[00:31] I have postgres 8.4 running, but I also have 8.3 installed. How can I shut down 8.4 and start up 8.3 ? [00:38] ah ok. got it. sudo service postgresql-8.3 start [00:52] kirkland: howdy! have run into this bug 504530? [00:52] Launchpad bug 504530 in euca2ools "euca-register fails to register an image: register_image() takes at least 2 non-keyword arguments (1 given)" [Undecided,New] https://launchpad.net/bugs/504530 [00:52] kirkland: this is with the latest version of euca2ools in lucid [00:56] hi are there any chance i can increase my /var directory on my ubuntu server,during installation what i select on partion is the default.. [00:57] this is my df -h--->http://pastebin.com/m7c3cf665 [00:58] on my volume group i have this--->http://pastebin.com/m199f7cc0 [00:59] and my lv display--->http://pastebin.com/m7c83e8bd [00:59] anyone have idea..? [01:03] so, no free lvm space? [01:03] ivoks: how do i check it..? [01:04] you did vg and lvdisplay [01:04] how about pvdisplay [01:04] PV Size 297.85 GB / not usable 3.64 MB [01:04] why would you increase /var? [01:04] it's not separate partition [01:05] atm, it's size is 293GB :) [01:05] coz im having recordings on that directory and it getting full. [01:05] ? [01:05] im wrong [01:05] your / (which contains /var) is 4% in use [01:06] what i mean is, does /var directory gets its size on the main partition..? [01:06] yes [01:06] so it geeting the size of my 295.28 GB [01:07] *nod* [01:07] ow ok....i guess i dont need to increase it [01:07] you can't [01:07] you allocated all disk space to / [01:07] ok thanks.. [01:07] /var isn't separate partition anyway [01:07] its part of the /root directory right..? [01:08] nope [01:08] its part of / [01:08] ow ok whihc the size is 295 GB [01:08] /root is part of / [01:08] got it...more clear now [01:09] new to ubuntu? [01:09] http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/c23.html [01:10] yes, but i been studying it..it my server on production [01:10] http://www.pathname.com/fhs/pub/fhs-2.3.html [01:11] ^^^ good read [01:11] ivoks: Error: "^^" is not a valid command. [01:12] 'night [01:12] ivoks:one more thing where did you find that /var is only using 4%...? [01:12] ow ok goodnight [01:12] whole / uses only 4% [01:13] ok, i guess you are familiar with windows [01:13] imagine C = / [01:13] and C:\Windows = /var [01:13] ok thanks..yes clear now [01:14] notice that, unlike windows, /var (or any other path) can be directory or partition [01:15] so / is one partition, but /var can be another - this is not the case with your setup [01:16] ok [01:22] mathiaz: hey [01:22] mathiaz: hrm [01:22] mathiaz: let's poke nekro === dendrobates is now known as dendro-afk [02:32] Is anyone here? === dendro-afk is now known as dendrobates [02:38] Is anyone here? [02:50] hello to all. i have by incident deleted the mail log files in /var/log and now after restarting postfix the files are not created again. do i need to restart syslog to get the files back or do i need to create them itself? [02:51] *by accident [02:51] If the logs are created by syslog, you do not need to create them -- you might need to restart syslog, though. [02:51] If the logs are written directly by postfix, I don't know -- I would expect restarting postfix to be sufficient. [02:53] twb: thanks for your reply ! have restarted postfix and the files are not recreated. will try now to restart syslog [02:58] twb: after the restarting of syslog the files were recreated. need now to test if the logs are really written to the files. have the strange root:root ownership for this files. normally it should be if i am not wrong syslog:adm [02:59] Here, /var/log/mail.{err,info,log} are root:adm [02:59] That's on 8.04 with traditional syslog (not rsyslog) [03:00] twb: you are a genius. thank you a lot with helping me with this problem. the log files were recreated after the restart of syslog and postfix write to them with no problem :-) === dendrobates is now known as dendro-afk [03:07] It probably would have been sufficient to just pkill -HUP syslog [03:07] That is what logrotate will do each day, so this would have fixed itself tomorrow [03:18] did not thinked syslog is such intelegent to do this. love intelegent self repairing systems :-) [03:35] It's more an happy accident than "self healing" [03:46] twb: i have problem with recieving mails on my system. till yet i have used in main.cf the aliases file for maping the incoming mails to my user account now i have decided to use recipient_canonical file for this but this does not work. === erichammond1 is now known as erichammond [03:47] xperia: that is beyond me knowledge. [03:47] xperia: try #postfix [03:47] okay [03:47] they are only heavy snoobish :-) [03:48] xperia: you mean they don't want to answer your questions? [03:51] if you ask too much specific question mostly you will read answer like noob ubuntu user :-) to be a ubunut user how want a good bleeding edge system is not easy today. [03:52] Well, Ubuntu isn't about the bleeding edge. [03:52] If you want the very latest versions, you could try LFS, or SourceMage, or Gentoo. [03:55] with bleeding edge i mean having a full configured system that allow you to have a good mailserver for recieving and sending with multiple email adresses. web and dns server with multiple domains and other things that are beyond the default setup. [03:56] if you have question related to this tasks geting a answer as a ubuntu user in other chanels isnt that easy. [04:00] xperia: have you read the "Smart Questions HOWTO"? [04:03] twb: ☺ [04:04] JanC: some context, please [04:04] I also doubt the Postfix people care about what OS you use ;) [04:05] twb: the problem is that the people dont are interested to help when somebody ask a specific question. i have asked this question here since 5 min in postfix and till yet nobody answered. [04:05] do i need to have the aliases file in main.cf for maping incoming mails to my user account or can i use only this line here for this purpose. [04:05] recipient_canonical_maps = hash:/etc/postfix/recipient_canonical [04:05] xperia: don't expect a response within five minutes. [04:05] eh, 5 min is not exactly a long time [04:05] xperia: if the channel is slow, it might take HOURS to get a reply. [04:08] yeah this could be. maybe the ubuntu-server irc chanell is unique as nearly all the time people are here with good knoweledge and answer question fast. if all chanelles would be this way a lot of prblems could be allready solved and time can be saved [04:12] That's because #ubuntu-server isn't a specialist channel, and isn't full of newbies. [04:12] #emacs is similar. === dendro-afk is now known as dendrobates [05:14] okay have solved my postfix problems now. the bug was related to the file /etc/mailname [05:15] if this file contain a different location than localhost mails recieved will be not resolved on the mashine. [05:15] thanks for the help see you next time [05:16] New bug: #504615 in eucalyptus (main) "add redirect to default index.html on CLC" [Wishlist,In progress] https://launchpad.net/bugs/504615 [05:17] hi all, I am trying to fix a problem. apt seems to be broken and apt-get update keeps telling me that the files I am downloading are not bzip2 files [05:17] Get:5 http://security.ubuntu.com karmic-security/multiverse Translation-en_US [210B] [05:17] 98% [2 Translation-en_US bzip2 0] [Waiting for headers] [Connecting to bzip2: (stdin) is not a bzip2 file. [05:18] Ive tried these exact same repos on another box (ubuntu karmic desktop) and it works fine [05:19] anyone have any idea on what might be wrong? === erichammond1 is now known as erichammond === dendrobates is now known as dendro-afk [07:12] ttx: hey [07:12] kirkland: hey, still up ? [07:12] ttx: yeah, been busy [07:12] ttx: i just uploaded eucalyptus [07:12] ah ok [07:12] so it probably didn't make the last cd iso build [07:13] will ask for a manual trigger [07:13] 2am at alpha2, 3am at alpha3... [07:13] ttx: heh [07:14] kirkland: did you spend some time testing ? [07:15] kirkland: aka "did it start an instance" ? [07:15] kirkland: kernel team fixed the tun thing yesterday [07:15] ttx: i did some testing [07:15] ttx: i didn't get an instance running, though [07:15] kirkland: any issue you want to pass on ? [07:16] ttx: might not have had the kernel [07:16] kirkland: ok, then that's the subject of your tomorrow and of my today [07:16] kirkland: good night :) [07:16] ttx: i'm finishing the minutes now [07:16] kirkland: arh :) [07:17] https://wiki.ubuntu.com/MeetingLogs/Server/20100106 [07:18] ok [07:18] ttx: oh, one more ... [07:18] ttx: https://launchpad.net/bugs/461202 [07:18] Launchpad bug 461202 in eucalyptus "After purging and removing /var/lib/eucalyptus image store is out of sync" [Low,Fix released] [07:18] ttx: i couldn't reproduce that one [07:18] ttx: so i marked it fix released [07:18] ttx: it would be nice if you could confirm that [07:18] ttx: it's pretty easy [07:19] kirkland: will try [07:19] ttx: i did test your changes [07:19] ttx: they did install, start up [07:19] ttx: but there are still some upstart issues [07:20] kirkland: ok, will look with todays daily [07:21] ttx: now i'm calling it a night [08:30] Hello! [08:30] does anyone know whats gping on here? http://pastebin.com/d3e621f6b [08:44] I want to show the grub boot menu on startup what file to i need to edit? [08:46] hello to all. i have problem building software with bitbake on my uuntu server. after some 20 to 30 minutes i am getting this error message here [08:46] running task 1 of 2542 (id: 15, .../recipes/shasum/shasum-native.bb, do setscene) [08:46] running task 2 of 2542 (id: 75, .../recipes/coreutils/coreutils-native_7.2.bb,do setscene) [08:46] Out of Memory: Kill Process 11335 (python) [08:46] ERROR: Task 15 shasum-native.bb do setscene failed [08:46] it looks like python use nearly all the ram and somehow the kernel shut it down in the middle of the build proccess [08:47] people told me this is related to my operating system [08:47] how can i fix this ? [08:50] put more RAM in the computer? :-) [08:52] obviously there's some memleak, or you don't have enough ram for that build process if it's eating up all your memory [08:54] cemc: thanks for sugesstion. cant beleve that ram is the problem as the machine has 1GB RAM. have tryed to build it on a other machine iwth debian/ubuntu and i am getting also the same error message [08:58] 1GB is not a lot nowadays [08:58] did you try building it on a machine with more RAM ? [08:58] like 2GB, 4GB ? [08:59] hmmm not till yet will ask other people how had successfull builded the software how much ram did they had. [09:00] xperia: is that some opensource software, maybe I can download it and try building it if it's not too complicated, have 4gb ram in here [09:00] also on Ubuntu Karmic [09:03] cemc yeah it is open source software. it is called open embedded and is used to run linux on pocket pc phones like this here [09:03] http://www.youtube.com/watch?v=U9vg2TU0wew [09:04] give me just a moment to post the few instructions === erichammond1 is now known as erichammond [09:05] cemc: http://wiki.openembedded.net/index.php/Getting_Started [09:07] okay have aksed now the other people too and they have told me that they build this on machines with about 3 to 4 GB RAM [09:07] so need a new machine in this case hmm bad [09:08] cemc: you dont need anymore to do this. you will need a lot of time only to clone the repo with git. time for setting up the environment will take 2 to 3 Hours. [09:09] xperia: is there any readme file with requirements for building? [09:09] where it says maybe ummm... RAM: 2+GB ? :) [09:11] yes that i have also searched but didnt find it till yet. it dont even prove if the build system has enoght recources for doing this. cant answe this question as i am self searching the answer for it :-) [09:11] I can't find menu.1st what do i used to edit the grub menu [09:12] I can't find menu.1st what do i used to edit the grub menu [09:12] need some gamer system with quad core cpu [09:12] sabgenton_: what ubuntu version? [09:12] server kamic [09:12] karmic [09:13] !grub2 | sabgenton_ [09:13] sabgenton_: GRUB2 is the default Ubuntu boot manager in Karmic. For more information and troubleshooting on GRUB2 please refer to https://wiki.ubuntu.com/Grub2 [09:13] oh yeah its version 2 now [09:13] indeed :) [09:14] I did a distro-upgrade and when it got to the grub-pc package it screwed up cause I had my ide cables reversed [09:14] stupid thing [09:15] well it might be my fault I'll tell lmore later [09:17] morning [09:47] crazy on install of karmic /boot/grub/grub.cfg had hardrives based on /dev/sd somthing [09:48] i did an up date an now it's based on a UUID [09:48] I guess its so it knows what hardrive what even if there /dev/sd* location changes [09:49] !uuid | sabgenton_ yes [09:49] sabgenton_ yes: To see a list of your devices/partitions and their corresponding UUID's, run this command in a !shell: « sudo blkid » (see https://wiki.ubuntu.com/LibAtaForAtaDisks for the rationale behind the transition to UUID) [09:50] is this feature newer than the initial karmic release [09:50] ? [09:51] guntbert: why didn't karmix have this when i first installed? [09:51] was it not release at that point? [09:51] sabgenton_: its defintely not new - why it is implemented in step - no idea === erichammond1 is now known as erichammond [10:02] guntbert: in step? [10:03] sabgenton_: ah "in steps" - sorry [10:03] ah sorry [10:05] has anyone ever had a glich where at the ubuntu login prompt it gets sorta superimposed with a root prompt? [10:05] sabgenton_: no - care to !pastebin that? [10:07] i tryed typing something at this root@ubuntu prompt and then it said invaild comand [10:07] and sometimes it said invalid login or whatever l [10:07] like it was switching back and forth [10:07] guntbert: can't it's gone now [10:07] but it's kind a worrying :/ [10:08] sabgenton_: you never should have a prompt root@ubuntu - don't enable the root account [10:08] its not enabled! [10:08] !noroot | sabgenton_ [10:08] sabgenton_: We do not support having a root password set. See !root and !wfm for more information. [10:08] and i hadn't log on and gone sudo -i ether [10:09] guntbert: I rember one time it was a real pain not being able to scp as root [10:10] but anyway i'm not using the root at all [10:10] sabgenton_: ah that was you ... [10:10] ? [10:10] lol [10:10] couple of years ago [10:10] 5 [10:10] ish [10:11] sabgenton_: ok - my error - sorry - we had someone with that last week [10:11] ah [10:13] guntbert: did you tell him to passwd root then when done sudo usermod -p '!' root back [10:13] ? [10:13] (no harm?) [10:14] sabgenton_: no - I only listened to the conversation [10:14] ah see [10:14] well did someone else sugest it? [10:15] or was it considered a no no [10:15] sabgenton_: I really don't remember details [10:15] seems a bit strange not to suport it [10:16] is it somthing strange - the reason [10:16] or does ubuntu just not suport people acdently typing things with no protection of "do sudo first" [10:17] what if I need to rsync some stuff only root has access to to/from an ubuntu box (or between two ubuntu boxes)? rsyncd isn't too recommended either ;) [10:18] hmm! [10:18] whats the big deall in not suporting it [10:19] the only reason i can thing of is users stuffing things cause they don't have the little promt warning "use sudo" [10:19] probably it opens up a lot more problems in general, if people (who don't have some degree of knowledge) start using root and breaking stuff ;) [10:19] yeah that I understand as above^ [10:21] but i I need to rsync like you said and then I have a problem with rsync (unrelated to sudo vs root) are they now going to "not support me" because i didn't use sudo (imposble / impactical) [10:21] ? [10:21] seems strange [10:21] to use the linux term [10:21] FUD [10:22] I'm not really trying to fight or anything just want to know if theres something I'm missing in the tecnical [10:22] out side of lots of noobs getting carryed away [10:22] another way is to go to #linux and treat your problem as a general linux problem, disregarding the distro a bit ;) [10:23] just be careful not to mess up anything really ubuntu-specific ;) [10:23] heh [10:23] hehe [10:23] it's just linux :) [10:23] with some flavouring, heh [10:24] cemc: its funney the politics in distro chanels :) [10:24] / irc in general [10:24] if you need root, use root, just know about the dangers, and don't recommend to other people :) [10:24] sabgenton_: you did read https://help.ubuntu.com/community/RootSudo of course [11:23] is there any difference between SATA I, II and III? [11:23] are they compatible? [11:39] is this correct for crontab? 05 0-12,17-23/2 * * * (every other hour except from 12-17) ? [11:42] cemc: http://adminschoice.com/crontab-quick-reference [12:18] hi, i need help in setting up thin client [12:19] i m planning to use linux as thin client [12:19] is xubuntu good to go? [13:07] New bug: #497831 in eucalyptus (main) "Eucalyptus-SC doesn't ask the cluster-name question" [High,In progress] https://launchpad.net/bugs/497831 [13:14] Hey guys, throughout the night someone has been attempting to log in to ssh on my server, what's some kind of good software to automatically block an IP after a certain number of failed logins? I [13:14] am using Ubuntu 9.04 [13:15] Hellsheep: fail2ban [13:15] Thanks :D [13:15] works well - interfaces with iptables to block IPs for some time after a certain amount of failed logins [13:15] Cool cool, sounds like exactly what i want [13:16] :) [13:16] Is it quite easy to configure? [13:16] Bit of a noob with servers :P [13:16] yeah [13:16] brb [13:16] kk [13:17] Hellsheep: The defaults work fine for most people [13:18] Or just use SSH keys for login and block off passwords. [13:20] Thanks :) [13:20] I found some docs on it and am reading it now :) [13:20] Hellsheep: I'm using that on 20+ servers with linux and solaris - works well :) [13:21] Awesome :) [13:21] I had something like [13:21] 500-700 login attemps [13:21] I cant count exactly :P [13:22] I dunno how to [13:22] xD [13:22] Is it really worth reporting the IP's at all? [13:22] nah [13:22] it's just worms [13:22] ah i see [13:22] just block them for 30 minutes or something [13:23] blocking them too long will make it hard if you fail yourself :þ [13:25] :) [13:26] there was some thought of adding cumulatively increasing blocks, but I don't think it's been done yet [13:26] anyway - should be trivial [13:26] Is [13:26] Maxretry the number of login attemps [13:26] Before getting banned [13:27] oh - it's in already? [13:27] * RoyK checks [13:27] Hmm [13:27] I have "maxretry" [13:27] Below the time to ban them [13:29] maxretry means how many times they can retry before getting banned [13:29] Ah cool [13:30] what I meant was 'if x.x.x.x is banned and the ban is removed and x.x.x.x tries another n times, the next time it fails, it's banned for x*2^failcount minutes [13:30] Ohhh yep i see [13:30] not a big issue, though [13:30] All configured :D [13:30] That was easy [13:31] Hellsheep: what's the ip? [13:31] My IP? [13:31] yeah - lemme try :D [13:31] For the server [13:31] kk [13:31] 69.162.115.201 [13:31] Anyone know how to resolve the issue where an Outlook user sends email to a blackberry phone and its all gobbedly gook because all of the tags are showing and you are basically reading a bunch of HTML. [13:33] Hellsheep: hm. eight attempts and still not blocked [13:33] O.o [13:33] Hmm [13:33] check logs [13:33] It was set to 6 [13:33] i see the logs [13:34] set it to 3 [13:34] Jan 8 16:34:52 basetek sshd[20080]: Failed password for root from port 61249 ssh2 [13:34] no ip? [13:34] I removed it [13:34] in case [13:34] :P [13:34] Jan 8 16:34:52 basetek sshd[20080]: Failed password for root from 81.191.198.164 port 61249 ssh2 [13:34] Thats it [13:34] that's mine [13:34] people can find my IP if they want it anyway :) [13:34] Hellsheep: is fail2ban running? [13:35] Yep [13:35] One question [13:35] Could it be related to iptables not being configured? [13:35] no [13:35] it configures iptables [13:36] iptables -I INPUT -s x.x.x.x -j DROP [13:36] that sort of thing [13:36] ah okay [13:36] try iptables -vnL [13:36] see if that shows anything [13:36] Chain fail2ban-ssh (1 references) [13:36] pkts bytes target prot opt in out source destination [13:36] 0 0 RETURN all -- * * 0.0.0.0/0 0.0.0.0/0 [13:36] Only thing relating to fail2ban in there [13:37] hm.. seems fail2ban isn't parsing the logs, then [13:37] Lemme check i got it right [13:37] Okay, webmin reports my failed ssh in /var/logauth.log [13:38] I think i have it set to /var/log/logauth.log [13:38] Thats why. [13:38] My bad lol [13:38] :) [13:38] set max to 3 [13:38] lemme test again [13:38] Okay go again [13:39] 2010-01-08 16:40:48,197 fail2ban.actions: WARNING [ssh] Ban 81.191.198.164 [13:39] :) [13:40] Chain fail2ban-ssh (1 references) [13:40] pkts bytes target prot opt in out source destination [13:40] 17 912 DROP all -- * * 81.191.198.164 0.0.0.0/0 [13:40] 0 0 RETURN all -- * * 0.0.0.0/0 0.0.0.0/0 [13:40] Much better [13:40] :D [13:41] mm, fail2ban is nice. [13:45] Thanks for that RoyK [13:47] Hellsheep: :) [13:49] Hellsheep: looks like it parsed auth.log before I tried to login again [13:49] it was banned on first attempt [13:59] mathiaz: running a few mins late, sorry [13:59] jiboumans: ok [14:03] -24˚C [14:03] fuck this country [14:04] mathiaz: eta 1 min :) [14:04] RoyK: watch the language please [14:04] jussi01: do we have another anti-swearing bot? [14:05] jussi01: I use the English language. Any other prefferd? [14:05] preferred, perhaps [14:05] RoyK: the ubuntu channels have some guidelines, please follow them [14:05] !guidelines [14:05] The guidelines for using the Ubuntu channels can be found here: http://wiki.ubuntu.com/IrcGuidelines [14:05] jussi01: such as not using English? [14:05] !language | RoyK [14:05] RoyK: Please watch your language and topic to help keep this channel family friendly. [14:06] jussi01: most families knows English, sir [14:06] jussi01: or do you mean "palin friendly"? [14:07] RoyK: ok, please dont swear, as expressly stated in the guidelines [14:07] jussi01: can you please tell me what is wrong with weighting expressions? [14:08] jussi01: explain why the hell you don't want me to use certain amounts of certain languages [14:08] there's no kids in here anyway [14:08] none that haven't grown accostumed to common language, anyway [14:09] jussi01: or are you just a shell script, unable to answer straight questions? [14:10] * RoyK finds it strange that some idiot like jussi01 just complains about the language and doesn't say anything else in here [14:10] fucking bot [14:11] RoyK: Calm down. [14:12] RoyK: thanks for your patience.... my connection died... :/ [14:12] I wonde wtf wrote those "guidelines" [14:12] wonder [14:12] does everything have to be church-friendly? [14:13] RoyK: Dude. Quit being an arse. [14:13] soren: I'm not [14:13] There you go again. [14:13] RoyK: if you wish to discuss the guidelines, please feel free to join us in #ubuntu-ops [14:14] soren: I'm just concerned that some people are more obsessed about certain parts of the English language than they are about the technical issues described in here [14:14] RoyK: You're the one spending most time arguing about it. [14:15] soren: well, mr jussi01 here hasn't said a word helping others as far as I can see from my logs, but was very quick to bitch me about language. is that fair? [14:15] RoyK: Give it a rest. [14:16] I will [14:17] Thank you. [14:21] hey guys i have 2 servers running 9.10 on them. Im trying to preform an Rsync transfer without the use of compression. one server is 100Mbit both ways the other is on a 20/1mbit line. Ive ran speeds tests from both servers and they do hit their max speeds. but when transfering from the 100mbit box to the 20mbit box i cant go any faster than around 200KB/s any ideas? [14:21] smoser: ping [14:22] i should just add scp transfers are the same thing [14:23] Doonz: I guess the 1Mbps uplink performs worse than defined [14:23] ROy the 1mbit link does not upload [14:23] its the 100mbit link that does and it preforms as expected [14:24] Doonz: does scp and rsync perform similar? [14:24] Doonz: where do you see that 200KB/s ? [14:24] identical top end speeds [14:24] going from 100mb it to 20mbit [14:26] did you try to test with say iperf between them? [14:27] could there be some other limitation between them? like say an ISP limiting some traffic ? [14:28] Hi! what inetd is installed by default on hardy? (8.04.3) [14:28] abli: none. [14:29] Ok, thanks. [14:29] <_ruben> rsync probably requires a fair ammount of uploading on the remote end as well, to verify/check/determine what to transfer [14:30] cemc: whats iperf [14:30] <_ruben> network performance tester [14:31] that won't read from disk or do any compressing/other things, it (should) just tests the bandwidth. install it on both ubuntus and try it [14:35] Doonz: also make sure that the 20/1 link is not uploading anything else, because if the upload bw is full'ish, the 20mbps download could suffer [14:35] yeah yeah yeah this isnt my first time on the inter [14:35] web [14:35] ok, just making sure :) [14:38] ttx: hey [14:38] kirkland: yo [14:38] ttx: so are you able to run instances with today's code? [14:38] culdn't on a CLC / Walrus separated setup [14:39] ttx: what about all-in-one? [14:39] I'm reproducing on a classic CLC+Walrus/CC+SC to see if that's specific to Walrus separate or not [14:39] Doonz: start a rsync/ssh, then on the 20/1 end put a dstat on the interface and see what's happening (upload doesn't get full). do you rsync through ssh? maybe ssh is limited, try a rsync:// [14:39] kirkland: test in progress [14:39] ttx: k -- i'm about to rebuild my cluster [14:39] all registration should work with my latest fix [14:39] ttx: is there a topology you'd like me to focus on? === dendro-afk is now known as dendrobates [14:39] kirkland: not really, any should work [14:39] k [14:40] ttx: btw ... [14:40] note that there is an issue with eucalyptus-cc separate [14:40] bug 504704 [14:40] ttx: i setup a pxe/tftp server and i've been doing netinstalls all week [14:40] ttx: Error: Could not parse data returned by Launchpad: The read operation timed out [14:40] ttx: it's super sweet [14:40] using anna/choose_modules=eucalyptus-udeb ? [14:40] bug 504704 [14:40] ttx: yeah, preseeding that too [14:40] Launchpad bug 504704 in eucalyptus "[lucid] On CC-only setups, eucalyptus-cc fails to start at boot" [Medium,Triaged] https://launchpad.net/bugs/504704 [14:40] cool [14:41] ttx: it makes me wonder ... [14:41] I need to bring a disk and sync with your mirror [14:41] ttx: if we should create a eucalyptus-netboot package [14:41] so that I don't download 200 Gb on my small DSL line [14:41] ttx: optional, suggested, that can be installed on the CLC [14:41] ttx: sure [14:42] ttx: i have totally hands-off, no-touch installs working now [14:42] ttx: takes minutes [14:42] kirkland: interesting [14:42] ttx: an in fact ... [14:42] kirkland: what would the package specifically do ? [14:42] ttx: install tftp [14:42] ttx: set up /var/lib/tftpboot [14:43] with a few example preseeds ? [14:43] ttx: yeah, add some preseeds [14:43] sure, sounds like a good timesaver [14:43] ttx: one for each of our defined, well-supported topologies [14:43] ttx: mathiaz mentioned that there's a similar spec, for autoinstalls [14:44] ttx: anyway, i think UEC is interesting because you never install just one machine [14:44] ttx: once you've installed >= 3 machines, i find it's faster to have setup a network install service [14:44] ttx: there is one other, simpler option .... [14:44] ttx: we could make the CLC recommend squid [14:45] ttx: and make all dependent nodes use it as their caching proxy [14:45] ttx: *that* would save you a lot of time, if you had a transparent caching proxy [14:45] ttx: we did this at Intel [14:45] ttx: it worked *marvelously* [14:45] sounds like a nice best practice [14:45] ttx: i think we should default the CLC to being a squid proxy, IMHO [14:46] ttx: since all of your nodes will pull the same packages over and over and over again, on updates, and such [14:46] about bug 504704, its linked to the "eucalyptus-cc starts on started eucalyptus" issue [14:46] Launchpad bug 504704 in eucalyptus "[lucid] On CC-only setups, eucalyptus-cc fails to start at boot" [Medium,Triaged] https://launchpad.net/bugs/504704 [14:46] ttx: and since the CLC serves a preseed file, we could easily set that in one place [14:46] on a -cc only install that will fail [14:47] ttx: right so i was too tired to explain this to you last night [14:47] kirkland: so we are back at "should eucalyptus upstart script really be used to manage all eucalyptus in any configuration" [14:47] ttx: but i worked for several hours on that one [14:47] kirkland: for NC, right [14:47] ttx: right, and i have a reasonable hack that makes that work, sort of [14:47] basically, CC has the same issue, only more annoying [14:48] since CC is regularly installed with other stuff [14:48] ttx: yeah [14:48] ttx: so i'm pretty sure that something is broken with: [14:48] so having a single "stop eucayptus" is really nice there [14:48] start on (started foo and started bar) [14:48] yes, that's "The Upstart Bug" [14:48] ttx: i'm going to talk with keybuk about that now in -devel [14:49] just some background before you start [14:49] https://bugs.launchpad.net/ubuntu/+source/eucalyptus/+bug/503850/comments/1 [14:49] Launchpad bug 503850 in eucalyptus "Upstart publication scripts no longer run" [High,Fix released] [14:50] kirkland: I'll follow the discussion [14:51] kirkland: but apparently it's a known issue [14:51] kirkland: that is not planned to be fixed in lucid [14:51] kirkland: that doesn't mean there is no other way to achieve the same result [14:51] kirkland: so picking Keybuk's brain about it is good. [14:52] ttx: okay, well our other option is emitting signals [14:52] ttx: initctl emit "eucalyptus-is-running" [14:52] ttx: we can start on that [14:53] kirkland: that's worth a try [14:53] ttx: yup, apw says that's "The Upstart Bug" [14:54] not sure keybuk is around [14:54] it sounds like TUB to me. not sure if it is a bug, or intended [14:54] but its confusing what it means for sure [14:57] zul, i'll be in in ~30 minutes [14:57] smoser: k === robbiew1 is now known as robbiew [15:14] jdstrand, I was going to ask you about BUG 274530 but looks like you are already working on it. [15:14] Launchpad bug 274530 in openoffice.org "cell with german umlaut incorrectly displayed" [Undecided,Invalid] https://launchpad.net/bugs/274530 [15:14] jdstrand, Ummm...meant bug 274350 [15:14] Launchpad bug 274350 in likewise-open "apparmor HOMEDIRS not adjusted for likewise" [High,In progress] https://launchpad.net/bugs/274350 [15:14] coffeedude: yeah-- it will be a small debdiff. I'll have it soon (it's building locally) [15:15] jdstrand, cool. I'm assuming it is too late for alpha2. or not? [15:15] coffeedude: I was thinking if you approved it, I'd upload [15:16] jdstrand, Sure. should be a trivial diff. Sounds good. Just ping me and I'll review it as soon as you are ready. [15:16] coffeedude: thanks! [15:18] coffeedude: btw, as I'm sure you know, the source was accepted, and I just accepted the binaries about 30 minutes ago [15:19] jdstrand, for the new likewise-open_5.4 packages? [15:19] yeah [15:19] * coffeedude cheers [15:19] https://launchpad.net/ubuntu/+source/likewise-open/5.4.0.39949-2 [15:19] coffeedude: ^ [15:21] jdstrand, cool beans. Makes me happy. [15:21] :) [15:24] some one here to help me with ldap [15:24] ?? [15:25] hello some 1 dr to help [15:26] We can't help if you don't ask a question. [15:29] kirkland: my testing blocks on bug 504530 [15:29] Launchpad bug 504530 in euca2ools "euca-register fails to register an image: register_image() takes at least 2 non-keyword arguments (1 given)" [High,Confirmed] https://launchpad.net/bugs/504530 [15:29] ttx: oh, yeah, damn [15:29] ttx: we need neil for that one [15:29] ttx: would you send an email to Canonical-Eucalyptus? [15:30] kirkland: ok [15:30] Sorren [15:30] hi [15:31] I have got stuck in ldap from amonth [15:31] Can you tell me how to ssh using an ldap account [15:31] And while I do machine authentication is autofs nedded to load a particular home directory for remote user? [15:32] kirkland: done [15:32] smoser: im testing my ec2-config scripts on ec2 instance ill be updating the debian packaging as well fyi [15:33] ttx: cheers [15:34] kirkland: fwiw, reverting to the previous euca2ools doesn't seem to solve it [15:34] so it might indeed be a eucalyptus issue instead [15:34] ttx: i looked at the source [15:34] ttx: it's in a library somewhere [15:34] ttx: the bug is [15:35] ttx: as that one tool doesn't take more than one argument [15:35] ttx: but it sources something that makes it think that it does [15:35] arh [15:37] zul, thats cool. try starting witih my build in my ppa [15:39] smoser: ill get the things working here and then ill merge your branch and re-test them sounds ok? [15:39] bzr+ssh://bazaar.launchpad.net/~smoser/ec2-init/ec2-init.devel.pkg is pkg branch at this point. i just pushed it. [15:39] err, try bzr push lp:~smoser/+junk/ec2-init.devel.pkg instead. [15:39] zul, if you'd like i can do that work. [15:39] i can just merge your package into mine [15:40] smoser: sure ill ping you when im ready [15:40] ok. i'll be back in a bit now [15:40] ok === dendro-afk is now known as dendrobates [15:45] I'm trying to install the Java EE SDK on my 9.10 32bit server but it says it requires the DISPLAY environment variable be set - do you know what I should set it too? === dendro-afk is now known as dendrobates [15:48] zul: pastedeploty doesn't seem to have been reviewed yet ? [15:48] python-pastedeploy [15:48] its been renamed to paste === dendro-afk is now known as dendrobates [15:49] zul: what does that mean ? There is another bug about it ? [15:49] ttx: just a sec [15:51] ttx: crap hold on [15:51] ttx: yeah it hasnt been reviewed yet [15:51] ok [15:51] sorry i got confused [15:51] it needs to be reviewed [15:52] same for pastescript, right [15:53] pastescript has [15:54] it was done this morning [15:54] zul: ah, right [15:55] mathiaz: hi... is the server team using Papercuts project to fix trivial bugs on the server side or using a different project? why I ask is regarding Bug #194472 , mpt mentioned we need to keep the desktop goal of not using the terminal and it needs to be decided by the server team... [15:55] Launchpad bug 194472 in hundredpapercuts "Entering password in Terminal gives no visual feedback" [Low,Fix committed] https://launchpad.net/bugs/194472 [15:55] * ttx doesn't refresh sufficiently fast [15:56] mathiaz: to my knowledge , there havent been any server "papercuts" ... should i cancel the papercut task? [15:57] vish: we plan to have server papercuts. I'm not sure that one would be accepted as a "server papercut" though === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [15:59] ttx: from what i hear it fix is to show the stars only during user entry and it disappears once the users hits "return" ... [personally i would like that , but my main concern is regarding the papercut task] mpt mentioned to check with you guys first [15:59] s/it/the === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [16:02] vish: I'd reject it as a onehundredpapercuts bug and nominate it for the server papercuts project, whenever this starts (in a few weeks) [16:02] i'll send an email about that project soon [16:02] ttx: cool , thanks :) [16:02] we are still in the process of defining what would make a server papercut :) [16:06] ;) [16:07] ttx: isnt that bug a security risk? should it be marked so? === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [16:08] vish: depends on implementation, I suppose [16:09] hmm , ok.. i'll leave it as such === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [16:16] i have a question i'm running ubuntu-server on qemu/kvm [16:16] i'm connecting to the internet via a wireless network is there a way to attach the virtual server to the phisical network? [16:16] i want to access my server from the internet [16:18] engine252: How is the host computer connected to the network/internet? [16:18] now it is connected with a NAT configuration [16:18] oh no , wireless [16:19] the guest through NAT [16:19] guest = ubuntu-server [16:20] Pici: [16:21] engine252: I'm not sure sorry, I just wanted to make sure that the others here had enough information to answer [16:23] see i also have vmware workstation installed and there i can configure a bridged network but i can't seem to configure the same for kvm [16:40] Hello, I am having issues with likewise-open5 on Karmic. I can sucessfully log the machine onto the domain, it registers with Active Directory; however, when I reboot, I can no longer auth to LDAP. I have to log the machine off/on the domain without rebooting to be able to login in. What could be the issue? Thank you for your help. [16:44] dasunsru1e32, do you mean you are using pam/nss_ldap on your system as well as likewise-open? [16:45] dasunsru1e32, or justy that you cannot log in with domain credentials after the reboot? [16:45] I have installed likewise-open5 [16:45] and used this guide to setup: https://help.ubuntu.com/9.10/serverguide/C/likewise-open.html [16:45] it works, until I reboot [16:46] It was fine, I am not sure what is happening or why all the sudden it is failing after reboot [16:46] dasunsru1e32, can you verify that the following processes are running after reboot: lsassd, netlogond, npcmuxd [16:47] sure === dendro-afk is now known as dendrobates [16:47] hey dendrobates [16:47] root 3250 0.0 0.2 198844 7296 ? Sl 09:30 0:00 /usr/sbin/lsassd --start-as-daemon [16:47] root 2047 0.0 0.0 82428 2268 ? Sl 09:24 0:00 /usr/sbin/netlogond --start-as-daemon [16:48] root 2065 0.0 0.0 88312 1064 ? Sl 09:24 0:00 /usr/sbin/npcmuxd --start-as-daemon [16:48] Now, I have not rebooted yet [16:49] dasunsru1e32, Ahh,...yeah. Make sure after the reboot then. [16:50] ok, I will be right back, since I am on the machine that is affected [16:53] jjohansen: is it still the plan to fix bug 494565 today ? [16:53] Launchpad bug 494565 in linux "support ramdiskless boot for relavant kvm drive interfaces in -virtual" [Low,Triaged] https://launchpad.net/bugs/494565 [16:53] dasunsrule32, wb [16:53] thanks [16:54] one sec [16:54] I am checking the services [16:54] ttx: sorry that won't hit today [16:54] jjohansen: so that won't hit for alpha2 ? [16:54] right [16:54] what were the other two services? [16:54] jjohansen: ok. [16:54] lsassd is running [16:55] dasunsrule32, npcmuxd and netlogond [16:56] dasunsrule32, those 3 should be the only required ones. [16:56] They are all running, I tried to login and I can't [16:56] dasunsrule32, ok, moving to private channel then so we can debug. [16:56] ok [16:57] how do I pm in irc? [16:57] sorry, [16:58] dasunsrule32: "/query nick". But I'll open up a private chat so you should see it now. [16:58] dasunsrule32: /msg nick here comes your text [17:01] dasunsrule32: /msg google irc commands :þ [17:11] kirkland: neil explanation doesn't sound very good to me, looks like we'll have to revert to a previous boto [17:13] * ttx is not really here anymore, will be back later for catchup [17:15] ttx: urgh [17:15] ttx: well, i think it's important that we have this working for A2 [17:15] ttx: i milestoned the bug [17:22] ttx: looks like smoser did the merge [17:22] ? [17:22] smoser: so looks like the newer python-boto is breaking euca-tools, according to neil [17:22] carp [17:23] smoser: see the mail on Canonical-Eucalyptus from Neil [17:23] smoser: he recommends reverting for Lucid [17:23] smoser: i don't know how that sits with you [17:23] smoser: isnt that suppose to be "crap"? [17:24] well, i wasn't aware of "the boto upstream is going to be releasing another version soon from what I understand." [17:25] 1.9 makes parts of ec2-init much easier [17:26] its not something that couldn't be worked around, but basically, I want the boto.utils.get_instance_metadata from boto 1.9 [17:29] we can revert, but then i have to get the functionality it was giving me. [17:32] kirkland, smoser: hmhm [17:32] yeah, [17:33] i will admit that it is probably less work for me to work aroudn than to work around in euca2ools, ie, ec2-init uses less of boto than they do [17:33] smoser: yes, but you don't need the extra layer of work [17:33] but debian has 1.9 now [17:34] well, right. but neither do eucatools folks. [17:34] * ttx is kinda happy to see API issues not biting only java libraries [17:34] so , here was my thoughts [17:34] here is why i wanted 1.9 [17:34] - i needed to be able to crawl the metadata service, getting all the data that was there. [17:35] - boto1.8 has a boto.utils.get_instance_metadata() that does that, but has bugs where it doesn't get it all. [17:35] - boto1.9 gets it all [17:35] so i was looking at [17:35] a.) reimplementing it [17:36] kirkland: we need to know the extent of the APi breakage in euca2ools, to see how feasible it is to patch on our end [17:36] kirkland: if it's limited to a few functions, it might be doable for us to patch euca2ools for compatibility [17:36] b.) copying it (licensing of boto is MIT, ec2init is GPL) [17:37] kirkland: I understand that upstream doesn't want to support both [17:37] so i guess i could copy to ec2init [17:37] i dont know tha ti have the insigght to declare that from a boto perspective 1.9 will be more easily supported for 5 years [17:37] than 1.8 [17:39] kirkland: we need more detail to come up with the best solution [17:39] I've got to go -- be back later to discuss it if needed [17:39] i'm almost certain that it'd be easiest fo rme to copy parts of boto [17:40] smoser: how about overriding the functions in 1.8 with some imported 1.9 code ? [17:40] python allows for some nice function = my_function [17:40] true. yeah, its largely standalone i think. i just have to copy it somewhere. [17:41] smoser: depends on how much your desired function depends on other parts of the code, obviously [17:42] I'm gone now, back in 100 minutes or so. [17:42] 100? :) why not 2 hours? :) [17:43] metric system [17:44] lol === luis__lopez is now known as luis_lopez [17:58] le french est weird [18:57] smoser: if we are going to have runurl suppor in ec2-config then it should really be in the archive :) [19:02] zul: I've gotta run, but at some point I'd be interested to hear what is meant by "runurl support in ec2-config". [19:04] i kind of wonder what that is too. [19:04] i think the goal was to, from cloud-config support something like: [19:04] run-url: http://short.url.asdf arg1 arg2 [19:04] zul, if you want me to implement run-url, its fairly easy [19:05] smoser: yes please :) [19:05] runurl was intended to be a command line extension where you can pass it parameters and do things with return values and output just like any other shell command. If ec2-config has a way to run shell script blobs, and runurl is installed, then that gives more power than just letting them specify a URL and arguments. [19:05] smoser: i put it in my yaml file [19:06] E.g., we do things like: if runurl SOMEURL SOMEPARMS; then ...; else ...; fi [19:06] or: runurl SOMEURL SOMEPARMS > SOMEFILE [19:07] sure, easy enough to add a command to do that. i'm not really sure what the intent of that item was... lost in UDS brain [19:07] smoser: A command to do what? [19:07] if we had a 'command' named runurl [19:07] that then the cloud-config would just invoke [19:07] and could be used elsewhere too [19:08] Agreed. The AMIs I build have a command named runurl. I publish it in the Alestic PPA. [19:09] If you're going to rewrite it, please let me review so that we can make sure it is backwards compatible. For example, I use wget which enables some cool features like being able to drop the "http://" for easier to read command lines. [19:10] https://launchpad.net/~alestic/+archive/ppa [19:10] I'll be back on in a few hours. [19:12] yo ttx [19:12] zul: sssshhh. I'm not here ! [19:12] riiiight [19:16] New bug: #504897 in nut (main) "megatec_usb problem (did not claim interface)" [Undecided,New] https://launchpad.net/bugs/504897 [19:19] Hello all! I was wondering if there was a place where I could track development of LXC in Ubuntu? [19:19] phitoo: there's a blueprint for it [19:20] phitoo: https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-contextualization [19:22] jiboumans: Thanks. Now looking... :-) [19:26] jiboumans: OK! Is there some sort of status report? Some way to report issues? I expect to be testing it in the next Alpha release. [19:33] phitoo: the spec is targeted for lucid but not yet accepted. We're considering if we can add it for the next milestone (alpha3) which will be completed by late february [19:35] ttx++ # sex lead, ha! [19:36] jiboumans: way more sexy than DX or DUX === A_Tuin is now known as A-Tuin_ [19:41] jiboumans: Is there anything I can do or say to push for acceptation? I can test but not develop. [20:10] Quick question - we're about to replace a drive in an HP ProLiant server with Raid 1+0. Is there a way that we can monitor the raid rebuild progress within the Ubuntu command line? [20:13] I thought you could cat /dev/mdm$device_number [20:13] phitoo: you can propose your testing skills to the spec assignee [20:13] phitoo: I'm sure he will make good use of you ;) [20:14] Hmm - I just looked under /dev, and we have no mdm devices. I'm guessing that's for software Raid, and we're using hardware raid built into the Proliant server [20:16] cjwatson: hey, idea i wanted to run by you ... [20:16] cjwatson: tell me if i'm off the reservation [20:17] cjwatson: i find it really useful to install a caching squid proxy on my CLC [20:17] bkonkle: you would need to see if HP has any software that would let you monitor it then I believe [20:17] mdm would be s/w raid [20:17] cjwatson: and point all of the UEC components to use the CLC as its proxy server [20:17] Okay, got it. [20:17] Thank you! [20:17] no prob [20:17] cjwatson: since we're necessarily downloading the same stuff over and over and over again for potentially dozens of NCs [20:18] cjwatson: what would you think of my making that the default behavior (preseeding that proxy info into the hosted preseed files), but making it debconf (priority low or medium) changeable? [20:25] smoser: lp:~zulcss/ubuntu/lucid/ec2-init/ec2-init-config [20:29] zul: nice work ! [20:29] ttx: its really simple but its enough to get us started working on it [20:30] ergh...i mean improving it [20:30] kirkland: hmm, I don't really have an opinion either way. I imagine some would find it useful but some (many?) would have a corporate mirror already [20:31] cjwatson: agreed [20:31] kirkland: if you did it you would also have to cope with the case where the CLC itself needs a proxy to see the outside world [20:31] kirkland: seems like something to look at as low priority [20:31] cjwatson: that's already the case, though, no? [20:31] and for those folks it would result in additional load on CLC [20:31] cjwatson: yes, definitely low priority [20:31] ? [20:31] ttx: proxy work is pretty cheap [20:32] ttx: i have a powerpc mac mini that's my squid proxy :-) [20:32] network load ? CLC is hit pretty hard by clients, therically [20:32] kirkland: proxy's already handled in some ways, but it would require additional handling to also configure squid to use another proxy [20:32] cjwatson: gotcha [20:32] ttx: sure, some load [20:33] ttx: i'm talking mostly about caching debs and package updates/installs [20:33] ttx: that doesn't happen often [20:33] kirkland: hm, right [20:33] ttx: -nc's don't really talk to the outside world very often [20:33] ttx: but when they do, would be nice to suck those down at gigabit speed [20:33] ttx: especially when doing this over and over and over [20:33] ttx: which is necessarily the case with a cluster [20:34] salve a tutti [20:34] !! [20:36] ragazzi c'è qualcuno [20:36] ho problemi con Bin9 [20:36] Bind9 [20:36] una cosa molto semplice [20:37] !it [20:37] Vai su #ubuntu-it se vuoi parlare in italiano, in questo canale usiamo solo l'inglese. Grazie! (click col tasto destro sul nome del canale per entrare) [20:37] sorry :P [20:38] jjohansen, ping [20:38] pong [20:39] i think maybe kernel is locking up in ec2 instances [20:39] at least i can't explain it. it just goes away [20:40] hrmm, it boots and then dies [20:40] ie, system there, i'm using it, and then ssh console stops [20:40] reboot brings it back. [20:41] hrmm, I assume you have tried sshing back in [20:41] actually, so that means the kernel isn't dead. because i do see a [20:41] [ 32.429018] Restarting system. [20:41] yeah, i've tried again and again. [20:41] what of pinging [20:42] ping external to ec2 never works. [20:42] i'll try interneal to internal if it goes down again [20:42] where would a put a shell script so that it loads on first boot and runs as root? [20:43] I was using the --firstboot flag for ubuntu-vm-builder but it just doesn't work [20:43] I essentially want to build a vm, mount its drive, inject some files, then start they vm somewhere else [20:44] smoser: it could still be part of the kernel locking, there are some updates I need to finish rolling in [20:45] Hello, I have a local cloud vs. public cloud question. [20:46] cab938, easiest thing to do is probably to just add an rc.local like init.d script [20:51] how do i kill a process that doesn't die with kill -9? [20:52] malloc64: sudo kill -9 [20:52] okay, how do i kill a process that doesn't die with sudo kill -9? [20:53] malloc64: what process? show its line from ps aux please [20:53] /usr/bin/perl -w /usr/share/debconf/frontend /var/lib/dpkg/info/xserver-xorg.postinst configure 1:7.4~5ubuntu18 [20:53] root 6998 0.0 2.8 56352 14044 pts/2 D< 10:40 0:00 /usr/bin/perl -w /usr/share/debconf/frontend /var/lib/dpkg/info/xserver-xorg.postinst configure 1:7.4~5ubuntu18 [20:53] kirkland: there is a spec about providing an easy to setup apt proxy [20:53] sorry, there's the full [20:54] mathiaz: yeah, that's one step beyond what i'm suggesting [20:54] mathiaz: as that would require a few hundred gigs of disk [20:55] mathiaz: just a squid proxy on the CLC, to cache updates and debs, could be done pretty cheap, without requiring a lot of disk [20:55] kirkland: well - IIUC you wanna run a squid proxy on the CLC to cache the packages? [20:55] mathiaz: right -- just the ones needed, though [20:55] kirkland: right - IIRC that's what the apt proxy spec is about [20:55] mathiaz: hmm, okay [20:55] kirkland: granted - you may wanna pull down a complete archive [20:55] kirkland: but that's just another step IIRC [20:56] kirkland: having an easy way to setup an apt proxy seems to be the first step [20:56] kirkland: and could be usefull in any environement beside CLC [20:56] guntbert: root 6998 0.0 2.8 56352 14044 pts/2 D< 10:40 0:00 /usr/bin/perl -w /usr/share/debconf/frontend /var/lib/dpkg/info/xserver-xorg.postinst configure 1:7.4~5ubuntu18 [20:56] mathiaz: right [20:58] malloc64: see http://linuxgazette.net/issue83/tag/6.html [21:00] mathiaz: ttx: i'm interested in your opinion about this ... [21:00] http://paste.ubuntu.com/353624/\ [21:00] gunbert: sounds like i need to reboot. but i was in the middle of a jaunty to karmic upgrade. am i hosed? [21:00] mathiaz: ttx: to solve https://launchpad.net/bugs/487275 [21:01] Launchpad bug 487275 in eucalyptus "eucalyptus.conf should not be a conffile" [High,Triaged] [21:01] mathiaz: ttx: I'm creating /etc/eucalyptus/eucalyptus-defaults.conf [21:01] mathiaz: ttx: which will be a conffile [21:01] mathiaz: ttx: managed by us [21:01] mathiaz: ttx: and sourced at the top of /etc/eucalyptus/eucalyptus.conf [21:02] mathiaz: ttx: users (or more likely euca_conf) will change values in /etc/eucalyptus/eucalyptus.conf [21:02] mathiaz: ttx: which won't be a conf file any more [21:02] mathiaz: ttx: i'd like your sanity check on that pastebin, to make sure those values are at least *reasonable* defaults [21:03] looking [21:04] kirkland: hm - I don't understand the comment: If you want to change these values, you almost certainly [21:04] kirkland: want to either edit /etc/eucalyptus/eucalyptus.conf [21:04] kirkland: if /etc/eucalyptus/eucalyptus-defaults.conf [21:04] kirkland: is supposed to be conffile, it means that this is the file end users are supposed to edit [21:04] I'm not sure I'm getting it either... [21:05] kirkland: I think it's the other way around [21:05] doesn't eucalyptus.conf contain all values, so defaults end up being useless? [21:05] kirkland: conffile are only edit by sysadmin and should be source *last* [21:05] kirkland: the package provide a default configuration, which values can be overridden by a local sysadmin (via a conffile) [21:06] kirkland: euca_conf and package maintainer scripts should modify the package file, not the conffile [21:07] * kirkland steps back to think [21:07] ideally euca_conf would modify only a state file in /var/lib, but maybe that's a lot to ask the euca folks [21:08] and the conffile in /etc would only be modified by sysadmin [21:08] getting late for me, especially for a friday [21:08] ttx: i'm afraid that "/etc/eucalyptus.conf" is pretty thoroughly hardwired into their code [21:08] kirkland: ^^ right [21:08] mathiaz knows that stuff better than I do, I leave you in good hands :) [21:09] ttx: so we could move that to /var/lib, and symlink it back to /etc/eucalyptus for compatibility [21:09] ttx: thanks [21:09] kirkland: you'd probably want three files [21:09] mathiaz: have a moment to finish this discussion? [21:09] * kirkland is all ears [21:09] kirkland: sure [21:09] kirkland: something like that, yes [21:09] * kirkland has been looking at this for too long [21:09] * ttx has a headache [21:09] kirkland: 1. one file that is only modified by sysadmin (a conffile) [21:09] * kirkland too [21:09] kirkland: 2. one file that is modified by euca_conf [21:10] kirkland: 3. a default file that sets all the options [21:10] mathiaz: and what order or they sourced in? [21:10] kirkland: 3. sources 2, then 1 [21:11] mathiaz: is (3) a conffile too? [21:11] kirkland: as 1 is more important than 2, which is more important than 3 [21:11] kirkland: nope - only 1 is a conffile [21:11] kirkland: 3. should be in /usr/ [21:11] mathiaz: okay, let's assume that (2) *must* be /etc/eucalyptus/eucalyptus.conf (which might be a symlink to /var/lib) [21:11] shipped by the package as a regular this is the default options from the maintainer perspective [21:11] mathiaz: 3 is something in /usr/share ? [21:12] kirkland: yes [21:12] mathiaz: and 1 is actually in /etc [21:12] kirkland: as this is a default file shipped by the package maintainers [21:12] kirkland: yes [21:12] mathiaz: (while 2 and 3 are symlinks to /usr and /var) [21:12] mathiaz: got it ... [21:12] kirkland: and 2 is in /var/lib/ since euca_conf would modify it [21:12] mathiaz: now, names .... [21:12] mathiaz: yup, all good [21:13] kirkland: all of these files are shell scripts right? [21:13] mathiaz: suggestions on what to call 1 and 3? [21:13] mathiaz: yes, shell syntax [21:13] kirkland: and by default eucalyptus looks for /etc/eucalyptus/eucalyptus.conf? [21:13] mathiaz: right [21:14] mathiaz: i'd be very hesitatant to change that too much [21:14] kirkland: traditionally 1. is in /etc/defaults/ [21:14] mathiaz: ah, right [21:14] mathiaz: 1 == /etc/defaults/eucalyptus [21:14] * mathiaz nods [21:14] * mathiaz checks the FHS and debian policy [21:15] mathiaz: is (3) necessary? [21:15] mathiaz: seems like /etc/eucalyptus/eucalyptus.conf -> /var/lib/eucalyptus/eucalyptus.conf would just need to source (1) [21:16] kirkland: hm [21:17] kirkland: the problem is on package upgrades [21:17] mathiaz: i'm trying to figure out what the value of (3) is for our situation specifically [21:17] kirkland: how do you handle new options when euca_conf has already modified the file? [21:17] * kirkland thinks [21:18] kirkland: if new options are introduced, you'd have to edit (2) *without* loosing the modification from euca_conf [21:18] kirkland: from the postinstall script [21:18] kirkland: if you have (3), you just ship the new options in the new configuration file [21:21] zul, so you were leaving packaging to me? was that the plan? [21:21] mathiaz: and why would you do that in (3) rather than (1) ? [21:22] kirkland: you don't wanna touch (1) [21:22] kirkland: you start with an all commented file for (1) [21:23] kirkland: or actually - you have a comment in (1) that points to (3) for the complete list of options [21:23] kirkland: since (1) is a conffile, you wanna minimize changes made to this file [21:23] mathiaz: right, so the documentation i wanted to purge from the configuration files and put into a manpage [21:24] kirkland: that another good solution [21:24] kirkland: in which case it becomes like (3) [21:24] kirkland: ie the man page is always the refernce [21:24] mathiaz: yeah, i'm going to do that for sure, a manpage for euca_conf, and perhaps another one for the configuration options [21:24] mathiaz: possibly one manpage, with a symlink connecting the two [21:24] mathiaz: as they're closely related [21:25] kirkland: except that you need to have some similar to the man page for eucalyptus the program (that's (3)) [21:25] kirkland: *something* similar [21:26] mathiaz: hmm, okay [21:26] kirkland: on upgrades, if there a new options, you add them to the man page (for the user) and to (3) for the program [21:26] kirkland: that way you don't have to update the conffile [21:26] kirkland: and thus diminish the number of prompts [21:26] mathiaz: okay ... what needs to be taught to "use" (3)? [21:26] mathiaz: b/c everything currently sources/reads/writes (2) [21:27] kirkland: what used to read /etc/eucalyptus/eucalyptus.conf [21:27] kirkland: (3) is the main parent configuration file, which sources (2), then (1) at the end [21:27] kirkland: to have proper overriding capabilities [21:28] kirkland: so what I'd do is: [21:28] kirkland: (1) is /etc/default/eucalyptus [21:29] kirkland: (2) is /var/lib/eucalyptus/eucalyptus.conf [21:29] kirkland: (3) stays in /etc/eucalyptus/eucalyptus.conf [21:29] kirkland: however (3) is no longer a conffile [21:31] kirkland: http://www.debian.org/doc/debian-policy/ch-files.html#s-config-files [21:34] mathiaz: alrighty [21:34] kirkland: hm [21:35] kirkland: I think you're right in that we don't need (3) [21:35] kirkland: well - (1) and (3) can be merged in /etc/eucalyptus/eucalyptus.conf [21:35] kirkland: the issue with bug 487275 is that euca_conf modifies a conffile [21:36] mathiaz: Error: Could not parse data returned by Launchpad: The read operation timed out [21:36] kirkland: which leads to useless prompts on upgrades [21:36] kirkland: so ship /etc/eucalyptus/eucalyptus.conf as a conffile (and let the sysadmin customize it) [21:37] kirkland: and have another file (2) in /etc/eucalyptus/eucalyptus-euca_conf.conf modified by euca_conf [21:37] kirkland: which is then sources by /etc/eucalyptus/eucalyptus.conf [21:38] kirkland: and then /etc/eucalyptus/eucalyptus.conf would source at the *begining* /etc/eucalyptus/eucalyptus-euca_conf.conf [21:38] kirkland: so that the sysadmin could still modify the result of euca_conf [21:39] kirkland: well - it actually depends on what euca_conf modifies [21:39] kirkland: or rather sets as options [21:39] mathiaz: okay, that was pretty much what i was going for in my original approach [21:39] mathiaz: although my comments in the top of that file were inaccurrate [21:41] kirkland: well - not really actually :) [21:41] Hi all [21:41] kirkland: it's just that you wouldn't need such a file anymore === luis__lopez is now known as luis_lopez [21:42] can sshd be configured to have nic-specific authentication methods ? ie. "PKA-only" for Internet-facing, "PKA/Password" for internal network ? [21:43] MagicFab: well - you could run two instances of sshd [21:43] MagicFab: each binding to an interface whith their own configuration [21:43] mathiaz, neat. Apparently LTSP requires password auth in its LAN. [21:44] MagicFab: I don't think you can achieve the same result with only *one* sshd daemon running [21:44] I thought it generate keypairs etc. automatically [21:44] kirkland: now the downside of having only 2 eucalyptus.conf, is that you need to set *all* options possible in eucalyptus.conf [21:45] kirkland: which makes sysadmin edition a bit difficult [21:45] kirkland: having 3 configuration file would solve the end user issue [21:45] mathiaz: that's pretty much where we're at now, no? [21:45] kirkland: right [21:46] kirkland: especially if you move the documentation to a man page [21:46] mathiaz: i'm definitely moving to a manpage [21:46] kirkland: then you'd end up with a long list of options in eucalyptus.conf [21:46] kirkland: if the sysadmin opens the file, it can be a little bit difficult to edit [21:47] mathiaz, tx [21:47] kirkland: having /etc/default/eucalyptus, you'd just provide the most used options for edition [21:47] kirkland: or the sysadmin can just add the options he wants to tweak [21:48] kirkland: for example, the samba configuration file has hundreds of options, but you only see a dozen of them in the default configuration file [21:48] mathiaz: okay, stepping back from this a bit ... [21:49] mathiaz: looking at screen ... [21:49] mathiaz: there's an /etc/screenrc [21:49] mathiaz: which is the global screen configuration, a conffile, managed by ubuntu, admins can change it, but let's say that most don't [21:49] evening [21:49] mathiaz: and each user can have a ~/.screenrc [21:49] ivoks: o/ [21:50] mathiaz: screen knows to source /etc/screenrc first, then soruce ~/.screenrc for user custom overrides [21:50] kirkland: right [21:50] mathiaz: ~/.screenrc is not package managed, and hence users (or programs) can read and write that all they want [21:51] kirkland: there is a also a list of all-possible options with their default value somewhere in screen (probably compiled in the binary) [21:51] mathiaz: and the screen(1) manpage [21:52] New bug: #504960 in samba (main) "package samba-common 2:3.4.0-3ubuntu5.3 failed to install/upgrade: Unterprozess installiertes post-installation-Skript gab den Fehlerwert 128 zur?ck" [Undecided,New] https://launchpad.net/bugs/504960 [21:52] mathiaz: so i was thinking /etc/default/eucalyptus would be like our /etc/screenrc [21:52] mathiaz: conffile, pkg managed, admin can change there, but it's recommended that they use euca_conf [21:53] mathiaz: euca_conf reads/writes /etc/eucalyptus/eucalyptus.conf -> /var/lib/eucalyptus/eucalyptus.conf [21:53] mathiaz: which first sources /etc/default/eucalyptus, and then sets it's overrides from the global package defaults [21:54] kirkland: /me ponders [21:54] * mathiaz ponders [21:54] mathiaz: basically, we'll ship a set of "ubuntu" defaults [21:54] mathiaz: which are "distro" specific [21:54] mathiaz: like "use KVM" [21:55] mathiaz: then the user has more stuff that they have to configure for UEC to work [21:55] mathiaz: "site" specific stuff [21:55] mathiaz: like PUBLIC_IPS [21:58] kirkland: hm [21:59] mathiaz: so programs will keep reading/writing /etc/eucalyptus/eucalyptus.conf as they always have, for backwards compat [21:59] mathiaz: to make more FHS friendly, we move it to /var/lib, and symlink it back [21:59] mathiaz: and make it not a conffile [21:59] kirkland: well - it's a configuration file, it should stay in /etc/ [22:00] kirkland: this is what http://www.debian.org/doc/debian-policy/ch-files.html#s-config-files says [22:00] mathiaz: processed by a program [22:00] variable state, i would call it [22:00] kirkland: PUBLIC_IPS is configuration data [22:00] mathiaz: okay [22:00] kirkland: I don't think it's an internal state [22:01] kirkland: anyway I don't think it's important right now [22:01] mathiaz: alright, i'm not insistent on this [22:01] mathiaz: okay, good [22:01] mathiaz: so we, the distro, set some defaults in /etc/default [22:01] kirkland: IMO we need to figure out how many configuration files we're going to ship [22:01] as a root user, how do I watch what another user is doing from a terminal? [22:01] New bug: #504963 in bind9 (main) "[Karmic] host -4 does IPv6 lookup -- times out" [Undecided,New] https://launchpad.net/bugs/504963 [22:01] preferably without their knowledge [22:02] kirkland: to go back to bug 487275 [22:02] Launchpad bug 487275 in eucalyptus "eucalyptus.conf should not be a conffile" [High,Triaged] https://launchpad.net/bugs/487275 [22:03] kirkland: the problem, is that euca_conf modifies eucalyptus.conf and where there is a *new* version of eucalyptus.conf shipped by the package, you're prompted by dpkg [22:03] mathiaz: righto [22:04] kirkland: how often does the shipped version of eucalyptus.conf change? [22:04] mathiaz: hrm, probably once per release for most users [22:04] mathiaz: within a development cycle, perhaps more [22:05] kirkland: and we want to avoid the prompt because most of the users won't have changed eucalyptus.conf [22:05] mathiaz: actually .... [22:05] mathiaz: let's say i move all the documentation out [22:05] mathiaz: then that file is just variables/values read/written by euca_conf [22:05] kirkland: it's just that because euca_conf changed it (without the user knowledge) the user suddenly get a prompt for nothing [22:06] mathiaz: and i make it not a conf file, but just a template created by the postinst (if not yet existing) [22:06] kirkland: or rather the user gets a prompt for a package he has never configured directly (he doesn't know about eucalyptus.conf) [22:06] mathiaz: then we can change the manpage all we want [22:07] mathiaz: and the user can change (directly, or via euca_conf) all they want [22:07] mathiaz: done? [22:07] kirkland: hm - what happens if new options are introduced? [22:08] mathiaz: they're put into the manpage [22:08] kirkland: would you have to modify the file to stick the new options in there? [22:08] mathiaz: you mean where default values *must* be defined? [22:08] kirkland: yes - exactly [22:08] kirkland: is eucalyptus.conf the file where default values are defined? [22:09] mathiaz: currently, user-chosen, and almost-universally-default values are intermixed [22:09] mathiaz: i tried to pull the latter out to that file i pastebined at the top of this conversation [22:09] kirkland: right - and that's the problem with upgrades [22:10] kirkland: you need to be able set the default values for new options without bothering the user about it [22:10] mathiaz: see my "don't touch this file" comment :-) [22:10] mathiaz: so let's move that to /usr/share [22:10] kirkland: right - that's why you'd put such a file under /usr/share/package-name/ [22:11] kirkland: well - forget my last comment [22:11] kirkland: it's irrelavant for the moment [22:11] mathiaz: okay, so we'll ship a sane set of static defaults to a file in /usr/share [22:11] mathiaz: source that at the top of /etc/eucalyptus/eucalyptus.conf [22:11] kirkland: yes [22:12] kirkland: have eucalyptus.conf be a conffile [22:12] mathiaz: okay [22:12] kirkland: and have eucalyptus-local.conf be modified by euca_conf [22:13] kirkland: and sourced by eucalyptus.conf as well - *after* /usr/share/, but at the begining of the file [22:13] kirkland: so that the sysadmin has ultimate control over the configuration of eucalyptus via eucalyptus.conf [22:13] mathiaz: okay, so a) static defaults in /usr/share [22:13] kirkland: if new options are shipped during an upgrade, there are made available in /usr/sahre [22:14] mathiaz: b) user modified values in eucalyptus.conf (conffile) [22:14] kirkland: and if a new eucalyptus.conf is shipped, the end user is not prompted for a change if he hasn't modified the file [22:14] mathiaz: c) machine read/written values in eucalyptus-local.conf [22:14] kirkland: yes - seems like a good option to me [22:15] kirkland: a) /usr/share/eucalyptus/eucalytpus.conf [22:15] kirkland: b) /etc/eucalyptus/eucalyptus-local.conf (modified by euca_conf) [22:15] mathiaz: getting euca_conf to read/write eucalytpus-local.conf instead is going to be difficult, and potentially contentious [22:15] c) /etc/eucalyptus/eucalyptus.conf - sources a), then b) [22:16] kirkland: well - if euca_conf writes directly to eucalyptus.conf then users would get a prompt on a package upgrade *only* if the package ships a new eucalyptus.conf file [22:17] kirkland: which would happen less frequently then shipping a new /usr/share/eucalyptus/eucalyptus.conf file? [22:18] kirkland: which would happen less frequently then shipping a new /usr/share/eucalyptus/eucalyptus.conf file (where new defaults values for new options are added)? [22:41] hi guys..are there ways i can make image of my windows xp client pc and restore it anytime form image using my linux server, like multiple client pc deployment...any suggestion..? === Jamash- is now known as Jamash [22:42] that isn't allowed by windows license [22:42] one license per machine [22:42] :) [22:43] if I understand ruben23 correctly, he only wants to backup and restore one machine [22:46] Jamash: yes backup and restore per machince only [22:49] urgh... drbd 8.3.7rc2 build process is... urgh... [23:30] mathiaz: still around? [23:30] kirkland: sure! and you? [23:30] kirkland: not shoveling snow down there? [23:31] mathiaz: no, but it's wicked cold here too :-) [23:31] kirkland: I'm not sure we'd agree on what *wicked* cold means... :D [23:31] mathiaz: okay ... [23:31] mathiaz: http://pastebin.ubuntu.com/353685/ [23:32] mathiaz: heh :-) [23:32] mathiaz: well, it's going to be -10C here tonight [23:32] mathiaz: which is a record [23:32] mathiaz: and i'm running 20 miles (32km) tomorrow morning in that weather [23:32] kirkland: ohhh - no more coyotte then? [23:33] mathiaz: heh :-) i'm sure they're staying warm [23:33] mathiaz: okay, so see that pastebin [23:33] kirkland: oh don't worry about that - there are cross country competions running at lower temperature then that [23:33] kirkland: you'll survive :) [23:33] kirkland: look good to me [23:33] mathiaz: i think on new installs, /etc/eucalyptus/euclayptus.conf should look like that [23:34] kirkland: line 6 - maybe expand the command [23:34] mathiaz: i need to think about maintainer scripts to handle upgrades correctly [23:34] kirkland: right - on new install it should look like that [23:34] kirkland: how easy will it be to teach euca_conf to modify /etc/eucalyptus/eucalyptus-local.conf ? [23:35] mathiaz: http://pastebin.ubuntu.com/353688/ [23:35] mathiaz: not too hard ... [23:35] kirkland: cool [23:35] mathiaz: it has: [23:35] FILE="@prefix@/etc/eucalyptus/eucalyptus-local.conf" [23:36] mathiaz: however, it also has "/etc/eucalyptus/eucalyptus.conf" hardcoded elsewhere [23:36] kirkland: and add a statement in /etc/eucalyptus/eucalyptus-local.conf pointing to the eucalyptus.conf [23:36] mathiaz: i'm going to create one patch that fixes the latter using $FILE appropriately [23:36] kirkland: right seems like the best option too me [23:37] kirkland: you also need to handle the case where eucalyptus-local.conf doesn't exist [23:37] kirkland: like on first install [23:37] kirkland: would euca_conf support that? [23:37] kirkland: if not you could install an default eucalyptus-local.conf on first install [23:37] kirkland: it would *not* be a conffile [23:38] mathiaz: in the postinst, you mean [23:38] mathiaz: such that it's not a conffile? [23:38] kirkland: yes [23:38] mathiaz: ack, got it [23:38] kirkland: in the postinst [23:38] mathiaz: okay, next one: [23:38] mathiaz: http://pastebin.ubuntu.com/353689/ [23:38] kirkland: and does euca_conf support an empty/reduced eucalyptus-local.conf? [23:39] mathiaz: i'm not sure about that [23:39] mathiaz: will need to do some testing [23:39] kirkland: well - if it sources eucalyptus.conf, we're good to go [23:40] mathiaz: i think i might need to teach euca_conf about eucalyptus.conf AND eucalyptus-local.conf [23:40] mathiaz: on read, it reads eucalyptus.conf [23:40] mathiaz: but writes to eucalyptus-local.conf [23:40] kirkland: ha - true [23:40] is it just my server... but i can't seem to get 100% utilization on an amazon ec2 small instance... i get at most 40% [23:41] kirkland: line 11: want to either *directly* edit /etc/eucalyptus/eucalyptus.conf [23:41] what do I do so my machine can be a server for the home calendar when it is booted and I am not logged in? I think I just need to make the wireless connection but I have not found how to do this. And the router will be the firewall. [23:41] mathiaz: ack [23:41] kirkland: line 12: or use euca_conf to add your customizations. [23:41] kirkland: don't mention /etc/eucalyptus/eucalyptus-local.conf. [23:41] mathiaz: got it. [23:41] kirkland: actually mention it on line 8 [23:42] kirkland: /etc/eucalyptus/eucalyptus.conf. [23:42] kirkland: line8: /etc/eucalyptus/eucalyptus.conf and /etc/eucalyptus/eucalyptus-local.conf. === rberger_ is now known as rberger [23:43] mathiaz: k [23:44] kirkland: http://pastebin.ubuntu.com/353688/ - why not mention euca_conf? [23:44] kirkland: kirkland something like: you can also use use euca_conf to add your customizations - see euca_conf man page [23:45] mathiaz: it's mentioned on line 6 [23:46] kirkland: well - I'd add to the end as well [23:46] kirkland: line 10 outlines one source of documentation to customize UEC [23:47] kirkland: line 11 could mention *another* way (euca_conf) to customize UEC [23:48] zul: Where do I find ec2-runurl referenced in http://bazaar.launchpad.net/~zulcss/ec2-init/ec2-init-config/annotate/head%3A/upstart/ec2-runurl.conf [23:49] exec /usr/sbin/ec2-runurl [23:49] mathiaz: http://pastebin.ubuntu.com/353694/ [23:49] mathiaz: okay [23:50] mathiaz: this is what creates the non-conffile /etc/eucalyptus/eucalyptus-local.conf [23:51] mathiaz: this can be pretty bare, but it needs to start out with EUCALYPTUS="not_configured" [23:52] kirkland: you should probably make sure it's only run on package *installation* [23:52] kirkland: if the sysadmin decides to delete /etc/eucalyptus/eucalyptus-local.conf, it would be recreated on next package upgrade [23:52] kirkland: which may break things since the default is EUCALYPTUS="not_configured" [23:53] mathiaz: hmm, that seems highly inadvisable, removing eucalyptus-local.conf [23:53] kirkland: agreed. but the sysadmin may wanna do it [23:53] mathiaz: besides, i think we need to create that file on upgrades from 9.10 [23:53] kirkland: sure - we can add code to do that as well [23:54] mathiaz: actually, on upgrades, i think we need to mv eucalyptus.conf to eucalyptus-local.conf [23:54] kirkland: if dpkg --compare-version ...; then ... create file... ; fi [23:54] mathiaz: right, we need to compareversion and mv the current, euca_conf written eucalyptus.conf to eucalyptus-local [23:54] kirkland: right - so you'd first install the default file [23:54] mathiaz: and seed eucalyptus.conf with the 2-source version [23:55] kirkland: and deal with the package upgrade afterwards [23:55] kirkland: note that upgrades will have prompt anyway [23:56] oh, right [23:56] kirkland: since the conffile eucalyptus.conf will be modified [23:56] mathiaz: okay, so you really i need logic to only create -local on install? [23:56] kirkland: I think so. [23:56] kirkland: the logic is actually already in the postinst script [23:57] kirkland: where it deals with installing euca_root-wrap [23:57] kirkland: and does euca_conf -d / /etc/eucalyptus/eucalyptus.conf [23:57] mathiaz: oh, crap [23:57] yeah [23:58] kirkland: you'd probably gonna have to look at that code as well [23:58] mathiaz: hrm, shoudl this go in preinst, then? [23:58] kirkland: this =? [23:59] mathiaz: the -local creating codde [23:59] kirkland: I don't think so [23:59] kirkland: in preinst, /etc/eucalyptus/ won't exist