[00:21] nijaba: it hasn't started as of yet actually [00:21] * faulkes- was in negotiations all day [00:22] I should have it going tomorrow as I have a free day, well, "mostly" free === c_schmitz is now known as c_schmitz|away [01:54] * kgoetz discovers he has no /proc/acpi under 7.10, but this doesnt crash - therefor the problem is with 8.04 :( [01:57] afk lunch === foo_ is now known as foo === freeflyi1g is now known as freeflying [02:38] back [02:44] Is courier-imap has Poppassd service? [02:44] rhineheart_m: what? [02:44] Bambi_BOFH: yeah.. Poppassd service... [02:44] Bambi_BOFH: I need it for change_passwd by PAM authentication for my squirrelmail.. [02:44] that didnt clear it up [02:45] i see. i suggest searching the web for info === Bambi_BOFH is now known as kgoetz === c_schmitz is now known as c_schmitz|away [05:59] pschulz01: we forgot http://ars.userfriendly.org/cartoons/?id=20080302 [06:55] could somebody help me with a question about setting up a routing table? [07:38] moin [08:10] Doesn anyone has any experience with dhcp3-server and failover? [08:13] sergevn: are you going to ask the every day? ;) [08:13] henkjan: yes :) [08:13] henkjan: ;) [08:13] do you realy need failover? [08:14] henkjan: yes [08:14] i've heard only horror stories about failover [08:15] henkjan: I know how to configure it, but it aint working :D [08:15] <_ruben> search the dhcp users mailinglist archives .. its a rather hot topic overthere [08:15] sergevn: yes. did that answer helped you!? [08:15] 2 dhcp-server witch their own scope isnt an option? [08:15] dhcp failover is a major pain into the ass... [08:15] henkjan: if the secondary takes over, it says it holds all leases [08:16] kraut: yeah im getting that picture the last 2 days [08:18] Hm... I've never tried it, but it sounds pretty straight forward? [08:19] Assuming some sort of monitoring mechanism takes care of switching them, what's left to do but copy the leases from the active to the passive node? [08:30] <_ruben> strange .. ps aux sorts procs by pid, tho im used to them being sorted by starttime (on suse boxes) .. guess i'll have to dive into the manpage :p [08:32] <_ruben> ps auxkstart_time .. lets see if that can be made a default somehow [08:40] sergevn: just configure 2 dhcp servers with both the same scope [08:41] sergevn: a client request an adress by broadcast. Both (or more) dhcp answer: 'pick this adress'. Client takes the adress and answers in broadcast, 'i take this adress'. Both (or more) dhcp servers update their leases file [08:41] thats how we are running it [08:41] Yeah, I really don't see the problem. Identical configuration, sync leases info from active to passive node. When disaster strikes, start dhcp server on the passive node. Go. [08:42] you don't need to sync leases [08:42] henkjan: AH, yes, that's even better. [08:42] henkjan: I never tried. Didn't know dhcpd was clever enough to do that. === waver_ is now known as WaVeR [08:53] henkjan: so if you run 2 dhcp servers in the same network [08:54] henkjan: they know about each others leases by defaul? [08:54] hi, eth0 isnt starting on boot (ifup eth0 starts it correctly though) where do i start looking? [08:55] pppZero: In /etc/network/interfaces [08:55] morning [08:55] pppZero: It needs to say "auto eth0" somewhere. [08:56] soren let it be known for the record, i love you! [08:56] heh [08:56] sergevn: a client will answer with an broadcast. Both dhcp server receive that broacast en place the mentioned ip the leases file [08:56] henkjan: aaaaaaaah ok [08:56] sergevn: thats exactly how we are running it with 4 dhcp servers [08:57] * soren_ hugs pppZero [08:59] well, did it earn an VT enabled pc with this answer? === soren_ is now known as soren [08:59] henkjan: problem nr2, it has to be compatible with dynamic dns [09:00] hmm [09:00] sergven i've got a setup going here with dhcp + djbdns + a perl script that adds hostnames to djbdns as windows boxen take out leases, give me a minute or 10 to find the link [09:03] if you want to use bind, I've got no idea how its done, probably some changes to how it writes files should suffice, but heres the link: [09:03] http://www.thismetalsky.org/projects/dhcp_dns.xml [09:05] henkjan: im testing it right now, 2 dhcp servers in 1 subnet, same range etc. [09:06] henkjan: but if i get lease from 1 server, shut that server down. [09:06] henkjan: do another dhcp request, im getting the same ip [09:06] thats what you want, right? [09:06] henkjan: yes, but not duplicate leases [09:07] hmm, how are you testing it? [09:07] henkjan: havent released the old one yet, and getting the same one (as old) from other server [09:07] Morning everybody :) [09:07] a proper dhcpclient will answer with a broadcast [09:07] henkjan: using dhclient [09:08] sergevn: thats exactly what you want. The second dhcp server knows your lease, so it gives you the same ipadress [09:09] henkjan: hmm [09:13] the dhcp server seems to give the same ip to a machine as long as it can, i've got a computer here thats been on the same IP for about 3 years, and thats though endless reboots, OS installs and power outages === \sh_away is now known as \sh [10:12] hello everyone...i need help with ssh [10:13] slafko: what is your problem? [10:13] i want to connect to server with ssh, but I got error "Host key verification failed" [10:13] the server seems to have a different hostkey than the key you have in .ssh/known_hosts [10:14] how to fix it? [10:18] ok, I solved the problem [10:18] thanks! [10:18] you can remove the corresponding line from ~/.ssh/known_hosts [10:18] ah, to late :) [12:02] what is an good application for emailing system reports [12:02] no, i dont like to create manual scripts :P [12:03] logwatch? [12:04] sergevn: yep, logwatch is great, even though rules can be a pain sometimes to get right [12:13] morning nijaba [12:14] hello faulkes- [12:16] well, that was nice, I think I actually slept for 12 hours solid [12:17] nijaba: logrotate is installed by default i see, doesnt lolrotate have the same functionality as logwatch [12:18] nijaba: lolrotate = logrotate :D [12:18] sergevn: not IIRC. logrotate changes the logfile and archive old one only [12:18] sergevn: it can mail them eventually, but with no filtering at all [12:19] * faulkes- nods [12:20] nijaba: ok, so i installed logwatch, i see it puts it script in cron.daily. [12:20] logwatch will allow you to set regexp to be looked for in multiple log file and periodically mail you the matching lines of all watched log files in a single email [12:20] nijaba: but aint i getting the same log each day now? [12:21] sergevn: nope, only the lines that matches your regexp will be sent [12:21] sergevn: look at the default config, it is a good safe start [12:21] nijaba: noob Q: what is regexp [12:21] regular expressions [12:21] sergevn: regexp = regular expression [12:22] nijaba: /etc/logwatch doesnt contain any config files :D [12:24] sergevn: we have a customer running logcheck on his colocated machine [12:29] sergevn: I realize I have been speaking of logwatch when thinking of logcheck [12:29] * nijaba feels silly [12:32] nijaba: so you didnt mean logwatch in the first place ? :D [12:33] sergevn: both seem to do similar things but with a different architecture. yeah, I meant logcheck as it is what I am using, but logwatch should work as well [12:33] logwatch - log analyser with nice output written in Perl [12:33] logcheck - mails anomalies in the system logfiles to the administrator [12:33] nijaba: you need more coffee [12:33] nijaba needs to send me an VT enabled computer :) [12:33] zul: right... I'll be back after another load of caffeine [12:34] zul: jetlag... [12:34] nijaba: sure sure :) [12:35] both are going to match on regexp though iirc [12:35] logcheck-database - database of system log rules for the use of log checkers [12:35] logchech has a database available [12:35] don't know about logwatch [12:35] rules just get interpreted [12:37] if you look in /etc/logcheck you'll notice that the files in there are all regexp based [12:37] henkjan: I personally don't have any to send, but if you plan on doing something usefull with a vt machine for the community, let's talk about it [12:39] nijaba: it was just joke, hence the smiley. But with an VT enable pc i could test a lot more on virtualisation [12:39] currently running on 1Ghz PIII :) [12:41] henkjan: I understood it as such. And I think we do need some tester on PIII machines as well, just to make sure the devs don't get too sloppy ;) [12:41] hehe [12:42] but KVM will only run with VT support right? [12:43] henkjan: yep, we had to remove the full qemu emul because of it depending on a old gcc 3.4 version, IIRC [12:43] ah, to bad for me :) [12:43] but i'm happy with Xen [13:41] morning all === cjwatson_ is now known as cjwatson [13:48] morning sommer [13:49] hi sommer [14:14] hello all, how to know the running service on ubuntu server and how can i stop any of them ? [14:15] start/stop with /etc/init.d/ thx henkjan , i mean how to stop any permenant [14:16] that next time don't start [14:16] remove the symlink in /etc/rc3.d/S [14:17] or you can use update-rc.d [14:17] ah, didnt know about that one [14:18] When installing Ubuntu server, how can I disable automatic network configuration and enter a custom IP instead? [14:18] Apollo9: edit /etc/network/interfaces [14:19] man interfaces will give you enough information about what needs to be in that file [14:20] Apollo9: you can also press esc during install when the dhcp config is taking place [14:20] it will ask you what IP you want [14:20] ah, during install time its easier indeed [14:21] but in case if he connect to dhcp server then network will configured auto the first time, in this case maybe he disconnect network cable while install [14:23] Thank you all :) [14:30] Btw, how large should the swap area be on a relatively old PC (I'm installing it as a test server to toy around a bit)? Should it be large (like 700MB) to compensate for lacking RAM? [14:33] mohamed_, you can cancel the dhcp configuration (just press cancel when trying to get IP from dhcp) [14:34] thx, WaVeR , for this info [14:34] Apollo9: how much ram do you have? [14:35] welcome [14:35] ram is cheap nowadays, you should avoid swapping [14:35] avoid? yes? always possible? no [14:35] * faulkes- whistles innocently about jvm app servers [14:36] eeew [14:36] henkjan: 128 MB, but only 800 MHz [14:36] 4626 tomcat5 15 0 4874m 1.5g 59m S 2.7 39.0 116:35.54 java [14:37] Apollo9: 512M swap should be enough [14:37] if you only have 128mb, I would have at least 512mb of swap [14:37] OK, thanks. [14:38] is there a command to know the running services ? [14:39] the ps command [14:39] but using that much swap makes your server unusable [14:39] although that will list all processes [14:39] see the ps man page [14:40] it all depends on what you plan to do with the server as a test [14:40] if it's just a LAMP stack, you should be fine [14:40] if it's a jvm app server test, you'll be in trouble [14:41] if it's just general learning and testing, you'll be fine [14:41] "mostly" [14:42] they mostly comem out at night... mostly [14:42] henkjan: I'm pretty new to Linux. Can you please explain why using more swap makes it unusable? [14:43] Apollo9: swap is slow [14:44] if one use more ram , this mean that can't crate swap partation ? [14:45] crate = create [14:50] you can always create a swap partition [14:51] the point I believe henkjan is trying to make is that swap is much slower than actual ram [14:51] because it is disk based [14:51] e.g if i have 2 G ram in this case i can avoid swap ? [14:52] all depends on what you are doing with the server [14:52] I have yet to encounter any server that has not had a swap partition [14:53] even if you feel you will never run out of ram, best practices dictate that murphy's law will apply ;) [14:54] the short and sweet of it, create a swap partition just to be safe, even if it is never used [14:54] and swap partition sizing all depends on how much ram you have and what you expect the server to do, so its a variable amount (IMO) [14:54] if the server had 2gb of ram, I would personally give it 1gb of swap [14:55] this is good then one safe in this case [14:55] safety is good ;) [14:56] of course [14:56] you can't anticipate what or how much memory a given application will consume or if it has memory leaks [14:56] faulkes-, mohamed_: there's a couple of different theories as to how much swap is the right amount [14:57] when I have lots of hard disk, I give 2x RAM size to swap, which handles the theoretic worst case scenario of having to swap ALL of RAM out [14:58] and disk is cheap [14:58] faulkes-: yep, most of the time [14:58] kirkland, but if the system ram is enough and swap also exist system will not use swap till ram resources not enough ? [14:58] faulkes-: I do have a mythtv machine that only has a 2GB SD card for a disk, and there, I have no swap at all [14:58] mohamed_: right [14:59] it is very dependent, my opinion, on what the server is expected to do [14:59] absolutely, I agree with that [14:59] lately, I've been using swapfiles instead of swap partitions [14:59] hello...Is there a way to upgrade squirrelmail using terminal? [14:59] allows me to allocate more and reclaim my swap space far more easily [15:00] kirkland: yes although thats a more advanced technique, for a new user [15:00] http://en.wikipedia.org/wiki/Paging recommends 1.5x RAM size, fwiw [15:03] lol, after installation: /dev/hda1 has gone 49710 days without being checked, check forced. [15:07] that's typical [15:25] how to add users to imap without adding them into the system users? [15:58] <\sh> does anybody work on grinder, e.g. for inclusion in ibex? [16:00] !grinder [16:00] Sorry, I don't know anything about grinder - try searching on http://ubotu.ubuntu-nl.org/factoids.cgi [16:03] <\sh> henkjan: http://grinder.sourceforge.net/ [16:03] ah, i see === \sh is now known as \sh_away [17:41] question: once I create a project in LP, how do I create/add the trunk? after following it's directions I keep getting "not a branch" error from bzr [17:44] faulkes-: --> #launchpad [17:44] faulkes-: did you push your branch to LP ? [17:45] faulkes-: I think you first need to push your branch and then you can set the trunk serie in the project. [17:47] hi all [17:48] mathiaz: it's a bit confusing as to how they want it done [17:48] what is ubuntu recomended server hosting(Virual Server)? [17:48] scottk: thanks, I'll ask there [17:48] zyx386: kvm [17:48] i will bay Virtual Server [17:49] zul, can give the company link, [17:50] zyx386: sorry its not a company i gave you the wrong info misfired [17:51] ok thanks, i choose company with virual server, but ubuntu basis? [17:51] for example www.server4you.com [17:54] can everyone tell me :) [18:29] does a regular server install come with a GUI (DE and x-server)? [18:30] if not, what is a lightweight window manager to install and what package/s? (Fluxbox?) [18:34] engida: you can find details here: https://help.ubuntu.com/community/ServerGUI [18:34] sommer: Thanks [18:34] np [18:35] sommer: is there a factoid for the server gui ? [18:35] !servergui [18:38] mathiaz: not that I know of [18:38] but that would make a good one... wouldn't have to lookup the site :-) [18:48] !servergui [18:48] Sorry, I don't know anything about servergui - try searching on http://ubotu.ubuntu-nl.org/factoids.cgi === \sh_away is now known as \sh [20:31] hrmm, 30 min until meeting? [20:31] zyx386: personally, I use a2hosting.com. They do have Ubuntu VPS options: http://www.a2hosting.com/vps_hosting.php [20:43] hey all, is the statement regarding apt-get dist-upgrade correct: http://doc.ubuntu.com/ubuntu/serverguide/C/apt-get.html [20:44] with regards to upgrading packages? [20:44] wait never mind I'm just reading the lead in wrong I think [20:50] sommer: the statement about upgrading from one release to another should be removed [20:50] sommer: do-release-upgrade should be used instead of dist-upgrade [20:51] mathiaz: ya, I removed the second few paragraphs... as was reading the first one, and it started to not make sense :-) [20:51] mathiaz: it's replaced with a link to the upgrading section under installation [20:51] It's true that's the supported way, but I've never had a problem with apt-get dist-upgrade on servers. [20:51] sommer: dist-upgrading from dapper to hardy won't work as dapper's dpkg doesn't understant the new Breaks: field available in hardy [20:52] mathiaz: yep, we should be covered [21:03] anyone seen ivoks recenlty? [21:03] zul: today he replied to an email a sent him last week, he's been sick for the last 10 days [21:03] ah ok thats unfortunate [21:04] ya, sounded pretty bad [21:27] mathiaz: Since you uploaded cyrus-sasl2, cyrus-sasl2-heimdal needs to be updated too or it's broken. [21:42] just did a 6.06 server install and am wanting to add 2 new users to the admin group. I have looked at adduser and useradd, was wondering if there was a command to list current users/grous [21:43] engida: cat /etc/group and cat /etc/passwd will accomplisth that :-) [21:44] cat /etc/group | grep [21:44] which command is better to just create a normal user that belongs to an admin group (or a sudoer), useradd or adduser? [21:45] engida: I think adduser is pretty simple: adduser username groupname [21:45] but really whatever works for you is the "correct" way [21:45] but I don't want the new user to have a home directory... just a shell and being able to sudo... [21:47] engida: for that you'll want to check the options for useradd, then you can always go back and run adduser [21:47] at least that's what I'd do === Apollo9 is now known as g00f [21:57] ScottK: I haven't uploaded cyrus-sasl2 yet [21:57] Hmmm [21:57] ScottK: kirkland and I still have to complete the merge IIRC [21:57] OK [21:57] ScottK: someone else uploaded it ? [21:58] No. slangasek uploaded another Ubuntu revision. [21:58] So you'll need to update your merge, but the -18 merge isn't done [21:59] When you do, please include an update of cyrus-sasl2-heimdal (in Universe) so it won't be broken. [21:59] sommer: Is it possible for the new user to belong to the admin group and still have its own password to be identified by? [22:00] I envoked the -p argument in useradd and was not promted for a password when creating a new user [22:00] engida: sure, you just need to set it using passwd [22:01] engida: I think the -p argument expects the password to be part of the command [22:01] or standard input [22:01] hmm.. so a should do the trick? [22:02] engida: should [22:02] Thnks :) [22:02] np === \sh is now known as \sh_away [22:02] ScottK: ok [22:02] mathiaz: Thanks. [22:05] mathiaz: do you want to set aside some time this week to finish that cyrus-sasl import? === g00f is now known as go0f [22:06] mathiaz: or, if you want to just do that one yourself, that's okay by me--it looked more complicated than a run-of-the-mill import [22:06] s/import/merge/g [22:08] It can't be uploaded until after the alpha 6 freeze is over, so you have some time. [22:08] kirkland: I think it's good example of a merge. [22:09] kirkland: even though it includes more work, you get a good intro to the merge process. [22:09] kirkland: we could actually upload it as it is, but we've started to clean up the packaging a little bit. [22:10] kirkland: so I think we should keep working on it - get the merge ready and once the archive is opened, I'll upload it. [22:16] soren, zul: can you look this over, and let me know which of xen-3.[012], qemu, kvm need updated? https://bugzilla.redhat.com/show_bug.cgi?id=433560 [22:16] Red Hat bug 433560 in vulnerability "CVE-2008-0928 Qemu insufficient block device address range checking" [Medium,New: ] [22:17] keescook: All of them, probably. [22:17] soren: that's what I suspected. :P === go0f is now known as Apollo9 === Apollo9 is now known as Go0f [22:19] keescook: Don't use Ian's patch, but I assume you noticed that? [22:20] soren: yea, saw comment 22 there. [22:20] keescook: Right. I'll grab it for hardy kvm. [22:20] I'm hoping to get a "blessed" patch from either upstream or you/zul :) [22:24] keescook: Er.. How high priority is this? Can you wait until tomorrow. I want to talk to Dan. === Go0f is now known as Apollo9 [22:29] soren: it's "fix soon" in my book, but needs to be "fix right" due to the invasiveness of the patch, so I'm happy to get additional confirmation. [22:30] keescook: There's one of the cases that looks wrong in his updated patch, and I just want to check with him which one of us is on crack :) [22:30] keescook: None of this is in main, is it? [22:30] keescook: Apart from kvm in hardy, that is.. [22:41] soren: xen? [22:41] keescook: In main? Really? [22:41] * keescook remains unclear [22:41] libxen, perhaps, but not xen. [22:41] I hope. [22:59] libxen is not affected but xen-utils is affected [23:14] zul: xen-utils? Hm. Interesting. [23:17] hello! How to configure http://www.example.com/tick to be accessible thru http://www.tickl.example.com? [23:19] rhineheart_m: in your sites-enabled apache conf file: [23:19] [23:19] ServerName www.tickl.example.com [23:19] DocumentRoot /var/www/html/tick [23:19] [23:21] kirkland: thanks for that. How about hosting multiple sites with each domain? [23:22] rhineheart_m: you can put as many of those sections as you want in there [23:24] you mean? all I have to do is to confgure external nameserver like www.tickl.example.com to point to my IP? [23:26] rhineheart_m: yes [23:26] lamont: I'm reading this http://ressukka.net/blog/posts/20070909_postfix_defaults/ and thinking default subdomain matching for relay'ed domains is probably not the best plan. [23:28] kirkland: where can I find the sites-enabled apache conf? [23:29] rhineheart_m: /etc/apache2/sites-enabled/* [23:31] kirkland: I have only two available files there.. 000-default and squirrelmail [23:32] rhineheart_m: you want 000-default then [23:32] rhineheart_m: you'd better create one file in /etc/apache2/site-available/ for each virtual hosts [23:33] rhineheart_m: and then symlink them into sites-enabled/ [23:33] mathiaz: what do you mean with that? [23:33] rhineheart_m: and then use the command a2ensite to enable the sites [23:34] rhineheart_m: the code that kirkland posted above - put in a file in /etc/apache2/site-available/ [23:34] kirkland: right - I think it's just the standard way to do it. [23:34] kirkland: but you should edit your files in /etc/apache2/site-available/ [23:35] kirkland: site-enabled is just a directory with symlinks. [23:35] mathiaz: ;-) [23:35] mathiaz: agreed. [23:35] the two of you! great job! how to create symlinks by the way? [23:35] ScottK: reading [23:35] rhineheart_m: use a2ensite [23:36] rhineheart_m: there is a manual page for a2ensite that explains what it does [23:36] rhineheart_m: you don't need to know about symlinks. [23:36] Actually I'm not sure I like default subdomain matching at all. [23:37] ScottK: You made me thinking deep? Please say something.. [23:37] rhineheart_m: I'm talking to lamont about a postfix thing. [23:38] Sorry it sounded like your topic. [23:38] * ScottK knows very little about web servers and tends to ignore talk about them. [23:38] NP. Thanks for letting me know.. [23:38] ScottK: yeah. it's a ponder thing, for sure. [23:39] ScottK: and nfc how the 2.4.5 backport built, either. [23:39] lamont: Accepting mail for some domain I don't explicitly list seems so Postfix 1.0 [23:40] and yet my mail server _wants_ that behavior (iz many hosts it takes mail for) [23:40] lamont: I'm pretty sure the whole Main/Universe thing is supposed to be ignored for backports. It's really a non-sequitor. [23:41] ah, that could be it [23:41] And jdong told me that he thought that got fixed a year ago. I think someone 'improved' the system. [23:41] yeah - quite possible that a recent rollout regressed it... you should file a bug against launchpad [23:42] lamont: I have a strict policy against filing or looking at bugs against launchpad. It just hurts my blood pressure. I'm heckling jdong into doing it. [23:43] ScottK: fish in a barrel. [23:44] It's just such a freaking amateur hour project combined with extreme hubris that it drives me nuts. [23:45] mathiaz, kirkland: got an idea how to enable https in squirrelmail? [23:45] My favorite recent example was being told I'd have to give up my opinion that the pre-beta U/I was better if I wanted my complaints to be taken seriously. [23:46] So I've given up on making complaints. [23:47] rhineheart_m: https://help.ubuntu.com/community/forum/server/apache2/SSL [23:53] ScottK: generally, I find that it's getting better, and that it does what I need. [23:54] lamont: There are things that are getting better. I find that as the system is growing more complex the U/I is getting more confusing and it's getting harder to find stuff. [23:54] The amateurish part is I think they put way to much into implementation and not enough into design and testing. [23:55] In the most recent release they made a change to have PPA uploads start building immediately before publisher runs. [23:55] What they failed to notice in design, implementation, or testing is that before publisher runs on an empty ppa there is no releases file. [23:55] Soyuz will barf when it tries to upload the binaries. [23:56] Pretty basic thing I'd say. [23:56] It's currently marked low and the only developer comment I saw was something along the lines of ooh - let's call this the builds to fast bug. [23:57] I just barely restrained myself from a really sarcastic comment and that was the last LP bug I looked at.