[00:00] Mmmmm, I LOVE XFS filesystem. [00:00] xenoterracide: OK... and... did it work? :) [00:00] Hehe.. ktorrent can reserve the exact right amount of disk space needed for each file in a torrent, in a mere second, where-as ext3 took several minutes, dragging ktorrent down to almost nothing while it did it. [00:01] yeah it worked [00:02] xenoterracide: OK, now create a script that just has #!/bin/bash and then /bin/date in it, and put it in the same dir as your backup script, and put 5 * * * * /var/www/oblivionet/com/scripts/testscript.sh [00:02] and see what happens in a few minutes :) [00:02] if I leave the MAILTO in there will it mail me all output that should go to stdout? (never messed with cron's mail before) [00:02] Yes. [00:03] man 5 crontab for the details [00:03] Sweet! [00:03] My tcrules it at least so far, working. ;) [00:04] jmarsden: You know shorewall traffic-shaping stuff any? [00:05] No, I've not used shorewall, I'm afraid. [00:05] Oh, it wasn't you that was the shorewall guy here? okay. Drats. ;) [00:05] yeah that script works too [00:06] I wonder what it's doing or not doing right [00:06] xenoterracide: OK, so now put /bin/date at the top of your backup script and see if that shows up... [00:06] and check file perms on the two scripts are the same [00:07] they are [00:07] Then I'm guessing the script (commands within it) depend on something about the environment that cron does not set up for you... [00:08] could be [00:08] maybe postgres or git needs some info.. [00:09] What are you trying to do, xenoterracide? [00:09] ... well of course that time it would work... wtf [00:09] xenoterracide: Pastebin the script, if you want? Ah... :) [00:09] other than adding /bin/date to the script I would change it [00:09] I didn't* [00:10] xenoterracide: Okay. I can't help you if you don't give details as to what you're trying to accomplish. [00:10] Overall, what are you trying to do? [00:10] Psi-Jack__: basically I'm dumping the db commit-ing the dump to git and pushing the git repo to origin [00:10] Okay. What part seems to be failing? [00:12] I'm assuming just now, you've tested cron, and it is running, so the next step would be to see if it's connecting to postgres, yes? [00:13] actually it seemed to just work now. the last 3 times I did it it failed [00:13] this time it worked [00:13] Okay, just out of curiosity, what are you using to do the dump? [00:13] pg_dump [00:13] Okay, and you're using environment variabled, like PGUSER, PGPASSWORD? [00:13] In your script? [00:13] yeah [00:14] @ cluster heads: here's a question for ya; from the perspective of a load balancer: if one wanted to just notice and log a connection from a list of ip's or where it was routed to; [just off the top of the head] would this add a lot more load on the lb since it has a match list? call it 200 or so ip blocks/single ips [00:14] well actually I made my own variables and passed them to the command [00:14] but yeah [00:14] Well, pg_dump and psql both use PGUSER and PGPASSWORD, so you can do it more securely. [00:15] I didn't set up a password. mostly because the only user that can connect on a socket is this one. and connections aren't allowed from remote [00:15] Sounds to me,, though, like it might've, for the times it failed, had troubles connecting to the database, perhaps the connection limit was maxed? [00:15] I doubt it. very lowtraffic site... [00:15] I can't rule it out though [00:15] Not setting a password is VERY VERY BAD! [00:15] well a password is set [00:16] for the site to connect. I just use ident to verify for socket connections [00:16] Again, NOT USING a password is VERY VERY BAD. ;) [00:16] ident is the most insecure method of all. [00:16] actually trust would be ;) [00:17] xenoterracide: You beat me to that one :) [00:17] and since I'm the only one with cli access to the box. and ssh is on a weird port with key auth only... if you'd like to explain how they'd get in... [00:17] Heh, yeah, true, but ident is the second most insecure methods. ;) [00:18] At work I'm in the process now of getting them to use SSL as well. Originally they'd used ident, too. ;) [00:18] xenoterracide: Defence in depth is always wise. This is the Internet, and there are bad people on the Internet :) [00:18] true [00:18] but if they have access to my user account they could just delete the whole site [00:19] Not necessarily always. ;) [00:20] if they can find it's directory and they know how to run rm -r [00:20] its* [00:20] Not if you use extra security methods to prevent it. ;) [00:21] One good thing about Linux is, it's a preemptive multi-tasking enviroment, so, some ways to prevent mass deletion is with kernel-level security enhancements. ;) [00:21] heh [00:21] remember never implement a security system that costs more than the value of the things it's securing [00:22] You know...that's a good concept. [00:23] chattr +i is fairly inexpensive :) [00:23] * Psi-Jack__ nods and grins. [00:28] and it's pretty easy to revert. [00:29] I'm familliar with it [00:29] there are all sorts of things I can do [00:30] Yes. Including using passwords. [00:30] has it occured ot you on the topic of database that the php cms stores the db password in a plain txt file readable by the same user that I've given no passwd access to? [00:31] and that variables and files would also be readable by this user? [00:31] It's your server. If you like it the way it is, fine. [00:31] and that the only way to login to this db without the password is by logging in as this usr [00:32] is there any decent documentation on GnuTLS? [00:32] I'm not finding any good docs on how to set up a certificate authority. [00:33] all I'm saying is that if someone gets access to this user account, no matter what security measure I implemented they could get in. now I can spend a day or 2 making that impossible. but is it worth the time [00:33] mushroomblue: The gnutls-doc package is not sufficient? [00:36] Heh. [00:36] TinyCA is a useful tool. [00:37] anyone??? [00:39] am trying to mirror ubuntu repo with debmirror --method=http --host=ie.archive.ubuntu.com --arch=i386 --source --dist=hardy,hardy-security,hardy-updates,hardy-backports --section=main,main/debian-installer,universe --ignore-release-gpg --root=ubuntu --allow-dist-rename /srv/ubuntu but it dies out soon with the warning : releasing 1 pending lock.. what can be the problem? [00:40] shame TinyCA doesn't work with GnuTLS [00:40] that looks handy. [00:41] nippz: dunno. good question, tho. [00:41] It doesn't? [00:41] no it says Duplicate Codename hardy. [00:41] s/no/now/ [00:43] Psi-Jack__: I could be wrong, but the homepage isn't mentioning it. [00:47] weeb_: Did you create a dir /srv/ubuntu/ubuntu before running that command? [00:48] nope just that command only:) [00:48] s/command/directory/ [00:49] I'm running Ubuntu 9.04 Jaunty here and tried it but used /tmp/srv/ubuntu instead of /srv/ubuntu and it worked as long as I first did mkdir -p /tmp/srv/ubuntu/ubuntu [00:49] I also got rid of the --allow-dist-rename since my debmirror says it doesn't know about that option. [00:50] It has mirrored 149MB while I have typed that :) [00:50] what should be the file path at the last of all the debmirror options? /srv/ubuntu/ or /srv/ubuntu/ubuntu/ [00:51] while /srv/ubuntu/ is a symlink to my external harddisk [00:51] Just /src/ubuntu. The root=ubuntu is what is adding the second ubuntu, i think. [00:51] btw creating that directory doesn't sole the problem :( [00:52] OK, I'll try again actually using /srv/ubuntu and recheck... [00:54] Hmmm. You are correct. But running the command again then works fine, for me. My command line is sudo mkdir -p /srv/ubuntu/ubuntu ; sudo debmirror --method=http --host=ie.archive.ubuntu.com --arch=i386 --source --dist=hardy,hardy-security,hardy-updates,hardy-backports --section=main,main/debian-installer,universe --ignore-release-gpg --root=ubuntu /srv/ubuntu [00:56] well well /srv/ubuntu was linking to /media/disk/mirror here.. and i just rm -rf /media/disk/mirror and rerun debmirror with target directory at /media/disk/mirror its working:) [00:56] that is without having the directory created.. [00:57] Well, as long as it works, cool :) [00:57] now i may go timed out.. it will pull all my bandwidth:) [00:57] For quite a while... probably 30+ GB or so download there. [00:58] i only have a 2mbps connection it will take a month or so to complete!:) [00:58] Then that's not a sane thing to do, is it? [00:58] no other go:) [00:58] And your ISP may kick you off if you use 100% of bandwidth for that long. [00:59] Get someone to make you a pile of DVS and mail them to you, or something... [00:59] s/DVS/DVDs/ [00:59] well not like that.. i have free net time for 2-8am every day and for other time i have a 2.5gb/month limit so can't run debmirror on those time [00:59] Do you absolutely need a full Hardy mirror? What will you do with it, and how many Hardy servers will use that mirror? [01:00] its for me to play with the repo... actually to play with the builder script for a custom distro.. [01:01] You don't need every package in universe for that?!!! [01:01] ehmm.. err..hmmm this is what i am following http://www.gnewsense.org/Builder/HowToCreateYourOwnGNULinuxDistribution ;-) [01:01] And why start with Hardy rather than with, say, Karmic? [01:02] Or, if this is a learning exercise, see http://www.linuxfromscratch.org/ [01:03] hmm.. yeah they are following LTS thats why still on hardy [01:04] just to play with the builder scripts of gnewsense.org may be i should just read the scripts.. but thought a real run would make it easier. [01:04] * weeb_ wish ubuntu to be fsf free [01:05] Well, sure, but a 40GB download is pretty significant for most home users... [01:06] yeah.. i am also trying to get a shell account to some other who is having the repo [01:07] * weeb_ may not make sense as its early morning here and he is yet to go to bed;-) [01:07] :) and they know you will be compiling the entire repo and will need several tens of GB of disk space? If you can get a shell acct like that for free, that's quite a deal :) [01:08] OK, go to bed and see how much of the mirror you have collected when you wake up :) [01:08] compile entire repo? its just a play with binary debs:) [01:08] afaik [01:09] Hmm, so what is --source doing in your debmirror command ? [01:10] If you don't need the source packages, don't mirror them. [01:10] * jmarsden thinks that perhaps you need to understand the commands you run a bit better, before running them? [01:11] yeah actually all these to understand the builder scripts.. so i thought i would just the downloading thing now so that later i can start learning on stuff.. [01:12] currently am learning more of c.. yet to enter shell stuff;-) [01:14] well well what made it work is that --dist=hardy,hardy-security,hardy-updates thing.. only --dist=hardy works! [01:16] even --dist=hardy,hardy-security gives me "Duplicate Codename hardy." [01:19] I'd suggest you drop universe and start with just main and the debian-installer -- you'll avoid mirroring all the optional stuff you don't actually *need* to create a distro :) [01:19] But even that is a large download. [01:54] i'm trying to start a python script at startup on a headless box using crontab. Do i need to do anything special to get it to start (since it won't on adding a crontab line with the path/to/run.py ) and where do i check to see what went wrong? it dosen't seem to be in syslog [01:55] faileas: Add MAILTO=you@example.com at the top of the crontab file so it emails you any ouytput from the script. [01:56] jmarsden: i can't mail out of that box for some reason [01:56] (thats the next thing i need to fix ;p) [01:56] Then (a) fix that! and (b) set MAILTO=yourlocaluser [01:57] no way to get it to log to a text file? [01:57] Not that I know of. man 5 crontab for all the details of what you can do in there... [01:58] of course you can wrap your python in a bash script that does logger -p daemin.debug "it is running now..." [01:59] kinds of things and then runs the pythin program, if you want to debug it that way. [01:59] But I'd fix mail first, if I were you, mail is pretty vital for managing a server... logwatch will mail you stuff to take care of, etc etc. [01:59] *chuckles* I just want it to work ;p [01:59] Then do the work to make it work :) [02:00] well chances are since its based off a minimal install, i may not have a mail server installed... [02:00] which would explain some other problems i had. postfix isn't it? [02:00] Could be. Postfix is a reasonable choice of MTA, yes. [02:07] i think i worked out the original issue ;p [02:08] it shoulda been python 'path/to/run.py [02:09] Well, sure, unless your run.py has a #!/usr/bin/env python or similar header in it so the shell knows what to do. [02:10] hmm, that didn't seem to work either [02:14] mushroomblue i assume you saw my long arsed question then :P [02:16] faileas: Did you add MAILTO=yourlocalusername at the top of the crontab as I suggested earlier? [03:17] anyone remember the command to reload a users group in an open shell? [03:17] without logging in and out [03:24] xenoterracide: I'm not sure there is one? You could try exec $SHELL [03:24] jmarsden: I know there is [03:24] I just can't recall the command [03:24] I've used it [03:25] Other than starting a new shell (with exec $SHELL) I can't see how you would tell a running process to go re-read all its group info... [03:27] newgrp and sg do it [03:28] newgrp just switches the current primary group, it won't add a newly created group to an existing process [03:29] well it's now in my list of groups for my open shell [03:29] Per http://www.cyberciti.biz/faq/howto-linux-add-user-to-group/ you can try su –preserve-environment –command “$(which $SHELL) –login -i” $(whoami) [03:31] and I couldn't write to this directory if I wasn't in the http group [03:31] OK... something has changed from when I first learned that stuff then. [03:31] maybe [03:32] and it probably does change the active primary group [03:32] but it still works [03:32] xenoterracide: Yes, sounds like it has a side effect of adding the group if it didn't already exist, or something like that. [04:21] !find brctl [04:21] File brctl found in bridge-utils, selinux-policy-dev, selinux-policy-doc, selinux-policy-ubuntu, selinux-policy-ubuntu-dev (and 3 others) [04:21] Interesting. [04:22] gadmin-openvpn-server needs bridge-utils, but doesn't depend on it. [04:23] If it really *needs* it, file a packaging bug against gadmin-openvpn-server, I suppose. [04:23] Yeah, it doesn't even run unless it's there, so. Yeah, ;) [04:31] And why the heck does rsyslog-pgsql DEPEND on the server? [04:34] what is the ubuntu equivalent to redhat kickstart please? [04:35] Psi-Jack__: huh? which server? -- I don't see any odd dependencies there [04:35] rsyslog-pgsql, when I try to apt-get install rsyslog-pgsql I see it including postgresql-8.3 which is the server. [04:35] Psi-Jack__: ... how do you expect to log to a pgsql server without the server? [04:36] giovani: Over the network to a database on a different physical machine? [04:36] jmarsden: that sounds silly to separate the syslog from the database it's logging to -- a huge waste of bandwidth for no gain [04:36] giovani: It's called, CLIENT. [04:36] giovani: It doesn't need the server if you have that on a seperate machine. [04:37] so install rsyslog on the server with the database? why generate twice the network traffic? [04:37] anyway, it's a universe package, it's not well supported [04:37] giovani: If you have set up a corporate database server and want to log to it, but are not the admon of it... I can see why it might be useful in some settings. [04:37] giovani: It's not. It's on a gigabit network straight to the database server. [04:37] jmarsden: rsyslog is becoming the new default syslog for ubuntu in karmic. [04:37] but not rsyslog-pgsql [04:38] Psi-Jack__: rsyslog, but not the pgsql part... right :) [04:38] Yeah. [04:38] anyway [04:38] Which is natively part of rsyslog itself, just libraries. [04:38] oh [04:38] btw [04:38] you're wrong [04:38] it's not a dependency [04:38] it's a recommend [04:38] you didn't turn off auto-install recommends [04:38] so there's absolutely nothing wrong with the package [04:39] Oh? Hmmm.. [04:39] How do Iturn off that then? [04:39] sudo apt-get install WHATEVER --no-install-recommends [04:39] I've seen things recommended, but never had to forcefully tell it NOT to install them. [04:39] Or you can configure it somewhere under /etc/apt [04:39] Psi-Jack__: for a number of versions of ubuntu you have [04:40] Default is to install recommends now, it has been the default since... Hardy I think. [04:40] Hmmm [04:40] APT::Install-Recommends "0"; [04:40] APT::Install-Suggests "0"; [04:40] in apt.conf [04:41] Heh, guess I gotta create that, eh? [04:41] or conf.d [04:41] no [04:41] /etc/apt/apt.conf I think it is? Seems to be a zero length file here, by default. [04:42] Mine's non-existant. [04:42] jmarsden: or /etc/apt/apt.conf.d/ [04:42] All I have is the apt.conf.d [04:42] right ... use that [04:42] same thing [04:43] That's better. [04:43] And to note, Hardy did show the list of Recommended, and didn't install them by default. At least Hardy-server [04:44] it's existed for at least a few versions [04:44] Yeah... Definately not in hardy-server. I remember seeing Recommended packages almost always. [04:45] http://www.ubuntu.com/getubuntu/releasenotes/810#Recommended%20packages%20installed%20by%20default [04:45] maek: It may be out of date, but see https://help.ubuntu.com/community/KickstartCompatibility [04:45] Interesting though.. Why it doesn't depend on the postgresql-client-8.3, but /recommends/ postgresql, a virtual package for all of postgresql. [04:46] Psi-Jack__: what use is recommending/depending on the pgsql client? [04:46] Psi-Jack__: Because the expected common use case is not to move the syslog data over the network twice. [04:46] jmarsden: thanks. but preseed is the "ubuntu" way? [04:46] maek: look at http://www.ubuntu.com/products/whatisubuntu/serveredition/features/autodeploy [04:46] maek: Yes, I think so. I've not used it; years ago I used Kickstart on RedHat 5.x and 6.x :) [04:47] Still, though, it should depend specifically on the postgresql client libraries, and not just rely only on the recommends. [04:47] That is a bug. [04:47] jmarsden: I use kickstart now for rhel5, but I got a new job and its a 100% ubuntu shop :) [04:47] giovani: thanks [04:48] if you don't want to register for the pdf -- http://www.ubuntu.com/system/files/u1/AutomatedDeploymentsWP-20090126.pdf [04:49] Psi-Jack__: I'm not sure that it is a bug ... [04:49] there's no way to make a single depends/recommends setup work for every situation [04:49] How would it not be a bug? Recommends are optional recommendations, not absolutely required to run it. [04:49] they've found a middle ground by making the server/client a recommend [04:49] Psi-Jack__: not according to the debian manual [04:50] which is what the recommends being automatically installed thing is all about [04:50] No, no, that definately sounds incorrect. [04:50] That's not even the meaning of recommendations. [04:50] the english meaning of recommendation isn't relevant to what the debian manual describes them as [04:51] it's pretty common for technical terms based on english words to deviate from the definition [04:51] submit the bug report if you wqnt to ... [04:53] It is a bug, truely. [04:53] I wouldn't want to run postgreql on EVERY system I put rsyslog-pgsql on, thats just hideous. [04:53] And I'm speaking with the debian people now, about it. I am right., [04:54] Recommends are /optional/ and Not even Debian installs them by default, it just offers them. [04:54] Hi! I just set up a gui in Ubuntu server. But the /etc/X11/xorg.conf file is empty! Also, my Xorg process is taking up 100% of a cpu core. Can I resolve this? [04:54] I have the following card: [04:54] http://ati.amd.com/products/server/es1000/index.html [04:54] If rsyslog-pgsql absolutely cannot under any circumstances run without postgresql-client, then there should be a Depends: for it. [04:54] Correct. [04:54] Psi-Jack__: you should not be using multiple rsyslog-pgsql instances on different servers [04:54] poi77: Ubuntu server has no GUI. If you add one you are creating a non-standard installation. Maybe ask in #ubuntu ? [04:55] there should be one central syslog server responsible for writing to the db [04:55] postgresql-client is what it should minimally depend on, because it needs those libraries to work, period and simple, but it doesn't need the full blown server to work, as-is by itself. [04:55] jmarsden: My distributer set it up; but got the xorg.conf wrong [04:55] giovani: Wow.. Oh hell no, That'd be a HELL of a SPF! [04:55] giovani: Single Point of Failure. [04:56] giovani: rsyslog keeps track of what it cannot send to the pgsql server, and will re-spool them once it can connect. [04:56] giovani: Where-as if you used udp or tcp methods, you could loose some messages. [05:00] Any ideas? This is really important. If not, can I get paid support for ubuntu server [05:01] poi77: you're welcome to contact canonical -- but they'll likely tell you to run regular ubuntu desktop [05:01] What is canonical and I need to run server (my company specifies that) [05:02] Why would you run X on a server [05:02] That's the first question. [05:03] Canonical is the commercial company that does Ubuntu support etc if you pay them for it... the big organization behind Ubuntu, if you like. And ... the right way to run a server is to run it without a GUI. It is more secure, and leaves more resources available for doing real work, the thing the server is for... [05:04] poi77: Seriously.. Answer the question. Why would you ever run X on a server? [05:04] Ubuntu server is trying to remove everything... when runing any apt-get command it comes up with a shit ton of "no longer required:" programs to auto-remove .. only if I do autoremove them .. I wont have a working system. .. WTF! anyone else had this issue b4? [05:04] I am just configuring a machine for a customer [05:05] macrocosm: you probably removed the metapackage [05:05] poi77: you need to talk to your customer about their "need" for a gui then [05:05] Indeed. [05:05] which meta? I dont remember removing anything recently but small apps [05:05] Can you run ssh -X without a gui? [05:05] You should NEVER run X on a server. [05:05] poi77: Yes, you can. [05:05] How [05:05] ssh -X forwards X to the origin. [05:05] The GUI will be on your workstation, not on the server :) [05:06] Oh :-) they wanted a gui [05:06] I will tell them [05:06] why do they want a gui? we've asked this dozens of times [05:06] poi77: They can get it without installing Xorg on the server. They just need an X server on their workstations. [05:07] Sorry I don't know why [05:07] I think they want to attach a monitor for debugging and their tech guys need it [05:07] giovani: is there a way to convince ubuntu not to jump ... what should i google? [05:07] poi77: well like I said, you need to talk to them about this -- and learn yourself, then explain to them why it's a bad idea [05:07] poi77: plugging in a monitor is not a reason to need a gui [05:08] Why is it a bad idea? [05:08] If the GUI is used only intermittently [05:08] and not running in other cases [05:08] it's buggy, bloated code that offers no advantages whatsoever [05:08] poi77: Memory consumption, stability, the list is huge. [05:08] But of course it will only be started when needed... [05:08] so? [05:08] It should never be started, period. [05:09] that doesn't make it less buggy [05:09] there's NO reason to use it [05:09] you haven't offered one [05:09] you just keep saying it's needed [05:09] use webmin for a gui instead [05:09] no [05:09] do not use webmin, it breaks the debian way of doing things [05:09] ebox is the only supported web gui [05:09] well it works [05:10] no, it doesn't work [05:10] it's specifically not supported [05:10] Eh? [05:10] webmin is not a supported solution... use ebox, if you need a web UI. [05:10] ubuntu is clear on this [05:10] wtf is ebox [05:10] !ebox [05:10] ebox is a web-based GUI interface for administering a server. It is designed to work with Ubuntu/Debian style configuration management. See https://help.ubuntu.com/community/eBox [05:10] ive never had a single problem with webmin [05:10] macrocosm: you have a problem right now you're tryign to get us to help with [05:10] and thats using it for years [05:11] wonderful [05:11] anyawy [05:11] webmin isn't supported [05:11] Thanks for the suggestions [05:11] don't recommend it here [05:11] macrocosm: https://help.ubuntu.com/community/WebMin [05:11] on a test dev virtualized server that does not have webmin [05:11] Heh. ebox is nice, though. [05:11] I'm waiting for them to fix a few things before I truely can use it properly. ;) [05:12] Psi-Jack__: Why wait... fix them yourself and submit patches :) [05:12] I only use it for simple stuff .. so thats prolly why ive never had trouble with it [05:12] I will check out ebox tho .. sounds cool [05:12] jmarsden: It's been a loooong time since I messed with perl, let alone Mason. [05:12] macrocosm: webmin will totally ruin your clean configs [05:13] https://answers.edge.launchpad.net/ubuntu/+question/2873 has more pointers to why webmin was dropped... [05:13] My console seems to have fallen asleep [05:13] how can I wake it up [05:13] SJr: press some keys [05:13] nope [05:13] the screen isn't coming on [05:13] Turn it on? [05:13] hasnt yet ... check out . turnkey linux .. all their servers are ubuntu with webmin .. no issues [05:14] then it's probably an issue with your monitor [05:14] or your video card [05:14] or the server is not "asleep [05:14] " [05:14] well no the apm kicked in would be my guess. [05:14] uh [05:14] that's not related to the console [05:14] like to turn the monitor off [05:14] that would be a hardware-level issue [05:15] http://www.turnkeylinux.org/ how can they use it so successfully .. if you say its an ubuntu killer [05:15] macrocosm: nobody said "killer" [05:15] well .. an irritant .. perhaps lol [05:15] macrocosm: Itr isn;t us saying it, it is the Debian and Ubuntu communities, and we have provided pointers to official docs... read them :) [05:15] maybe they modify it [05:16] maybe they don't use webmin -- it's not mentioned on that page [05:16] the bottom line is [05:16] they all have webmin [05:16] it's not supported -- and hasn't been for a long time [05:16] or at least the ones ive used [05:17] if you're not willing to configure a server by hand -- you probably don't belong on linux anyhow [05:18] been using it for years ... and sometimes you start from scratch .. like my production server .. and other times a virtual quickstart is handly for testing [05:19] I couldn't agree less -- but it's a free world, and a free os [05:19] certainly .. time is money [05:20] proper, non-graphical configuration can be done 10s to 100s of times faster by someone competent [05:20] absolutly [05:20] macrocosm: BTW, does your server still have the ubuntu-minimal and ubuntu-standard metapackages installed? [05:20] this is proven by looking at all major, professional linux installations and seeing how they do things [05:20] I never use webmin for any of that .. but I like the file browser and the mysql admin pages [05:21] I asked the customer; he did killall Xorg and then ran Matlab bench. Performance is terrible [05:21] oh yes, open your files to a web interface -- that sounds safe [05:21] From the workstation I mean [05:22] poi77: So this machine is a graphical mathematics workstation? If so, install Ubuntu desktop on it. [05:22] Yes basically [05:22] Why the desktop [05:23] Does the desktop support raid [05:23] Because you seem to want local graphics. And yes. [05:23] absolutely [05:23] Is it configured automatically [05:23] Did you really do your research before offering this client Ubuntu server as a graphical math workstation?? [05:23] Heh [05:23] He asked for it; he needs virtualization [05:23] poi77: the same way it is on the server [05:25] I am running about 4 VMs on a Ubuntu desktop right here... what does virtualization have to do with server vs desktop? [05:26] Isn't server optimized for virtualization? We're talking hundreds, not 4 [05:26] Is it? In what way? How did you determine whether it it or is not optimized for your clients massively virtual workload? [05:27] Hundreds of virtual servers? on a single box? [05:27] Must be a supercomputer. [05:27] I didn't make the decision [05:27]