[00:35] where would I go to talk about beowulf clusters or any linux cluster information? [00:40] joebob777as7, maybe the best place to talk about clustering would be in mailing lists :) [00:40] joebob777as7, what do you need to know? [00:42] wondering how well vmware esxi or esx server would work on a clustered linux box :) couldn't find anything online [00:44] joebob777as7, so you want to create multiple VM's and create a cluster based on those VM's or something like that?? Otherwise.. you'll just have to try it and test it :) [00:45] RoAkSoAx, no I want to build a hardware cluster for more processing power and then install esx on the cluster to have vm's as a way to build a 8 or 16 core server without paying for the server hardware [00:46] joebob777as7, well i guess it will work since esx suits for that... you just need to give it a try ;) [00:46] yeah I just didn't want to buy the hardware to find out it didn't work :) [01:02] New bug: #271115 in samba (main) "Automatic upgrade in Ubuntu Hardy breaks Samba - 9-13-08" [Undecided,New] https://launchpad.net/bugs/271115 [02:07] I'm seeing a "426 upgrade required" for cups on a gutsy server. I found bug 44931, which says that it's fixed. I can see the directory and the symlinks that are described here: https://bugs.launchpad.net/ubuntu/+source/cupsys/+bug/44931/comments/14, but the cupsys user does not exist. I'm confused why this is the case. Suggestions? [02:07] Launchpad bug 44931 in cupsys "CUPS tries to auto-generate SSL key, fails" [Medium,Fix released] [02:07] Launchpad bug 44931 in cupsys "CUPS tries to auto-generate SSL key, fails" [Medium,Fix released] https://launchpad.net/bugs/44931 [02:09] I also note that the /etc/cups/ssl has these permissions: drwx------ 2 root lp 4096 2008-03-29 09:36 /etc/cups/ssl [02:09] Launchpad bug 4096 in meld "meld: merge new debian version" [Medium,Fix released] https://launchpad.net/bugs/4096 [02:10] Hmm, the 'bot thinks that I gave it a bug number to munch on :) [02:13] I can change ownership permissions, update the cupsys.conf file, add a user, none of which seems an obviously "correct" fix. [02:31] I've now compared this to another gutsy install, which shows that the cupsys exists there, but the second-working-gutsy is an upgrade from edgy-feisty-gutsy, where the non-working server was a fresh install. I'm still no closer to any idea why I'm seeing this, or how best to fix it. [02:32] s/that the cupsys/that the cupsys user/ [03:08] Hi server people :) [03:09] I wrote a how to about getting a upnp server together for the xbox 360 sharing media [03:09] Its at http://ubuntuforums.org/showthread.php?t=848144 [03:09] However, users on the server edition of Ubuntu cant compile it [03:09] And I dont know why, they meet the dependencies [03:11] We tried reinstalling a dependant lib but that didnt work [03:28] nullack: Any reason you're not just installing gmediaserver ? [03:29] owh - please gimme a few mins to answer your Q properly [03:29] Sure [03:31] owh - MS implemented a non standard upnp for the 360 [03:31] owh: gmediaserver does not use the special 360 mode [03:32] owh: also the amount of dev activity on gmediaserver doesnt seem as high as others [03:36] nullack: Is there a bug report indicating that gmediaserver isn't operating wtih the 360? [03:38] owh: Yes there is numerous entries on xbox 360 failures for gmediaserver [03:38] I'm asking because the packaged upnp server is gmediaserver. If it's broken people need to know. If it no longer provides required functionality and isn't being developed then a different upnp server needs to be found. None of this helps you compile your upnp server under u-server, but I'm getting to that :) [03:38] owh : The ushare server is packaged in Ubuntu [03:39] nullack: So, there are bug reports in launchpad that describe this problem? [03:39] owh : But that revision is highly broken [03:39] nullack: So, is there a bug report about that? [03:40] owh : to be clear, the gmediaserver problems with the 360 are reported yes [03:40] The challenge I have is while I want to help you fix your specific problem, right now, it doesn't fix anything long-term, so I'm loathe to do anything until I've figured out if the real - long-term problem is addressed before spending energy helping you compile a package from source. I'm not trying to piss you off, but I could stay here all... [03:41] ..day and help fix little problems one at a time, rather than make Ubuntu better, which is what I care about. [03:41] owh : I understand fully and I support your principles fully [03:41] Cool [03:41] owh : Ive been down this road [03:42] So, my second question was, is there a bug report about ushare? [03:42] owh : I cant submit a please update ushare right now due to ushare not having a new release [03:42] Ah [03:42] owh : upstream are slow [03:42] owh : I cant even compile the SVN of upstream [03:42] So, if you were to download the ushare source from ubuntu, that is apt-get src, can you compile that? [03:43] owh : the ushare revision in Ubuntu is broken [03:43] owh L there is patches, which I use [03:43] nullack: If the upstream SVN isn't compilable, then there are some major issues with their development process IMHO. [03:43] owh : totally :) [03:44] Right, so, the reason I'm asking about the source package within Ubuntu, is that it has most likely got a bunch of build-depends, so it can be compiled. [03:44] owh : Yes it can be, but the thing is, that revision is broken [03:44] I'm getting to the bit where you install those, then compile to prove that it works, then patch it, then have success. [03:45] So, lemmie get this straight. You can compile the ushare source from ubuntu, on ubuntu-server, regardless of the actual application being broken? [03:45] I dont have Ubuntu server, I dont know about compiling on Ubuntu server [03:45] It's all the same packages [03:45] nullack: There is very little difference where you'd notice. [03:45] Yup [03:46] hads : I know, and when we query apt-cache it says the dependencies are met [03:46] Even reinstalling the dependent lib doesnt work [03:46] nullack: Step back a moment. You told us that you cannot compile. What causes you to say that? [03:47] owh : its here http://ubuntuforums.org/showthread.php?t=848144&page=4 [03:47] owh : It does compile on Hardy and Intrepid desktops [03:49] Right, first of all, none of those comments show how they actually installed the source and it doesn't look like they did so properly. Gimmie a mo... [03:49] owh : thanks [03:49] `apt-get -s build-dep ushare` whill show you the dependancies [03:50] I haven't read that thread to see if you've done that or not. [03:50] That was what I was looking for hads, thanks. [03:50] The -s is a dry run BTW [03:50] * owh nods [03:50] The following NEW packages will be installed: [03:50] autoconf automake1.7 autotools-dev cdbs fdupes intltool libavcodec-dev [03:50] libavformat-dev libavutil-dev libdc1394-22-dev libdlna-dev libdlna0 [03:50] libgsm1-dev libraw1394-dev m4 [03:50] 0 upgraded, 15 newly installed, 0 to remove and 0 not upgraded. [03:50] Inst m4 (1.4.11-1 Ubuntu:8.10/intrepid) [03:50] Inst autoconf (2.61-7ubuntu1 Ubuntu:8.10/intrepid) [03:50] Inst autotools-dev (20080123.1 Ubuntu:8.10/intrepid) [03:50] Inst automake1.7 (1.7.9-9 Ubuntu:8.10/intrepid) [03:50] Inst fdupes (1.50-PR2-1 Ubuntu:8.10/intrepid) [03:51] Inst intltool (0.40.1-1 Ubuntu:8.10/intrepid) [03:51] Inst cdbs (0.4.52ubuntu7 Ubuntu:8.10/intrepid) [03:51] Inst libavutil-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] ;-( [03:51] Inst libgsm1-dev (1.0.12-1 Ubuntu:8.10/intrepid) [03:51] Inst libraw1394-dev (1.3.0-4 Ubuntu:8.10/intrepid) [03:51] Inst libdc1394-22-dev (2.0.2-1 Ubuntu:8.10/intrepid) [03:51] Inst libavcodec-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] Inst libavformat-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] Inst libdlna0 (0.2.3-0ubuntu2 Ubuntu:8.10/intrepid) [03:51] Inst libdlna-dev (0.2.3-0ubuntu2 Ubuntu:8.10/intrepid) [03:51] Conf m4 (1.4.11-1 Ubuntu:8.10/intrepid) [03:51] Conf autoconf (2.61-7ubuntu1 Ubuntu:8.10/intrepid) [03:51] Conf autotools-dev (20080123.1 Ubuntu:8.10/intrepid) [03:51] Conf automake1.7 (1.7.9-9 Ubuntu:8.10/intrepid) [03:51] Conf fdupes (1.50-PR2-1 Ubuntu:8.10/intrepid) [03:51] Conf intltool (0.40.1-1 Ubuntu:8.10/intrepid) [03:51] Conf cdbs (0.4.52ubuntu7 Ubuntu:8.10/intrepid) [03:51] Oh dear [03:51] Conf libavutil-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] Conf libgsm1-dev (1.0.12-1 Ubuntu:8.10/intrepid) [03:51] Conf libraw1394-dev (1.3.0-4 Ubuntu:8.10/intrepid) [03:51] Conf libdc1394-22-dev (2.0.2-1 Ubuntu:8.10/intrepid) [03:51] Conf libavcodec-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] Conf libavformat-dev (3:0.svn20080206-12ubuntu1 Ubuntu:8.10/intrepid) [03:51] Conf libdlna0 (0.2.3-0ubuntu2 Ubuntu:8.10/intrepid) [03:51] Conf libdlna-dev (0.2.3-0ubuntu2 Ubuntu:8.10/intrepid) [03:51] hads: This just proves our point :) [03:52] Im confused, sorry [03:52] Firstly, please don't do that again :) [03:52] !pastebin [03:52] pastebin is a service to post multiple-lined texts so you don't flood the channel. The Ubuntu pastebin is at http://paste.ubuntu.com (make sure you give us the URL for your paste - see also the channel topic) [03:52] Ok, why? [03:52] Oh, ok, sorry [03:52] nullack: The list that you just pasted is the list of things that are needed to build this application properly. [03:53] Right, dependencies [03:53] So, your howto is broken :) [03:53] For server? [03:53] In general, it's probably just the case that those dependancies are pulled in already on the desktop [03:53] nullack: Well, no, for the actual build of ushare. [03:53] Indeed. [03:54] How do I determine what packages are not installed for the server build? [03:54] by default :) [03:54] nullack: apt already worries about it, you should not. [03:54] nullack: You should install the build-deps for ushare, then proceed from there. [03:54] So you need to add that to your howto. `apt-get build-dep ushare` [03:54] nullack: The maintainer for ushare has already done this work for you :) [03:55] I know apt-get wont install packages allready existing so I guess theres no harm in putting them all there [03:55] nullack: Yes there is harm. [03:55] How so? [03:55] nullack: In a future universe, your howto will be incorrect, but apt-get build-dep won't be. [03:56] Right, smart [03:56] I get it [03:56] nullack: If it all of a sudden depends on shark-fins, your howto will need to be maintained, but the build-dependencies will already be up to date. [03:56] Thanks very much [03:56] Pleasure. [03:57] One thing though, why did it complain about a missing package that isnt actually missing? i.e. its missing something else [03:57] A program error? [03:57] nullack: Because the ./configure command is trying to make do with what it has. It's not that smart. It can only test for things it knows about. [03:57] Right [03:58] nullack: Autoconf and automake are smarter, but they were missing :) [03:58] owh : I promise as soon as Ushare gets a new release I will contribute to the Ubuntu long term future by following it being upgraded [03:58] Thanks to all who helped, legends :) [03:58] So, use the build-dep and your world will be happier :) [03:58] Yes :) [03:58] Good on you nullack :) [03:59] nullack: You are already doing it by asking questions, learning and writing howto's. [05:28] rubble rubble [06:06] Real quick question, when setting up a raid 1 system, does the drives the you want to "raid" need to unmounted? [06:12] didymus7777: Uhm, well, given that you're going to format it, yes. [06:12] didymus7777: I don't know how you would create a raid device without formatting, but I might be wrong. [06:13] OK that helps, kinda, my problem is though Its a clean format [06:13] and I thought I set it up right. [06:13] didymus7777: How did you set it up? [06:14] well I have to hard drives drive a / and swap. Drive be same size but linux raid for first partion and linux raid same size for swap [06:15] But I have a hunch that when I set up the first drive I did it wrong [06:16] didymus7777: Did you do this from a command-line, from an installer, using magnets -- how did you set up raid? [06:16] New bug: #271184 in openssh (main) "ssh client not using correct identity key" [Undecided,New] https://launchpad.net/bugs/271184 [06:17] Actally I used webmin, [06:17] Aaargh. [06:17] All I need is you to tell me to start over and I will :) [06:17] No problem. [06:17] I can and do use the command line [06:17] didymus7777: Well, before you do any of that... [06:18] Idea!! cool like thouse [06:18] didymus7777: I don't know how you'd use webmin on a running system, so there must be something you're not telling us. [06:19] At the moment using webmin to do this sounds like you used magnets :) [06:20] Well, the one thing that I didnt tell you is Yes I am new to this raid thing, and I would have to say yes I am using magnets ... I think. I do aplogise for my ignorance [06:21] The system IS running already, with a os on it "ubuntu server" and I thought I could just set up raid with that. [06:21] didymus7777: When I say "magnets", I mean "some magic process that you're not telling us about" - and webmin qualifies :) [06:21] I ran the install disk for the server [06:21] OH well then yes [06:21] didymus7777: Is there any actual data on the server? [06:22] other then the install no, and yes I am willing to start over agian. [06:22] Do I need sometype of hard ware to make sure this raid works? [06:22] or can it be all software [06:23] didymus7777: Well, I've just been reading a SuSE instruction set here which seems to indicate that you can do this on a running machine. For my money I'd let the installer do it, but since you already installed, have a look at it: http://linux.derkeiler.com/Mailing-Lists/SuSE/2003-11/3029.html [06:23] didymus7777: As for software raid vs. hardware raid, it depends on the load. [06:23] I am going to guess that hardware is better [06:24] didymus7777: Some swear by hardware raid and never touch software raid, others don't have the hardware and make do. [06:24] thats me [06:24] I just wanted to make sure I could be done [06:24] didymus7777: Well, better is often a matter of requirements, horses for courses if you like. [06:25] didymus7777: Personally, the commands on the URL I showed you give me the willies and I would not use them. I cannot think of a good reason to bypass the installer to achieve this. [06:25] didymus7777: I'm sure that if you ask often enough, you'd find others that both agree and disagree with me. [06:28] owh: OH that makes sence [06:28] If I was to do this on the install how would you go about doing that. Ubuntu server comes not with the mdadm software and that's all I was seeing as far as setting this up. If I set the partions up for raid it will not auto do it will it? [06:29] didymus7777: http://doc.ubuntu.com/ubuntu/serverguide/C/advanced-installation.html [06:30] thank you thats what I wanted right there. you have been a huge help [06:30] Pleasure. [06:30] I will fallow this and get it done. again thank you === freaky[t] is now known as fReAkY[t] === fReAkY[t] is now known as freaky[t] [09:01] New bug: #271254 in drbd8 (universe) "drbd doesn't start: "No response from the DRBD driver! Is the module loaded?" but module is loaded" [Undecided,New] https://launchpad.net/bugs/271254 [09:12] moin [10:23] hey [10:23] i've just installed ubuntu server 8.04 [10:24] and 'w' is not showing any users as being connected [10:24] when i'm connected from ssh [10:24] i have an identical setup on another box and it shows the users that are logged in === PanzerMKZ_ is now known as PanzerMKZ [10:51] anyone know how to setup remote desktop from windows xp to ubuntu server 8.04? [11:01] Popz: not really. You can use VLC which is enabled by default [11:01] s/VLC/VNC [11:01] sorry :) [11:02] (but ubuntu server has no x server by default BTW) [11:02] no like windows server one [11:03] so, you have to setup one if you really want to have a "graphical" access to your server [11:04] Graphics are overrated. [11:04] hads: +1 [11:05] it just adds an non-value and guide you into bugs you really don't want to deal with for a server. [11:15] it'd be quite useful for some of the gameserver control apps i run [11:15] the server has about 16GB ram [11:15] so the graphics overhead is no problem :P [11:15] 8GB physical, 8GB virtual [11:19] So, 8 [11:23] hads, well said [11:24] Popz, so what are you actually trying to do? turn a server intoa desktop, or ssh into the box? [11:35] i need a visual environment for some applications, most of the time i'll be using ssh [11:37] Popz Do this: [11:38] *gimme a min while i type all the commands out [11:38] sure... :p [11:38] sudo aptitude install xorg [11:38] sudo apt-get install ubuntu-desktop [11:38] startx [11:38] sudo apt-get install gdm [11:38] sudo /etc/init.d/gdm start [11:38] Ew [11:38] i'd rather not install ubuntu-dekstop [11:38] then try gnome-core [11:39] how can i view that remotely? [11:39] Once you have gnome-core installed [11:40] sudo apt-get install vino [11:40] Then get a VNC client on whatever PC you use as your desktop machine [11:40] ok thanks [11:40] i'll try it out [12:22] <_ruben> remote x ftw [12:34] Popz: Any luck? [12:43] _ruben: ? [12:46] hello, do anybody know how can I show the owner user in for each file in svn ? [13:04] <_ruben> remote: wasnt talking to you ;) .. "remote x" ftw :p [13:34] oh [14:26] I am looking for an imap server that does not require the user to have a local user account, accounts would come from ldap [14:27] !imap [14:27] IMAP and POP are protocols for fetching email. The officially-supported server in Ubuntu is Dovecot (packages "dovecot-imapd" for IMAP, and "dovecot-pop3d" for POP) - See also !MailServer for information on the SMTP protocol [14:28] dovecot does support ldap [14:34] hmm cyrus could well be the way to go [14:37] However dovecot is the one we primarily document and support in Ubuntu. [14:37] not worth it if you can auth from ldap [14:38] It all depends on how much help you want from Ubuntu docs and such. [14:39] i just don't want to sent ages setting up something that i can't manage from a directory server [14:39] oh seems like you can auth from ldap after all [14:40] I would love to stick all my config into an ldap server === dantaliz1ng is now known as dantalizing === chuck__ is now known as zul [16:25] kirkland: about service --status-all, there are two ways imho of making the output (more) useful [16:25] 1/ get rid of everything that doesn't have a status [16:26] 2/ try to uniformize the output rather than letting all status actions echo their own thing [16:26] one way of doing it is to call only scripts that have a status action, and just checking the return code, displaying all results in a table [16:26] Koon: right... ideally, we'd have much better policies in Debian/Ubuntu as to status reporting [16:26] Koon: and these problems would go away [16:27] Koon: but I agree, that --status-all gives us 1 place to solve that [16:27] you'd still need a way of telling which /etc/init.d/script should have a status [16:27] Koon: grep "status)" usually gets it [16:28] Hello to everybody [16:28] Koon: they're almost *always* handled by a case statement [16:28] Doing it in --status-all supposes an ideal world where all daemons have a status action that returns proper codes [16:28] we might need to have a way of marking conformant scripts [16:29] I installed poptop to my Ubuntu Server and everyhing is going ok, peoplle is able to coonnect remotly with VPN to our network, however we noticed that whenever sombody connects from outside, the people at internat networs side can't access our server, Can anyone help me please' [16:29] Koon: i'm not sure i like the idea of a static list .... would be better to be more dynamic than that [16:30] a specific comment in the initscript [16:30] running ./file.bin ... just returns me to a new line. It's supposed to start a setup wizard I believe ... trying to install KnowledgeTree [16:30] kirkland: or even a specific action that would just point to status) [16:30] Koon: if we're going to touch a bunch of init scripts, i would just as soon fix them and get their status actions working properly [16:30] Now sombody is connected with VPN to our server, but I can't get ping now from the server [16:31] Koon: before running the invoke-rc.d, i suggest we grep for "status)" [16:31] Pizarro: looks like a network conflict [16:31] Nevermind, I just ran it on my desktop and it started up a GUI... server has no gui :( [16:32] then use the return code to return [Started] or [Stopped] ? [16:32] Koon: and only run invoke-rc.d if a "status)" action is found [16:32] Koon: --status-all will show the status of every daemon running on my machine or i get the wrong idea? [16:32] nxvl: it will basically just run everything in /etc/init.d with the "status" action [16:32] kirkland: so yes [16:32] nxvl: unfortunately, not all Ubuntu init scripts have status actions [16:32] nxvl: which makes up a noisy useless output [16:32] nxvl: and also, not all report status in the same way [16:33] Koon: i suggest we add a "--uniform" option, or something [16:33] nxvl: lots of scripts in /etc/init.d are not about daemons, but rather one-time boot/stop actions [16:33] Koon: and if that is passed with --status-all, then we simply look at the error code [16:33] kirkland: yes [16:33] Koon: and print [OK] or [NOT OK] [16:33] Koon: but the message saying that there is no status action goes to stdout? i think it goes to stderr, and if it is, it's easier to filter IMHO [16:34] nxvl: depends greatly on each script [16:34] Koon: well, yeah, but you understand what i meant :D [16:34] kirkland: right [16:34] nxvl: which are not required to follow a standard [16:35] kirkland: sounds good, the only drawback is that it's difficult to point out a misbehaving script that doesn't handle return codes properly [16:36] Koon: which is why i suggest we leave --status-all as is, and add a compound parameter [16:36] service --status-all --uniform [16:36] kirkland: maybe have a check mode that displays both [16:36] Koon: where we try to intelligently interpret/clean the results of --status-all [16:36] * atd is started [NOT OK] [16:37] interpret/clean/filter [16:37] that would be easy to spot problems [16:39] Koon: i don't think we're going to get any of this into Intrepid [16:39] Koon: at least i would find it highly unlikely [16:39] kirkland: oh no, it's more a general discussion -- in the meantime we should not advertise --status-all too much [16:40] Koon: fair enough [16:40] * Koon looks up an example [16:42] I seem to remember that rc-status --all in Gentoo had a pretty output [16:42] can't find it, might need to reinstall to see ;) === frith is now known as incorrect [16:45] Koon: i have a Fedora 9 KVM [16:45] kirkland: search for rc-update --all in http://tinyurl.com/4ybsv6 [16:45] Koon: its status-all output is pretty ugly too [16:45] Koon: nice [16:45] Koon: one minute, i'll pastebin the Fedora output [16:47] I mean search for "rc-status --all" (no rc-update) [16:47] * Koon should install a gentoo vm, the output was nicely right-aligned iirc [16:48] Koon: http://pastebin.ubuntu.com/47798/ [16:48] Koon: that's from Fedora 9 [16:48] kirkland: ew. [16:48] Koon: well, it's better than ours [16:48] :-) [16:48] rc-update uses runlevels to determine what should be running -- that's one way of doing it [16:49] looping through /etc/init.d catches all those manually-started ones [16:50] that said, removed packages (not purged) usually leave their /etc/init.d/ script behind [16:50] could that screw up our output ? [16:53] kirkland: ^^ [16:54] Koon: definitely [16:54] * Koon is not a big fan of LSB putting init scripts in /etc [16:55] they are not configuration files. [16:56] Koon: it wasn't LSB that invented that one :-) [16:56] Koon: that's as old as Unix itself [16:56] kirkland: but they mandate it, rigth [16:56] Koon: yeah [16:57] kirkland: i seem to remember the rationale was that links in /etc/rc.d should not cross filesystems === no0tic is now known as no0tic`ciao [16:57] Did anyone configured POPTP?????????? for VPN connections? === no0tic`ciao is now known as no0tic [16:57] poptop sorry [16:58] the manual web site is down [16:58] and I just need to know how to grant VPN clientes to see each others not oblu the server [16:58] in OpenVPN this was just a command line [17:41] The number one MS Exchange replacement Zarafa is going 100% open source [17:41] http://www.zarafa.com [17:42] Bad news for Microsoft - good news for everyone else [18:21] Hi [18:22] I have set a VPN server, the clients can connect but they only can ping the server, PLEASE I NEED HELP, thanks [18:23] anyone else having problems with Ubuntu 8.04 LTS locking VMware ESX 3.5? === chuck__ is now known as zul [18:45] kirkland: do you think it would be possible to automate the testing of Boot-on-Raid option in -server ? [19:06] mathiaz: yes, i think so [19:06] mathiaz: i'd need some help with the preseed/kickstart file for the partitioning, etc. [19:07] kirkland: right - could this be added to https://wiki.ubuntu.com/Testing/Cases/ServerInstall ? [19:07] <\sh> any heartbeat crack there? /me has a problem with using heartbeat and some bonded interfaces [19:07] kirkland: I think I can sort out the automate part of it (preseed) [19:07] kirkland: I'm more interested in the testing procedure. [19:08] mathiaz: my KVM testing is pretty throughly described on the BootDegradedRaid wiki page [19:08] mathiaz: but it boils down to "kvm -hda disk1.img -hdb disk2.img" for the installation [19:09] mathiaz: and then the two test scenarios are "kvm -hda disk1.img" simulating loss of disk2 [19:09] mathiaz: and "kvm -hdb disk2.img" simulating loss of disk1 [19:15] kirkland: ok - I'll have a look at the wiki page and see if the testing instructions could be adapted to ServerInstall [19:15] mathiaz: you'd probably want to add a script in the preseed to put something in /etc/rc.local, to signal that it's up [19:16] mathiaz: wget http://10.0.2.2/AM_ALIVE :-) [19:16] mathiaz: and monitor the apache error log or something [19:16] kirkland: right - this use case is a bit more complicated that the other. [19:18] mathiaz: other? [19:18] mathiaz: other what? [19:18] kirkland: there are other test cases listed on ServerInstall [19:18] mathiaz: ah [19:18] kirkland: eg mailserver, lamp, etc... [19:18] mathiaz: well, it would be great to automate this [19:19] kirkland: I'd like to add a test case for RaidBoot [19:19] mathiaz: right, make sure that it boots from each disk [19:19] kirkland: but the testing instruction will be different [19:19] mathiaz: oh, there is one trick... [19:19] mathiaz: let's say you boot degraded on disk1 alone [19:19] mathiaz: actually, strike that [19:20] mathiaz: what i was about to say is wrong ;-) [19:20] * mathiaz forgets everything [19:21] mathiaz: right, so the test should be to first boot with both disks attached, ensure that both are up and actively synced [19:21] mathiaz: then boot from disk1, make sure it comes up [19:21] mathiaz: then boot from disk2, same [19:21] mathiaz: that's assuming you have BOOT_DEGRADED=yes [19:21] mathiaz: which you'll define in the preseed [19:22] kirkland: right - and what happens if you boot from disk1 only after booting from disk2 ? [19:22] kirkland: actually it doesn't make sense to test. [19:22] mathiaz: this is the complex scenario i was about to mention.... [19:23] mathiaz: so let's say you boot with just disk1 attached [19:23] kirkland: rather make sure that you boot from disk2 if you boot with disk1 and disk2 the third time [19:23] mathiaz: in the OS running on disk1, mdadm will mark disk2 as faulty [19:23] mathiaz: so if you then reboot with both disk1 and disk2 attached, you will still be running degraded [19:23] mathiaz: because the system doesn't know what went wrong, and doesn't trust disk2 [19:24] mathiaz: at that point, if you know that disk2 is good, or perhaps brand new, you can add it back to the raid [19:24] mathiaz: mdadm /dev/md0 --add /dev/sdb1 [19:24] is there anyone willing to help me privatly with an ubuntu 8.04 server instaall with samba? [19:24] mathiaz: then you need to wait for the disks to re-sync [19:24] mathiaz: which will take a while, i use `watch -n1 cat /proc/mdstat` to monitor [19:24] any help would be greatly appreciated [19:24] mathiaz: basically, all of disk1 will get copied to disk2 [19:24] kirkland: right - what about the following scenario: [19:25] kirkland: Boot1. disk1 only - Boot2. disk2 only - Boot3 disk1 and disk2. [19:25] kirkland: which disk should be used ? disk2 I'd guess [19:25] mathiaz: yuck, that's a really gross scenario [19:26] mathiaz: kvm will boot off of the first disk, -hda [19:26] mathiaz: so its mbr will be read [19:26] mathiaz: system will boot, but the data on those two disks have diverged [19:26] mathiaz: so it's really in a bad, unclean, dirty scenario [19:27] mathiaz: you'd want to force a rsync (if this were a real system) [19:27] kirkland: right - but with BOOT_DEGRADED to yes what would happen ? [19:28] mathiaz: i don't think that has any effect, if i'm understanding you correctly [19:28] mathiaz: wait, i see what you mean [19:29] kirkland: well disk1 and disk2 have diverged. BOOT_DEGRADED=yes means you want to boot no matter what - which disk is used ? [19:29]