=== robbiew is now known as robbiew_ [00:10] does eucalyptus provide the EC2 meta-data service? [00:49] Ng, yes [00:50] Ng, eucalyptus does provide meta-data service, but only on 2 of the netwroking setups [01:08] how to remove check package list n ubuntu server [01:08] ruben23: Can you ask that a different way or be more specific? What are you trying to achieve exactly? [01:09] remove a package and check it list..of the package [01:10] jmarsden|work: and verify if its unsinatlled already. [01:10] ruben23: Ah, OK. To remove it you can do sudo apt-get remove PACKAGENAME [01:11] To see if it is installed, try dpkg -l PACKAGENAME and if you see ii at the start of the output, it is installed. [01:12] ruben23: If you want to remove it and purge all config info associated with it too, do sudo apt-get purge PACKAGENAME instead of remove. [01:12] dpkg -| PACKAGENAME [01:12] is that right..? [01:14] ruben23: dpkg -l PACKAGENAME where l is the lower case letter L (say "ell") [01:15] ruben23: Not a vertical bar | which is what you had. [01:15] ruben23: I'd guess you need to use a better terminal font so you can see the difference :) [01:18] jmarsden|work: it not working..http://pastebin.com/m454defb2 [01:20] ruben23: Please read what I wrote earlier. dpkg -l where "l" is the lower case letter L as in abcdefghijkl [01:20] You are using a vertical bar which is incorrect. [01:22] jmarsden|work: ok [01:26] jmarsden|work: how to revert back when i tried to upgrade a module then it went wrong for that versions, how do i roll back..to old version..? [01:30] hi i have tried the apt-get upgrade and missed up one of my application, anyone can help how do i revert it..? [01:37] hi anyone..? [01:38] Where? [01:39] hi [01:40] Who? [01:40] What? [01:40] Why? [01:40] When? [01:40] Welcome to the Ubuntu Server personal help desk where we attempt to help you with your Ubuntu Server. Press 1 to continue. [01:40] :-D [01:41] !enter [01:41] Please try to keep your questions/responses on one line - don't use the "Enter" key as punctuation! [01:41] ruben23: downgrading packages is not supported. [01:42] ruben23: in some cases, if you know what you're doing, it can be done, and usually isn't catastrophic (except for things like RDBMSs). [01:43] Oh, I see you're talking about "modules", so presumably this is related to apache or something, and I missed the start of the discussion. [01:43] Is there a particular problem you're trying to solve ruben23, or was there a whole discussion prior to me arriving here? [01:43] twb:i have idea on this error------->http://pastebin.com/m11785a5d [01:44] and this error also..--->http://pastebin.com/m6581c75a [01:44] my whole system is down now hwihc used php and eaccelerator, what i ahve done earlier tried to do apt-get upgrade. [01:44] ruben23: did you install third-party PHP packages? [01:45] twb: eaccelerator is install by tar file. [01:45] ruben23: I'm sorry, I will not help you with that. [01:45] then i messed it up when i do the upgrade [01:46] I suspect you need to get a new tar file for eAccelerator. [01:46] ruben23: Is it available as a Ubuntu package? [01:46] owh1: not if the package has "accelerator" in its name. [01:46] :) [01:46] owh1:ill do the package now but how do i remove the source install..? [01:48] ruben23: That depends entirely on how it was installed. This particular problem is why we keep telling people to use packages, rather than just install stuff from all over the place. My personal suggestion would be to find out what the installer from the original package did and determine if A) it has an uninstall, B) if there is an install log, C) if there is support from the author(s). [01:49] I realise this is probably time critical, but that's the best I can give you ruben23. [01:52] ruben23: One thought. What else was installed when you did the upgrade? Does any of it depend on this particular version of PHP? If not, you could uninstall the currently installed version of PHP, then install the previous version. Bear in mind that this won't actually fix anything - since it's an issue waiting to happen, but it's something you might investigate - note that if your install has lots of things that depend on PHP [01:53] ruben23: You could also install the PHP source, then compile eaccellerator yourself, but I doubt you'd get any help for that here. [01:54] owh1:ok ill just recompile eaccelerator by package then trsy it again [01:54] ruben23: No need to compile PHP itself, all you're trying to do is get the eaccellorator binary. [01:54] yeah [02:07] owh1: ive done it, it worked, now system is working and back to normal..thanks on the advice.. [02:07] ruben23: Now document what you did, put it in the system administrator document file and use it next time - meanwhile go look for a packaged version :) [02:08] owh1: ok, thanks === ajaya_ is now known as ajaya === jtisme is now known as jtholmes [03:14] <_csmith> hi, i have a quick question about netboot [03:14] <_csmith> can i get it to install a base system with a server kernel, without all the x11/cups etc... junk? [03:14] <_csmith> via netboot - the servers don't have a cdrom === nxvl_ is now known as nxvl [03:41] I haven't tried it myself, but I'm pretty sure you can. [03:41] _csmith: Do they have USB ports? [03:41] <_csmith> yeah [03:41] <_csmith> i think i found the trick [03:41] If they do, you can also use usb-creator to put the ISO on a USB stick [03:41] <_csmith> if you type "cli" on the boot screen i think it may do a server install [03:42] <_csmith> i thought it meant cli install process :\ [03:42] Hey guys is it possible to mount a device to multiple folders? [03:43] Doonz: you want the -o bind or -o rbind option to mount. [03:46] <_csmith> nah, cli option installs all the shit, too [03:46] <_csmith> >:( [03:46] _csmith: So take the server ISO and usb-creator and put the server ISO on a USB stick. [03:46] usb-creator-kde if you like such things. [03:46] Doonz, this is a little off topic, did you have a bro in the military? [03:47] <_csmith> doh, on a mac [03:47] <_csmith> i have a 1gb flash drive around, will that do the trick? [03:47] <_csmith> got some docos on it? [03:49] twb: thank you [03:49] the differences between the server and alternate installs are only the preseeds [03:49] billybigrigger: no [03:49] You should be able to extract those trivially by comparing the isolinux.cfgs on both isos [03:50] From those, you can just type in the server CD's default preseed values at the alternate CD's boot: prompt, [03:50] (From memory, the chief differences are the kernel flavour and the absence of language-pack-*) [03:51] Doonz, i used to talk to a guy on irc with that nick, years ago, can't remember what channels though...where'd you get your name from? sorry for asking just that i could have sworn he was from edmonton [03:51] <_csmith> excellent tip, thanks twb i'll check it out [03:51] _csmith: oh, I see you're netbooting -- that makes it even easier -- just copy the appropriate values into the pxelinux.cfg/default [03:51] billybigrigger: i was in the military [03:51] _csmith: I think current server CDs also have a preseed FILE, which you can either copy into the pxelinux.cfg (see Debian Installation Guide's appendices) or supply via HTTP [03:52] <_csmith> cool [03:52] oh, did you ever send a message to some guy on irc telling him you were his brother and that you had died on a mission or something? :P [03:52] <_csmith> will take a look [03:54] <_csmith> ok found it [03:54] billybigrigger: I don't find that particularly funny. [03:56] well it wasn't meant to be funny, but it's true [03:56] OK. then I was confused by the :P [03:56] how to check the last install/upgrade on my system since a given date ? [03:56] New bug: #425386 in dovecot (main) "dovecot package missing dovecot-sieve LDA plugin" [Wishlist,Fix released] https://launchpad.net/bugs/425386 [03:57] but sorry for the off-topic banter, just saw the nick Doonz and the host from edmonton, and it gave my memory a flashback [03:59] billybigrigger: no [04:00] sorry maybe someone else? [04:02] possible i guess [04:16] <_csmith> Thanks heaps twb, i made myself a preseed file based on the ubuntu server one and modified the netboot menu config to load it via http [04:17] <_csmith> seems to have done the deed [04:17] _csmith: in the next release, you can supply the preseed file via TFTP. Yaay! [04:22] Are there any resources to help me migrate from one linux server (gentoo) to another (new ubuntu-server install)? I know what I need to do to migrate apache, but postfix and mysql present additional challenges in data integrity. [04:24] gnac_: For Postfix, with the exception of being chrooted by default, Ubuntu's Postfix works pretty exactly as upstream ships it, so to the extent Gentoo does the same, you should be able to reuse the same configs. [04:26] i just deployed ami-1515f67c (Karmic 32 bit server), and try to login as ubuntu@ with the appropriate key, and keep getting: Too many authentication failures for ubuntu -- any tips? [04:27] ScottK: ty [04:27] Where does postfix chroot to? [04:28] twb: /var/spool/postfix/ [04:29] Hm, so it doesn't create mboxes in /var/mail? [04:29] Oh, I guess only some of the postfix processes are chrooted [04:30] Yes. [05:22] I configured RAID1 on my bios but am unclear if my install of Ubuntu 9.04 is configure properly, cant install grub 2 and see the device as /dev/mapper/pdc_bgfghjgbh2, does this sound right? [05:25] mattgyver: you are using fakeraid. Don't do that. [05:25] twb, what do you suggest? [05:25] md RAID beats fakeraid [05:25] (sorry this is my first stint with this) [05:32] mattgyver: https://help.ubuntu.com/community/Installation/SoftwareRAID [05:33] jmarsden, excellent i didnt see this but did see a different guide, thanks! [05:33] You're welcome. [05:35] jmarsden, I can use ext4 correct? This guide is using ext3 [05:36] mattgyver: Yes, you can use either one. Most server admins are conservative about new filesystems though... [05:36] Are you sure ext4's benefits for your application are worth the risk its newness inevitably brings? [05:37] Well, in my case this is really just an upgrade to an existing home server that i have setup, everythings running fine on ext4 on that box [05:37] OK. [05:38] I guess, at the same time its still kind of an experimental box so im okay with what it brings === johe_ is now known as johe [05:53] jmarsden: Thanks for the link to the bind9 documentation from yesterday. Reading it though only indicates that the behaviour of named-xfer is now included, but I've yet to find out how I can do a simple zone backup. Does this mean that I now need to become a bind9 expert to make a simple zone backup 8-) [05:54] expert? Probably not. Just set up a slave zone in your named.conf with the IP of an appropriate master server and it should pull the info in for you... [05:55] jmarsden: Doesn't that imply that I'm running a DNS of some sort, rather than a workstation where I'd like a backup of the zone data, or am I misunderstanding? [05:56] owh1: Yes, but you can run a caching DNS server on a workstation just fine. What is the value of a zone backup if you have no server to service it with anyway :) [05:59] I'm migrating the zone to different infrastructure, from a DNS server not under my control, to an ISP run DNS server where I have full web-control, just no access to the raw zone data. At the moment I have no way of determining if the previous DNS operator had some hosts or addresses defined that I know nothing about, which is why I wanted to do a zone level backup... [06:00] I'm wondering if I can do a full query using dig instead - one that gives back everything, that would suit the purpose too. [06:00] Ah, OK... I don't think dig will do that. [06:01] Yeah, I didn't think so either. Last time I recall trying with my own domain and not discovering the IP address of a defined host - knowing full well that it was there because I put it there :) [06:02] If you can control where the "old" DNS server will send full zone info to , you could use one of the free public secondary DNS servers to grab the zone info and then use its web interface to view it... but IMO yu'd do as well just to set up a little caching DNS server on your workstation and then make it a slave for the zone concerned. [06:03] I have absolutely no control over the "old" DNS server at all. [06:03] I have to do this for 20 odd domains :( [06:04] So how do you know it would permit you to use named-xfer to grab the zone, even if you had a named-xfer binary?? [06:04] I don't. I also don't have such a binary to TIAS :) [06:05] Most sane admins do not configure DNS servers to allow everyone to xfer the full zonefile these days. if you want to tell me one of the domains I can TIAS for you :) [06:11] New bug: #490683 in net-snmp (main) "package libsnmp-base 5.4.2.1~dfsg-3ubuntu2 failed to install/upgrade: subprocess installed post-installation script killed by signal (Interrupt)" [Undecided,New] https://launchpad.net/bugs/490683 [07:14] <_ruben> aww .. wiki.ubuntu.com down :( [07:26] _ruben: FYI, I can confirm that here. [07:29] I filed a bug on rt.ubuntu.com about it already. [07:30] Hmm, and it just came back. [07:31] Heh, Ubuntu doesn't use LP for its internal BTS? ;-P === dendrobates is now known as dendro-afk [07:38] twb: I suspect it is too slow for real time sysadmin tickets like that. [07:38] Ha! [07:38] I wonder if anyone has actually deployed LP (now that it's AGPL'd) outside of lp.net [07:40] twb: In private for testing, sure they have, I saw discussion about doing that in the couple of days following the open-sourcing of it. In public for production use? I don't know... perhaps not yet. [07:40] In production, but not necessarily publicly accessible [07:41] e.g. my ISP uses rt, but I can't browse to their rt's web interface [07:41] I'm not counting toy deployments [07:56] New bug: #490664 in samba (main) "Samba stopped working between 9.10 LiveCD and current installed system." [Low,Incomplete] https://launchpad.net/bugs/490664 [08:17] <_ruben> twb: i also checked with http://downforeveryoneorjustme.com/wiki.ubuntu.com ;) [08:18] I hate that site. The name is huge, it's .com instead of .net, and it's not "everyone" -- it only performs the test (ICMP or HTTP?) from one node. [08:29] hi, i want to choose a control panel for a ubuntu server [08:29] any suggestions? [08:30] As long as it's USB, any of them should work. [08:37] Shubuntu: There is none. eBox is built for Ubuntu, but in my opinion it is too much oriented to Microsoft networks. [08:39] alvin, so i'm basically screwed [08:40] Shubuntu: No, why would you? The command line is still the best tool. (but I admit eBox looks nice) [08:40] Shubuntu: And of course, I forget... There's Landscape, but it's not free. [08:40] alvin, i don't want to give command line access to people [08:41] Shubuntu: No, that's just his opinion. Try ebox. If you can't live with it, try any other panels you want, but be aware some are unsupported and some don't fit well with how Ubuntu/Debian manage packages and package upgrades. [08:41] i don't want something to manage packages [08:41] just something that people can use for ftp, user, email, and website management [08:42] That's the problem with all those control panels. None really fit in. [08:42] Shubuntu: a "control panel" tends to modify configuration files... which package updates also have to deal with... [08:42] eBox can do user and email [08:42] can ebox create virtual users and emails [08:42] and set up proftpd virtual users? [08:43] and create new sites and set their directories [08:43] Users are ok, but it can not manage ftp. Take a look at their website. [08:43] k [08:43] thanks [08:43] I don't use ebox :) webmin can do those kinds of things but it is incompatible with Debian package management... so use at your own risk. [08:45] I think twb talked about some management tool a while ago. Let me see if I can find that one. I still have to test that too. [08:45] The official recommendation is that you learn to use the CLI [08:45] I think [08:46] twb: Agreed. But alvin, i don't want to give command line access to people [08:46] Ah, I missed that line [08:48] webmin (or something like it) used only for email/web virtual hosting setup, and good backups of /etc/ just in case, is probably his only real option given that constraint, I think. [08:48] Shubuntu has a point. I install servers for a living and explaining ssh and adduser to Windows users can be hard. eBox can do a nice job there, but I wish it could handle NFS too and integrate even better with Ubuntu. (e.g. when you have an existing samba installation) [08:48] is there a how to, to create my own control panel? [08:48] I'm ashamed to admit that my boss makes me deploy webmin (and GNOME!) on customers' servers. [08:48] like the things I want to do are limited [08:48] if i find a good how to i'll code it myself [08:49] the requiremnets i have aren't as deep as webmin [08:49] Given that all they REALLY need is the ability to add/remove user accounts and such, I suggested using the gnome-system-tools tunneled with putty and xming32, but I didn't get budget approval to develop the idea [08:49] Shubuntu: In that case, you're back to eBox. It's mostly perl and it 'should be' documented. [08:49] webmin does a whole lot of things I don't care to use [08:49] i looked it up [08:50] it's horrible [08:50] You can suppress most of the webmin buttons from the end user's UI [08:50] ok :-) [08:50] Shubuntu: It is pretty modular, you can deconfigure most of it... [08:50] ebox is one of the worst things i've ever seen [08:50] very bad documentation [08:50] Shubuntu: OK, you can write your own panel :) [08:50] The problem is that anyone clueful enough to build a web UI is already too clued to WANT a web UI [08:50] hmmm, gnome with NX? nxclient runs good on Windows. (But installing gnome on a non-xdmcp server? bah!) [08:51] So all the web UIs are by clueless buffoons. [08:51] twb: Nice theory :-) Could be true. [08:51] Now, if you could get a web UI that read and wrote puppet manifests, that'd offload a lot of the backend effort to a separate tool, so the web UI could focus on the UI [08:51] I just need like a how to [08:51] twb: Some clueful people need to give away limited config access to servers to less-clueful people... that seems to lead towards development of a GUI for performing such tasks... [08:51] As it is, the backend code for webmin, at least, is utterly awful. [08:51] jmarsden: yeah, I know. [08:52] Shubuntu: If you need a howto, you don't know enough to write your own panel. [08:52] jmarsden: but normally I'd just give the less-clueful a VM or something, and let them shoot themselves in the foot [08:52] i wanna code either in perl, python or php [08:52] and I do know some [08:52] but still i need some guideline to follow [08:52] and then modify things as i see fit [08:53] jmarsden: either that or I'd write a menu-driven curses app, make it their login shell, and give them putty [08:53] Shubuntu: OK, take webmin or ebox and improve them into a panel you like :) [08:53] thanks, i guess that's the extent of your help [08:54] Shubuntu: The chances of you finding a "howto write your own web control panel" howto are very small. [08:54] jmarsden: for e.g. point-of-sale stuff, people I talk to are still perfectly happy with their DOS-based UIs, and hate it when they're replaced with web-based solutions. As long as it's a TUI and not a line-oriented interface, they seem to be happy enough with text... [08:55] twb: True, that could work. But once people have used a web UI to add email users/set passwords, mess with their web sites, etc it would probably be hard to persuade them to use a text interface for those functions. [08:56] Meh [08:56] I think it'd be easier to convince them than to fix webmin [08:56] the ideology of command line being good is unix, not web [08:56] the developers need to grasp that [08:57] Shubuntu: you're the developer, man. If you think you can do better, I ain't stopping you. [08:57] Shubuntu: OK, you are a developer, so go for it :) [08:57] it is not acceptable to expect end users to become os experts to be able to do web [08:57] Shubuntu: if you do a good enough job, I might even be able to get you some funding. [08:57] for real? [08:57] if i get money i'll do it [08:58] i'll create cream de la cream of all control panels if i get enough funding [08:58] Shubuntu: you get the money after you're done, if you do a significantly better job than the existing solutions, you solution is FOSS, and I can convince my boss that you'll stop developing it if you go back to your day job. [08:58] and i'll make it all open source but well documented so people can modify and use [08:58] Shubuntu: it obviously wouldn't be enough to live on [08:59] I mean, how do you think stuff like OpenLDAP gets developed? Groups like Red Hat and MIT pay the FOSS developers to work on it. [09:00] twb: Usually only after a version is created for free that mostly works... [09:00] jmarsden: granted [09:00] jmarsden: or gets open-sourced [09:00] i expect to be able to make a working version in 1 week [09:00] Shubuntu: then you're insane [09:09] ok if i set it up, I won't support antivirus and other things [09:09] only the web feature sets [09:09] i think it's not appropriate for people to try to cram everything into a control panel [09:10] control panel should handle web related material [09:10] smoser: ah right, so because I'm futzing around with SYSTEM I'm going to get a bit stuffed by ec2-init, aren't I ;) [09:13] in my opinion web control panel shouldn't deal with system set up [09:13] it should only be a front to configuration of behaviour [09:14] a mistake all these panels are making is they try to be a cross platform system control panel [09:14] which is basically impossible [09:14] even if they only develop for linux [09:15] the only way a control panel would be able to do such a thing is if it becomes a part of a kernel family [09:15] Shubuntu: well, they could offload the application of change management to a specialist backend like puppet or cfengine [09:15] and everything is developed for it [09:15] yep, or daemons [09:16] and the control panel will act as a ui module only [09:16] which makes it light [09:16] and it won't need to have a lot of system specific object files [09:17] i do agree with the idea that linux is unix based and all root / admin management is better done using command line [09:17] that just goes with the philosophy of linux [09:17] but [09:17] web is a different feature [09:17] and trying to drag web into the linux philosophy is a grave mistake [09:23] How can one request a backport? I asked for libsys-virt-perl in a wishlist (bug 462688). It's set to Fix Released now, but that is only for Lucid and that's too far off. [09:23] Launchpad bug 462688 in ubuntu "[needs-packaging] libsys-virt-perl" [Wishlist,Fix released] https://launchpad.net/bugs/462688 [09:26] Five months is too far? [09:28] twb: of course! I want to use it on production machines. [09:28] As for me, I wouldn't use non-LTS on a production machine... [09:29] twb: Good point, but they are kvm servers, and our virtual guests crash on Hardy (and Jaunty). They don't crash on karmic, although karmic certainly has more visible bugs in other areas. [09:30] Yeah, I understand. I was just grumbling, really. [09:30] I don't actually know about Ubuntu backports, either [09:31] So, out of no where my apache2 server is not working correctly. If you go to server, it sends a file to the browser that is named somethingblahblah.part... when you download the file it IS the index.html file... but for some reason it does not load as a website unless you go domain.com/index.html then it loads fine. Any suggestions? [09:32] crohakon: did you think to version control /etc (e.g. with etckeeper)? [09:32] And it is only for the / directory. All sub-directories load the index.html or index.php fine [09:33] twb, what do you mean version control? [09:33] twb, got a wiki on the subject I can take a look at? [09:34] crohakon: apt-cache show etckeeper [09:35] I have not done any upgrades [09:36] I was moving around some of the directories for the websites, reorganizing, and after that I started having this problem. [09:37] "I was cleaning up and there was this big file called `vmunix' that nobody seemed to be using..." [09:38] =) This is a simple install of apache2, I was only organizing the files in var/www/ and it should not have caused an issue. [09:38] Clearly SOMETHING changed and broke it. [09:38] This is why version control (or even change management) is Good Juju -- it allows you to ask the system "what changed?" [09:39] When you try to go to the domain it sends a file ---> 9TcSwWXi.~.part <--- the first part of the name is always just some random numbers and letters. If you download the file and open it, it is in fact my index.html file. If I got to domain.com/index.html it loads it just fine. [09:40] crohakon: I have no idea what "it sends a file" means. Try curl -o/dev/null -v URL [09:40] soren: proposed is the correct state for it at this point, I'd say. we'll plan the next chunk of work when we finish this one, and we can consider the rest of the blueprints which couldn't fit into alpha 2 [09:40] soren: and you aren't a minion, you're an engineer [09:41] crohakon: the URL you /msg'd me works for me. [09:41] crohakon, perhaps it has to do with mime types? [09:42] i know for a fact that, such things happen on a lot of file name that aren't recognized [09:42] such as doc files [09:42] mime types? I have never messed with mime types. Got a good tutorial on it? [09:43] oh my... [09:43] I cleared my cache on my browser and all is working good... I feel like an idiot... [09:43] * crohakon bows head in shame [09:44] crohakon, http://httpd.apache.org/docs/1.3/mod/mod_mime.html#addtype [09:44] mdz: If that's how we're doing it, I'm not sure I understand difference between "Series goal" and "Milestone target" setting of blueprints. I thought setting the series goal was something like "We expect to work on this project to some extent for release X", while setting the target milestone was more about setting a deadline for deliverables. [09:44] oh ok then [09:45] I will read the link anyway, it is good knowledge to have. Thanks for the help guys. Sorry for my stupidity. [09:46] yep, if in future you would like to create pdf, doc etc on yourserver, that would be useful === cjwatson_ is now known as cjwatson === kane__ is now known as kane_ [09:56] Suppose I am foolish enough to "ufw enable". How does one later see the rules added at the "ufw level", as opposed to iptables-save? [09:58] twb: ufw status === david is now known as Guest6473 [10:01] Oh. Obviously it's not reporting anything because I haven't added any rules. [10:03] Though it doesn't list the default policy for each chain... === tdn_ is now known as tdn [10:05] twb: ...because it's disable. [10:05] d [10:06] Oh, sorry. [10:06] twb: ufw status verbose [10:07] Nah, still doesn't do it on 8.04 [10:14] There's also some peculiar default rules... [10:15] For example, who the hell is 224.0.0.0/240.0.0.0 ? [10:18] # allow MULTICAST, be sure the MULTICAST line above is uncommented [10:18] Oh, MDNS? [10:19] (I'm still trying to get Emacs to let me in to look at the source files.) [10:19] That's what I'm reading in /etc/ufw/before.rules. [10:20] There we go, I just had to delete the hung ssh process. [10:20] Yeah, mDNS is 224.0.0.251 [10:22] http://en.wikipedia.org/wiki/IP_multicast, and RFC 3171 [10:25] hi, my mysql-server-5.1 fail to start on ubuntu : http://pastebin.com/d36d3c136 [10:34] jpds: I'm also far less freaked out now I realize those --dports in the INPUT child chain are just -j RETURNing [10:53] how could I get more info on mysql-server start ? [10:55] lyhana8: tail /var/log/mysql.err [10:55] and mysql.log [10:55] but since it is failing, probably in .err [11:19] baccenfutter: they are empty [11:22] in response to Matthias Gug's RFP, I just added nginx to the list of proposed promotions to main on https://wiki.ubuntu.com/LucidServerSeeds [11:23] I hope that's an appropriate way to make the suggestion [11:44] dix [11:53] soren: does EC2 not bring the metadata server up immediately with instances? the half an hour wait for it is... surprising... if you're running a UEC that doesn't seem to present metadata services ;) === Pupuser402 is now known as shadow25 [11:54] lyhana8: have you checked syslog, dmesg, any logs that might contain detailed information about the fail? [11:58] nop, I just purge it to make a reinstall [11:59] Ng: We have no clue how long it might be. [12:00] Ng: Perhaps half an hour is too much. I don't know. [12:00] I /do/ know, however, that I'm absolutely starving. [12:00] * soren goes to make some lunch [12:00] soren: I was just futzing around at home and built a one-box UEC cloud in SYSTEM mode and I have to wait for the full timeout ;) [12:01] could the fact that /var/lib/mysql/ is a link cause trouble ? [12:03] baccenfutter: it was... [12:10] soren: the only use of series goal is that it has access control. what counts is the set of things which are BOTH targeted for lucid AND milestoned for alpha 2 [12:10] everything else is backlog [12:15] mdz: I see. [12:17] mdz: One problem I see with this approach is the burndown chart. The lucid one (as opposed to the lucid-alpha-2 one) will be quite misleading if we only accept stuff that's expected by alpha 2. [12:17] soren: that's OK, we're reworking things as we go to accommodate the new model [12:18] we don't plan the whole release in one go anymore [12:18] the alpha 2 burndown chart is what counts [12:18] we'll drive that down to zero, then start a new one [12:19] ttx: as you touch the various specs, could you start setting jib to be approver for them? [12:27] mdz: sure. [12:49] hi [12:49] how can I do some test/benchmarks on network speed? [12:49] I think I'm having issues in my lans, and some computers do not work as expected, and I'd like to identify where is failing [12:51] acalvo: mtr? [12:51] didn't heard of mtr [12:51] I'll check it [12:52] man mtr [12:52] I've seen some packages called iftop [12:52] !man mtr [12:52] ok [12:52] I'll give it a look [12:52] thanks! [12:55] baccenfutter: Yes, thanks. Didn't know that program yet. :-) [12:55] acalvo: You can also try iperf [12:56] iperf [12:57] let's see === crohakon is now known as crohakon|zZz [13:20] hi. i'm afraid i'm getting stymied by what i think is a simple permissions issue. i can't seem to get dhcpd to read a file that i believe it should be able to. if i could trouble someone to look at this pastebin, i'm hoping i'm just overlooking something basic. http://pastebin.com/d657509 [13:31] lunaphyte a second, plz [13:32] lunaphyte there is no group permissions on those files [13:35] bogeyd6: i think it does, no? read permission for the group? if i change the group to dhcpd, you can see it's able to read the file, which i think confirms that. [13:42] Ng, yeah, if you dont have the metadata service, ec2-init is not going to be too happy. [13:43] I'm not entirely sure why I don't, but I'm assuming it's because of SYSTEM [13:43] 2 things you can do... a.) htere is a file /etc/ec2-init/is-compat-env that has 'compat=1' . you can mount that image, and change it to compat=0 [13:43] Ng: right. [13:43] Ng: SYSTEM disables the meta-data service. [13:43] well, i guess there is only that thing that you can do. hm... [13:44] i think that .b) was going to be that i plan for ec2-init to behave nicer in such a situation for lucid, or at least be able to give you other wasy around it. [13:48] smoser: other than timing out sooner, I'm struggling to imagine a way it could reliably tell [13:48] Ng, right. it really can't. thats the problem. [13:48] we might just have to do some tests on ec2 and see how long is reasonable to wait. [13:49] right now i think its 30 minutes... [13:49] that's what soren's comments in his python code suggest [13:49] just wondering, I've been seeing oidentd as a recommended inclusion in main on the /LucidServerSeeds wiki page, but I can't really think of a great use for it. Could somebody enlighten me? :) [13:49] I didn't actually time mine to check, and I thought it had continued to fail from the console output, but it did actually finish booting at some point overnight [13:52] Ng: It's really a very annoying situation. Any suggestions would be appreciated. Ideally, we want to only query the metadata service if we're on EC2 or UEC, but the only thing that reveals whether you are on EC2 or UEC or not is exactly whether or not there's a meta-data service. [13:54] soren: we are, or are going to, run a bunch of automated testing on EC2 and UEC, right? [13:55] if the results of that could include a debug log from ec2-init that says how quickly the metadata service replied, you'd see fairly quickly whether it's *always* there when the instance boots and know how safe an assumption that is :) [13:55] if it's literalyl always right there and responds quickly, then a timeout of a few minutes seems safer [13:57] hello everybody [13:57] i upgraded my production machine to ubuntu 9.10 today [13:58] and now the network is unreachable as non root [13:58] so now almost all systems are down [13:58] does somebody know what is going on [13:58] why can i ping as root but not as non-root [14:00] if somebody has any hints just trow them :D [14:00] i am currently getting +100 mails a minutes that are now getting denied due to unable to resolve the domains the mails come from [14:02] /etc/init.d/apparmor stop? [14:02] that's a huge wag. [14:03] lunaphyte_: i purged apparmor [14:03] hmm. well, like i said, a huge wag. :) [14:05] rebooted the system to be sure [14:05] and now the network is unreachable as non root <--- completely unreachable or just ping? [14:05] cyphermox: all non root processes [14:05] just a thought, have you looked to see if you are part of the network group? [14:05] ping does not work [14:05] so you couldn't say, do w3m http://some.host.on.the.network/ / [14:06] cyphermox: wha tis the netwerk group? [14:06] is this something new? [14:06] Ng, it is not always "right there". thats why the retry-loop is in place. [14:06] cyphermox: bind9 is running as non root and it is also unable to resolve [14:06] I see. [14:06] smoser: well that just sucks ;) [14:07] tuxcrafter: If someone accidentally un-setuid'ed ping, only root would be able to ping. [14:07] postfix is also not able to resolve [14:07] https://bugs.launchpad.net/ubuntu/hardy/+source/ec2-init/+bug/308530 is the bug, Ng. [14:07] Launchpad bug 308530 in ec2-init "Wait for network before downloading ssh credentials or user-data" [Critical,Fix committed] [14:07] tuxcrafter, can you ping an IP directly? [14:07] lets test that [14:08] tuxcrafter, you keep saying resolve, so we need to be sure :) [14:08] cyphermox: ping 74.125.53.100 [14:08]