a bit less advanced? I'm already confused. [08:50] hex20dec1: http://ftp.isc.org/isc/bind9/cur/9.7/doc/arm/Bv9ARM.pdf [08:57] greppy: Thank you very much for your help, I really do appreciate it, but I want to setup my server not to be a server admin, it simply does not interest me, (no offense). I have created a site that needed a dedicated server due to the previous server always crashing, I built it on a shared hosting, and ever since I switched, I've been stuck. Do you know anyone that would be willing to either help me one on one, or ge [08:58] hex20dec1: do you need a dns server? or do you just need dns to point to your server? [08:58] greppy: I have a VPS at the moment and I just want everything to run as it did on my shared. [08:58] That is all. [08:59] hex20dec1: and I'm not offended that you don't want to be a sysadmin, it's not for everyone, but if you are going to have a server on the network, not knowing this stuff can make you a liability to others as well. [09:00] I will be hiring one as soon as I see income from the site, which would be very very soon. [09:00] Well, as soon as I get it up and running. [09:00] hex20dec1: like I said, what do you actually need? If you just need dns to point to your website, your domain registrar may be able to provide that, sometimes without additional charge. [09:00] It is currently running fine, I just noticed a few tiny problems, that aren't even noticeable. [09:02] greppy: If I knew what you mean by "dns to point to your website" I would answer you, but I don't really understand. [09:03] You want to see the zone records? Would that help? [09:03] ok. You understand that all that dns does is provide a name to point to an IP address, right? [09:03] Lol, I actually just did by you explaining that. [09:03] =] [09:03] ok [09:03] who is your domain registrar? [09:04] 1&1 [09:05] hex20dec1: instead of setting up a dns server, use 1&1 to point to your server. http://faq.1and1.com/domains/domain_admin/dns_settings/13.html [09:07] Okay, since you've stated that all DNS does is forward for example: mydomain.com to 23.343.345.23 [09:07] Correct? [09:07] yes. [09:07] hello all [09:07] I am using Ubuntu 10.04 LTS and have a configured a LAMP server, everything works just fine except my virtual hosts [09:07] this server is supposed to host multiple Vhosts, and this is is the configuration of one of them [09:07] http://pastebin.com/ryzxE2uK [09:07] when I access the localhost or the deimos.ma domain name I reach my hello world page without a problem [09:07] but when I try to access it with hope.deimos.ma as I setted it up in my ServerName directive of my Vhost I get an error from the browser [09:08] any help please ? [09:08] greppy: So, domain name is running fine. [09:08] Deim0s, post your virtualhost config file on pastebin [09:08] greppy: I am able to access my site, if that's all it means. [09:08] iclebyte, http://pastebin.com/ryzxE2uK [09:09] hex20dec1: what is the website? [09:09] iclebyte, I have disabled the default one [09:09] Deim0s: did you restart apache after making changes? [09:09] greppy, of course [09:09] reload then restart [09:09] Go private, I don't want to put it here. [09:09] Deim0s: had to ask, you'd be amazed how many forget that step. [09:09] Deim0s, so what doesn't work? [09:10] iclebyte, I want to access the vhost via the servename say : hope.deimos.ma [09:10] deimos.ma = localhost and it works [09:10] greppy, :) [09:10] what happens when you access hope.deimos.ma ? [09:11] iclebyte, server not found [09:11] error loading page [09:11] so you don't get anything from apache? [09:12] iclebyte, right ! [09:12] can you ping deimos.ma ? [09:12] i.e. is it setup in your /etc/hosts file to point to 127.0.0.1 [09:12] ye [09:12] yes [09:12] and can access my hello world page from deimos.ma [09:13] it works only the subdomain who doesn't work [09:14] can you pint hope.deimos.ma ? [09:14] pint = ping [09:14] iclebyte, no it gave me unkown host [09:15] you need to add hope.deimos.ma to you /etc/hosts file also =) [09:16] iclebyte, hmm and have to do so for the rest of vhosts ? [09:16] if they are on your local machine then yes. Your browser is trying to connect to hope.deimos.ma - which doesn't yet exist. [09:17] there are 2 parts to this. 1) making sure you have connectivity to the host i.e. hope.deimos.ma 2) apache knowing that when it sees an HTTP request for that host it should serve from /var/www/hope/ [09:17] iclebyte, then I will need to setup a DNS server later :s [09:17] iclebyte, nice [09:17] thanks a lot [09:18] no. you can do it in your hosts file [09:18] 127.0.0.1 hope.deimos.net [09:18] .ma even =) [09:18] iclebyte, but I need that other people in my private network access my server [09:18] iclebyte, lol ok [09:19] yes but to verify it add it to your hosts file. [09:19] ok [09:19] of course [09:19] then you've prooved your webserver is working with vhosts. then setup DNS for all your other users =) [09:19] iclebyte, thank you a lot :) [09:19] is it working? [09:20] iclebyte, of course :D [09:20] awesome. well done. [09:20] it works I will try now to setup a Dns server [09:20] iclebyte, because of you :) [09:21] also, i'd recommend logging your vhosts to different access and error log directories too. I try to isolate all my different vhosts as much as possible on our internal web app server. [09:21] makes debugging different sites easier than having apache dump everything into /var/log/apache/error_log [09:34] Does ubuntu supports apparmor profiles? [09:38] ohai___: yup [09:40] ohai___: ubuntu is the drive behind apparmor [09:41] henkjan: I'm not sure if I shall use Debian or Ubuntu [09:41] Ubuntu has upstart and apparmor [09:42] morning o/ [09:42] midday o/ [09:43] ohai___: i'm happy with ubuntu-server, running an 200+ hosts [09:44] I dunno what I should use [09:44] ohai___: it's personal choice, just try them see what works for your needs [09:44] People here say Ubuntu [09:44] People in #debian say Debian, in #centos they say CentOS [09:44] I don't know [09:44] that is natural as this is Ubuntu channel, you've got high hopes if you expect non-partial answers ;-) [09:44] ohai___: that's why it's YOUR choice [09:45] I want stable software (debian) but also newer software (apparmor (with profiles), upstart) [09:45] it's really up to your own tastes. I'd use LTS version of Ubuntu as it tends to be stabler than the every-six-months-release [09:45] ohai___: you need to work out which is more important then, [09:46] ohai___: and pick the distro which offers your priorities the best solution [09:46] Hmm I don't know [09:46] Can Ubuntu handle new hardware? [09:46] ohai___: then find out [09:46] ohai___: choosing a distro because someone says "use it" is bonkers [09:46] ohai___: no more/less then debian/centos [09:46] ikonia: true, but I don't know anything else [09:47] ohai___: then you need to use some and try to see what YOU like [09:47] I used botch [09:47] both * [09:48] then work out what you like [09:48] i don't know [10:05] <_ruben> ohai___: do you know *anything* ? :P [10:05] <_ruben> all i see is "i dunno" :P [10:17] _ruben: yes [11:10] New bug: #940172 in augeas (main) "-dev package missing Depends on libxml2-dev" [Undecided,New] https://launchpad.net/bugs/940172 [11:23] lynxman, around? [11:23] jamespage: I am indeed [11:24] lynxman: excellent! [11:24] lynxman: so - we have a small and late window of opportunity to upgrade nodejs in precise from 4.x.x to 6.x.x [11:25] however I don't want to spend time on it if no-one is interested [11:25] does this have interest for cloudfoundry? [11:26] I know it supports nodejs but not sure about version requirements... [11:26] jamespage: cloudfoundry doesn't use it, but nodejs is what the cool kids are using nowadays :) [11:26] jamespage: afaict no version requirements [11:26] jamespage: it just "imports" it [11:26] jamespage: cloudfoundry core is ruby [11:26] lynxman, right - so it just uses whatever nodejs Ubuntu us providing? [11:27] us/is [11:27] jamespage: yeah [11:28] * jamespage ponders this [11:37] I need an SSH client that tries to auto reconnect if my internet connection fails. [11:38] melvincv: http://www.harding.motd.ca/autossh/ [11:40] I installed that. how to use that? man page's confusing, as usual [11:41] 'autossh -M 1777 root@IP address' Is that all? [11:43] Uh, how about FireSSH in Firefox? [11:57] Hi. I'm using php5-cgi to handle php's in my server. I noticed that after a dist-upgrade the metapackage php5 upgraded installing the default libapache2-mod-php5. Shouldn't the installer be able to identify that I'm using php5-cgi and only update the other packages within the metapackage requiring upgrading? cheers [11:59] jandro: I'm not quite clear on your question. Surely the way for the installer to be able to identify that you're using php5-cgi only is to not use the php5 metapackage? [12:02] Is there a channel for newbie server admins? :) [12:05] rbasak: sorry, let me rephrase/clarify my question: libapache2-mod-php5 and php5-cgi aren't compatible and exclusive dependecies of php5, therefore when i installed php, I installed the php5 metapacke as well as php5-cgi in the same apt-get line, so it will select the php5-cgi and not the default libapache module, but the dist-upgrade put remove the php5-cgi and put libapache-mod... back. [12:09] Hmm. I wonder why that package exists instead of the four it depends on each using "Provides: php5". [12:10] jandro: what happens if you "apt-get install php5-cgi" now? Does it force anything apart from libapache2-mod-php5 to be removed first? [12:10] jandro: also, which releases are you upgrading from and to? [12:11] jandro: How can you be sure that it's not installing libapache2-mod-php5 because some other package depends on it? [12:14] * rbasak is thinking along the same lines as soren [12:14] ie. that there's some reason that installing libapache2-mod-php5 is needed to satisfy dependencies [12:16] Also, there's no reason php5-cgi and libapache2-mod-php5 shouldn't be able to coexist, afaict. [12:19] uh, autossh syntax guys? I need to have the connection up even if my internet connection is unstable. It does not make sense to me, the session will disconnect in that case, but I want it to stop showing me "Write failed: broken pipe") Maybe you know to make the impossible possible? ;) [12:27] i installed ubuntu-server-orchestra, and i have one node, installed profile is oneiric-x86_64-juju, i want to install Oracle Weblogic and deploy it all nodes,, how can i do that? [12:28] and juju package not installed together with orchestra-server, should i install it manually,,, [12:28] thanks [12:29] hi I'm trying to forward port 80 to another port using ufw. I have opened port 80 and added a few lines to /etc/ufw/before.rules at the top of the file http://pastebin.com/gyvfuWK2 but it doesn't seem to be working... have I missed something? [12:36] rbasak: I just installed php5-cgi and it removed libapache2-mod-php5 [12:36] jamespage: question about the reboot test mods I did.. [12:36] SpamapS, fire away! [12:36] jamespage: is it at all likely to be merged, or should I just setup my own test-rig to run the reboot tests so we get some coverage of beta1 ? [12:38] SpamapS, sorry - I thought it had - is it MP'ed? [12:38] * SpamapS should really go back to bed. :-P [12:38] SpamapS, must be early? [12:38] v early? [12:38] yeah, 04:38 [12:39] insomnia strikes whenever I give in to the sugar demons before bed ;) [12:39] jamespage: let me look again, I thought it had been sitting there [12:40] jandro: interesting. Did it remove anything else? [12:41] SpamapS, I see it - not proposed - lemme press the button (and why can I do that?) [12:41] rbasak: no, just that. [12:42] jamespage: anybody can propose to merge a branch into a project they are a comitter on.. IIRC [12:42] SpamapS, ah - makes sense [12:42] I wondered why I could delete other peoples merge proposals :-) [12:43] rbasak: would you advise against the use of metapackages when working on the server side? [12:43] jamespage: whats the plan long term? Create a new test framework next cycle? [12:43] SpamapS, yeah - the QA team are sprinting on it week after next [12:43] jandro: depends on what you want! [12:43] rbasak: just to keep things tight and controlled I mean, I'm relatively new to this so I don't know [12:44] jandro: I'd use them unless there's a specific reason not to. Otherwise you have to do too much management yourself. The point of them is to make life easier, but if they get in the way you don't have to use them. [12:45] jandro: I've just tried installing libapache2-mod-php5 and php5-cgi side by side in precise and it works fine. Installing php5 on top also works fine - it didn't remove either. [12:45] jandro: do you have any packages installed that are not from the distribution? [12:47] jamespage: awesome. It would be interesting to see how much we can align w/ OpenStack's QA team goals.. I think we share a lot of the same problem space with them. [12:48] SpamapS, I'm hoping to attend ODS for specifically that reason... [12:52] rbasak: no, It's quite a plain LAMP server [12:53] jandro: fundamentally your problem seems to be that libapache2-mod-php5 and php5-cgi have become mutually exclusive when they usually are not [12:53] rbasak: I'm using 11.10 oneiric btw [12:54] jandro: could you try "apt-get install libapache2-mod-php5 php5-cgi" and see if it gives you any detail on why they can't both be fulfilled? [12:54] rbasak: maybe dependencies are different in the LTS? [12:55] jandro: if you want to keep things tight and controlled, I'd suggest config management [12:55] jandro: puppet is quite popular. :) [12:55] * SpamapS would suggest juju.. but for 1 server.. its not the best idea ;) [12:56] rbasak: interesting, unlike the other way around, it doesn't complain about the installation of libapache2-mod-php5 regarless I have php5-cgi already installed [12:58] jandro: I just tried oneiric and am getting the same results. They don't seem to be mutually exclusive at all. I suspect you have other packages installed that are causing the issue, but I'm not in a position to track down which [12:59] rbasak: thanks a lot for your help [12:59] jandro: no problem. If you can find steps to reproduce, please let me know. [12:59] rbasak: I will closely look into it next time there's an upgrade of the metapackage php5 [13:00] jandro: can stick the results of "dpkg-query -W" into a pastebin? "apt-get install pastebinit && dpkg-query-W|pastebinit" or something. [13:01] Tricks_: did you also open port 8124? [13:01] jamespage: so.. if my changes are merged, then we'll also need to add those post-reboot-tests [13:02] SpamapS, thats not to hard - once it gets merged and in the lab I can template a new set of jenkins jobs to setup the testing. [13:02] Tricks_: (eg, sudo ufw allow 8124/tcp) [13:03] rbasak: http://pastebin.com/B1vJWRDU [13:09] smoser, Are the daily cloud images only available via http or is there some rsync 'able place (just wondering whether I could optimize the way to keep up to date) [13:18] smb: cloud-images.ubuntu.com lets you rsync. [13:20] jandro: how do you have libgd2-xpm and libgd2-noxpm installed at the same time? They conflict with each other. [13:20] soren, Ah. Stupid me, should not have used the http prefix in the script then. Thanks [13:27] smb, http://paste.ubuntu.com/855353/ [13:27] thats what i use to mirror. [13:27] it tries to pre-populate dirs, which saves some io [13:30] smoser, ok. not sure I already understand the pre-poulate part. But I will check the script. :) [13:30] it copies one daily to the next. [13:30] and then rsyncs [13:30] smoser: what does "[racb] merge his changes with cloud-stash-sandbox: TODO" mean? [13:31] assuming that some of the content between them is similar. [13:31] rbasak: no idea to be honest [13:31] rbasak, you had some scripts that did what cloud-stash-sandbox did. [13:31] you thought maybe you should merge them into that (kirkland's script) [13:31] it can be postponed easy [13:31] smoser, Ah, somehow I' assumed rsync being clever about that as long as I keep the old dailies around [13:31] smoser: ah, I see [13:32] smb, well, it wont know across directories. === Ursinha_ is now known as Guest31670 [13:32] it doesn't compare all possible files to all other files. [13:32] it just does those in the same path [13:32] smoser: thanks, I'll have a think about that [13:32] (and 'current' is a symlink and it recognizes that) [13:32] rbasak: I recall installing php5-gd when implementing Drupal, those libraries might be related to that maybe [13:32] rbasak, honestly, i'd just postpone it [13:33] smoser, Ah, right. my approach was just to look at current anyway. So I would adapt a bit [13:33] smoser: oh yes of course. I keep forgetting we have a cycle :-) [13:34] smb, since you're here... [13:34] http://paste.ubuntu.com/854831/ [13:34] smoser, smb: how about publishing zsync files? [13:34] that wasn't captured from a fail... but just a ps after umount [13:34] (for the dailies) [13:34] rbasak, i forget why we dont bother with that. [13:35] rbasak, it might make more sense now... honestly the only thing that likely does very well with using previous is the new '-root.tar.gz' files. [13:35] the rest of them are filesystems [13:35] which basically mean very little similar content in similar spaces in the image. [13:35] smoser: AIUI, if rsync can help, then so can zsync [13:36] (and if zsync can't help, then neither can rsync) [13:36] yeah. [13:36] but... [13:36] i dont know. [13:36] for the -root.tar.gz, you'd have to use --rsyncable, but that'd be sensible [13:36] it might be worth it. [13:36] we do use rsyncable [13:36] awesome :) [13:36] but we do not completely sort the contents. [13:36] i only sort the top level directories. [13:37] perhaps we should try to force a sorted contents. [13:37] smoser, Was that only ps w? There is a lot of udev but strangely not what it would probably call... [13:37] that'd 'make them do well. [13:37] smb, 'ps w' is just about all you have in busybox ps [13:37] (at least in the one in the initramfs) [13:37] it is a list of all processes though [13:37] (ie, dont worry about lack of 'a') [13:38] yeah... so that was on a reboot (which would always fixt the problem) [13:38] smoser, Ah, ok. Well at least it show quite a bit of udev activity, so one or the other may get into growfs's way [13:38] so, the only thing that looked odd in there was the ext4 process [13:38] which, at that point, there are no ext4 filesystems mounted. [13:38] but that didn't mean much, just that it maybe hadn't been reaped. [13:39] the udev, was explained to me that they're just forked in the initial fury of udev events. [13:41] Hm ok. And yes, the ext4 threads just may lag behind on being cleaned. [13:44] jandro: as far as I can tell, the issue is on your system. I can't reproduce. That you've managed to install two conflicting packages without noticing suggests this too. [13:47] rbasak greetings.. was curious if you had a chance to try juju on arm yet [13:47] rbasak: ok, thanks, i won't bother you anymore, I'd look into those packages. thank you! [13:47] hazmat: hello! I have it running in a local environment on LXC. That seems to work. [13:48] rbasak, awesome, good to know [13:48] hazmat: I also have cobbler working, so the next step will be to try juju's orchestra provider. I was about to start that today actually. [13:48] rbasak, cool, i'm looking over some of the maas provider work today, but from an arm verification perspective it should be equiv to cobbler [13:49] hazmat: I had been working around bug 914392 with a stragetically placed 'sed -i'. I need to verify that it works without my workaround now that the bug is fixed. [13:49] Launchpad bug 914392 in juju "LXC local provider does not respect 'series' (only installs oneiric)" [High,Fix released] https://launchpad.net/bugs/914392 [13:50] hazmat: am I right in thinking that I'm supposed to use cobbler to netinst the images first and then mark them as available and then shut them down, and then juju will pick them up? I didn't quite understand the expected lifecycle/workflow around the orchestra provider. [13:50] s/images/machines/ [13:54] New bug: #939212 in ec2-ami-tools (multiverse) "AMI created from latest i386 Lucid image fails to boot" [Undecided,New] https://launchpad.net/bugs/939212 [13:57] rbasak, i'm not really clear on the initial orchestra setup (sans juju), my understanding was that you register the machines with orchestra, set them up for pxe/netboot, and add a management class (to be spec in juju's environments.yaml).. the machines need to be powered down before handing over juju, wrt to getting images into orchestra i'm not clear, but this is the right place to ask [13:58] hazmat: I'm clear on how to set orchestra up, just not how juju claims and recycles machines. [13:59] hazmat: when juju gets a machine, it expects it installed and with a clean slate? And juju never reboots (currently), so does that mean that we just want to netinst and provide a clean slate on every reboot? [13:59] hazmat: in which case, when juju does support reboots, what happens? [14:00] rbasak, SpamapS has a nice explanation of it here http://askubuntu.com/questions/102874/juju-ssh-setup-in-orchestra-environment effectively juju searches for machines with the management class specified in juju's config. wrt to install, ideal is clean installs on boot, since we're not using lxc isolation on physical machines [14:01] rbasak, reboot support just landed, that's a slightly different use case though.. the inst on boot is for reuse of the machine, a reboot wants to keep the same identity through the powercycle [14:02] rbasak, atm i don't know that orchestra is setup to reinstall on a powercycle.. most of the efforts this cycle have been going towards the maas frontend, i think Daviey knows a bit more about the setup [14:03] rbasak, yes.. when juju gets the machine it expects it installed with a clean state (enough to run cloud-init at min which it uses to install itself) [14:04] hazmat, Daviey, SpamapS: ok, so this matters a lot on ARM as how we trigger a netboot or switch between netboot and local boot is different to Intel [14:04] I need to make sure I understand this fully [14:10] rbasak, juju removes netboot from the machine when it acquires/uses it [14:11] rbasak, it will re-enable netboot when its not using the machine [14:11] hazmat: OK, so cobbler on ARM needs to add and remove netboot correctly then. This could be interesting :) === masACC is now known as maswan [14:13] rbasak: will the box just not boot w/o PXE? [14:14] SpamapS: sort of :0 [14:15] SpamapS: when cobbler removes netboot on Intel, what happens? I assume that it's not using IPMI and actually netboots it but gets it to chain load local? [14:15] it can if you tell cobbler how to make an ipmi request [14:16] rbasak: so the pxe boot flag makes cobbler no longer boot into the "special" system-specific installer [14:16] rbasak: but if the system is still "PXE first" then it will pick up the default PXE profile, which is the menu.. which then boots from disk. [14:18] SpamapS: so right now, on a panda, the installer will rewrite the SD card so that it is no longer PXE first. At which point cobbler cannot switch it back to "PXE first" without shelling in to rewrite the SD card. But that's panda specific, and probably not worth working on. ARM server management hardware will need to support this. [14:19] Although it may be possible to seed the installer to not install a bootloader onto SD, pxelinux menus don't work on ARM, so the cobbler default target does not either. [14:20] rbasak: so, effectively, there's no way to reclaim a panda automatically. Bummer. But I have to think something more robust will have a way to remotely say "boot me again" [14:21] rbasak: or more specifically "net boot me again" [14:22] SpamapS: I've been asked by a vendor about what we need in this regard, and a way to remotely say "netboot me again" is near the top of the list. Right after "netboot capable" :-) [14:23] But now I'm worried about integration with cobbler. It's going to be awkward because cobbler is assuming a load of Intel-only mechanisms right now (which is reasonable of course but makes it awkward for me) [14:24] Incidentally, pandas can do usb boot and then I can make it do what I want, but this isn't integrated. I have played with it and could integrate it, but don't think it's worth the effort. [14:24] rbasak: its python. hack-xtend it :) [14:24] (it's be highly panda-specific) [14:25] SpamapS: it's not so much missing code, but fundamental state architecture. [14:25] rbasak: the translatio of the pxe boot flag to something that pxe boots a machine is surprisingly trivial.. [14:26] by hooking the power on command, I suppose? [14:26] well right now all it does is cause the 'sync' command to write out templates differently [14:27] its up to tftpd and dhcpd to pick those up and turn them into network actions. [14:27] so that's assuming that the pxe boot flag is a part of the machine configuration state that will take effect on next boot [14:27] rather than pxe boot being an action [14:27] its a flag that is accessible whenever any action is taken.. [14:28] so if you take the poweroff/poweron action.. you can see it.. and act accordingly [14:28] and when the machine is rebooted from inside, say by juju, which changed the pxe flag and then rebooted the machine and expects some behaviour as a result? That won't work. [14:29] rbasak: juju wouldn't change the flag and reboot the machine. juju would change the flag and tell cobbler to turn it off/on [14:29] So then we're relying on specific behaviour that happens not to trigger the perfectly valid-to-do but broken action. [14:30] sorry you lost me there [14:30] if you don't define a power control method for the profile.. then juju just *waits* [14:30] it assumes the machine is gone after telling cobbler to turn it off. [14:30] well.. off/on [14:31] rbasak: again, the pxe boot flag is more "reinstall me" than "net boot me" [14:32] rbasak: so if you have mgmt software that can netboot the machine, then on power on.. if you see that pxe boot flag.. you should do the netboot management software step then. [14:39] hey roaksoax, can we talk about "ARM Preseed fiel review (BOOT!)"? What do I need to do? === nxvl_ is now known as nxvl [15:14] Is cobbler enlistment documented anywhere? [15:14] Hello! I wanted to ask about the status of vm-builder in precise [15:15] has it been deprecated for good - is there a replacement? [15:15] or will it land in LTS and be supported 5 years? === Owner_ is now known as Owner [15:36] Yann2: live-builder seems to be the preferred tool for what vm-builder used to work for. [15:37] soren, last time I looked, live builder seemed to be to create .iso CD images? not vm images? [15:41] soren, ubnutu got a cloud offering based on openstack - what would be the recommended way to script the creation of 10 vms? surely there is another way than pxe or to boot from the cd? :) [15:43] ah sorry I forgot you werent working for canonical anymore :( [16:10] anyone seeing weird permissions issues with mysql on precise? I swear I granted the correct bits to my users, but all of a sudden I can now connect as any user without passwords :) [16:12] [16:12] jMCg: that was the enter key. [16:16] Hrm, in this byobu session I have, there is something new that wasn't there yesterday...between the distro name and the uptime, I see "4!" === Ursinha_ is now known as Guest56087 [16:26] New bug: #940420 in munin (main) "munin-node crashes when enabling tls. fix by updating munin packages to 1.4.5 or higher" [Undecided,New] https://launchpad.net/bugs/940420 [16:32] Hi all. I have a quick assumption I'd like verified, if possible. When running LXC containers (ephemeral ones in particular), I assume that any CPU they consume will be reflected on the host (e.g., in top). Is that assumption valid? [16:51] yes [17:01] Yann2: What I'd recommend these days, actually, is to use the cloud images Ubuntu publishes and customise them at boot time using cloud-init. [17:02] cloud-init, I ll have a look thanks :) [17:05] hallyn: I'll assume that that "yes" was directed at me (correct me if I'm wrong). [17:05] benji: sorry, yes [17:05] :) [17:05] hello all I'm newbe here [17:06] may be someone can help me [17:07] I've created an ubuntu server box, booting from USB key, with 3 disks in RAID5 with mdadm [17:07]