[00:00] SpamapS: do you know if perhaps koan —static-interface is broken? [00:00] SpamapS : ... [00:01] or anyone else [00:01] my xml is coming out with "ksdevice=link" instread of eth0 === Jasonn is now known as me === me is now known as Jasonn [01:04] what would be the best way to backup a full kvm based on lvm storage? [01:04] is there any good article about how to ? [01:05] I want a full snapshot backed up, with the ability to restore the full vm, any ideas? [01:14] scalability-junk, the best way? [01:14] shutdown the kvm [01:14] make a copy of the lv, using like dd [01:14] or just back it up like any other computer [01:16] patdk-lap: yeah I already got a backup from inside the vm, but I just wanted to have a way of backup up the whole vm once a month or so [01:16] so the best way would be shutdown, dd, and then backup to remot [01:16] *remote [01:16] if you don't care about data consistancy, and want to do it without a shutdown [01:16] and don't care about lvm snapshots slowing things down [01:16] then you can do a lvm snapshot [01:16] then do the dd backup from the snapshot [01:17] mh consistency would be better... any experience on how long the downtime would be of a 100G vm with dd? [01:17] LVM snapshots are O(1) [01:17] that all depends on what your coping from and where you copy to [01:18] from disk 1 to disk 1 :D [01:18] the speed of 1 disk to 1 disk [01:18] whatever speed your disks have to spare [01:18] It's faster if you write to /dev/null instead [01:18] twb mh great suggestion :D [01:18] And only slightly less reliable [01:19] twb, faster if you don't bother reading :) [01:19] patdk-lap: but then you don't have a backup! [01:19] One of my customers, they had a broken backup script [01:19] For FIVE YEARS they were dutifully swapping new tapes in each week [01:19] The backup script did nothing but eject the tape [01:19] ya, love those :) [01:20] I try to check atleast once a month [01:20] oh they could have saved so much money on tapes :D [01:21] hell, saved manpower to change them :) === Resistance is now known as EvilResistance [01:31] mhhh gzip is just using one cpu damn [01:32] well, use something better [01:32] scalability-junk: that's because it's sooper-optimized in assembly on x86 [01:33] p7zip, pigz, ... [01:33] what's better than just use dd if=/vm | gzip -9 > vm.dd [01:33] depends what you call better [01:33] patdk-lap: that's a funny way to say xz-utils [01:33] more space saved, or done faster [01:33] twb, never used xz [01:33] yeah -9 seems overkill probably [01:33] normally use 7z myself [01:33] mh I try with -1 first [01:34] xz is the same algo as 7z [01:34] But it doesn't stupidly conflate archiving with compressing [01:34] if you want faster, lzop gives ok compression at full throughput [01:34] I'm a big fan of mksquashfs foo foo.sq -compr xz [01:35] I think I go for using dd withoutg compression and then do the compression afterwards so the vm can be started up again [01:35] compression, using say lzop could make it faster [01:36] ok I'll check that out thanks [01:36] any compression that doesn't overload the cpu, means less disk writing [01:36] lzop is designed for real-time use [01:36] so it won't overload any normal cpu [01:36] I love using it for memcache type things [01:37] my estimation right now dd the whole vm would be a downtime of about 5-20 min [01:37] 40G [01:37] scalability-junk, should be around 6-7min [01:37] Just quiesce the VM, ensure the VM subsystem has flushed to disk, then LVM snapshot and back that up [01:37] unless the box is busy doing other things [01:38] Or worst case, stop the VM, make an LVM snapshot, start the VM, then do the backup from the snapshot [01:38] patdk-lap: yeah unfortunately it won't be too idle... [01:38] ya, stopping the vm, snapshot, start vm, would work also [01:38] but also make disk slower, till you remove the snapshot [01:38] would that be faster in terms of downtime? [01:38] patdk-lap: unless he's some stupid mysql user, who ccares [01:39] twb, unless you do any disk writing :) [01:39] twb: mysql is not the only thing writing to ram [01:39] I'm a fan of loading stuff into ram ;) [01:39] ram? we are talking about writing to disk, in the vm [01:40] Maybe he knows how reliable mysql databases are, so only uses them on tmpfs [01:40] * patdk-lap hasn't had a mysql issue [01:40] * scalability-junk me neither [01:40] * scalability-junk is using mysql master in ram and slave on disk sometimes :( [01:43] scalability-junk, shutdown kvm, snapshot, restart kvm [01:43] should only take very very little time [01:43] then do your backup, and remove snapshot [01:43] probably the best yeah [01:43] did that for mysql on freebsd to back it up [01:51] just to clarify for restoring .dd files it would be enough to do "dd if=vm.dd of=/path/to/new/lv/vm" right? and then start the vm with virsh --create vm.xml ? [01:52] maybe [01:52] but didn't know lvm's did that [01:52] you using files? or lvm? [01:52] lvm [01:53] so your using raw partitions, so like, /dev/mapper/vg-lv, or /dev/vg/lv [01:53] yeah [01:53] yep [01:53] or tell kvm to use the file instead of the partition :) [01:54] so instead of restoring to a partition I'm just using the .dd file? [01:54] but wouldn't that have performance issues? [01:54] doubtful [01:55] pretty sure kvm file access is about the same as partition access [01:55] there is some extra overhead [01:55] but normally not enough to bother about [01:55] the same with .img files? [01:55] now, if you used sparse files, and it gets heavily fragmented over time, that could really be a perofrmance issue [01:55] I used them before and they were a lot slower [01:55] dunno [01:56] that's why I switched to lvm ... [01:56] I haven't had an issue either way, but only tested with xen [01:56] ah ok [01:56] but then, you have to set it all up correctly [01:57] mh I'm trying the dd from file to new lv and hope it's working as supposed :D [01:58] oh damn I love gzip :D [01:59] 20G down to 200M, perfect for not fully used lvs :D [01:59] except you had to read that 20gigs :( [02:00] patdk-lap: yeah that's not perfect, but I have no other solution yet. [02:00] and it's not the main backup solution, the main one is from inside the guest which is much smarter... [02:14] any mdadm experts about [02:16] !anybody [02:16] A high percentage of the first questions asked in this channel start with "Does anyone/anybody..." Why not ask your next question (the real one) and find out? See also !details, !gq, and !poll. [02:20] ok have a raid 5 unit with 5 disks, rebooted and seems I have a non functioning raid with three disks and two spares [02:21] how can I fix my raid array [02:21] New bug: #885522 in samba (main) "package samba 2:3.5.8~dfsg-1ubuntu2.3 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/885522 [03:11] New bug: #885538 in samba (main) "package samba 2:3.5.8~dfsg-1ubuntu2.3 failed to install/upgrade: ErrorMessage: package samba is not ready for configuration cannot configure (current status `half-installed')" [Undecided,New] https://launchpad.net/bugs/885538 [03:30] trying to reassemble a raid array have four of the five disks and recreates the array, but cannot mount the array [03:40] what is the best way to create an RSA key pair in windows [03:48] i3luefire: use putty, ssh into a normal box [03:53] i thought the key gen had to happen on the machine to be used [03:56] Well, it is better to minimize the places the secret key occurs [03:56] If you move or copy the secret key, it could be intercepted at either end, or in transit [03:57] But RSA is just an assymetric crypto algorithm; it doesn't even have to be used on a computer [03:57] You could use it to encrypt paper notes to pass to your girlfriends in class if you really wanted to [03:59] so if i use putty do i need to type in the key passphrase everytime i log in [04:00] Um, it kinda depends what you want this for [04:02] hmm [04:03] well i used like a 30 char passphrase so i think im gna make it shorter [04:04] lol, i dont imagine myself sending some encrypted paper notes to my girlfriend :-) [04:04] that sound funny :-D [04:04] cos you ain't got one [04:05] girlfriend ? or encrypted message to send ? [04:05] :-P [04:05] Yes [04:07] LOL [04:21] New bug: #885554 in mysql-5.1 (main) "mysql-server-5.1 Unpacking Halts 11.04-11.10 Server Upgrade" [Undecided,New] https://launchpad.net/bugs/885554 [04:34] ok. i think i have it set up with reasonable defaults but i get the error connection refused === Guest96131 is now known as onre [07:37] hello [07:38] anyone there ? ² [07:41] nope [10:26] is there a technical reason why bash.autocompletion is disabled for root account but its enabled for regular users by default? [10:39] hello anyone there ? [10:40] dont ask to ask, just ask zetui [10:40] ask to ask what ask ? [10:40] if anyone knows it and its awake, they will probably answer you [10:40] ok what this command do : tar -C /overlay -cvf - . | tar -C /mnt/sda1 -xf - [10:42] i hate to be that guy, but: man tar , then press / and type -C (n for next ocurrences) and do so for every option... [10:43] as hints, | (pipe) redirect the output of a comand to the input of the next, - (hypen) means standar input, . (dot) means current directory [10:46] take it easy afuentes [10:46] of course i check on man tar [10:46] for -C it is written change to directory DIR [10:47] -c create a new archive [10:47] what part is unclear? [10:48] the first command changes directory to /overlay and starts archiving that dir to stdout, which is piped to another command which changes dir to /mnt/sda1 and starts to unpack stuff from stdin [10:48] sorry if i sounded rude zetui, something is not working here and its frustrating me lol [10:48] so, in a way, the command is quite similar to cp -a /overlay/* /mnt/sda1 [10:49] this one - . | [10:51] the '-' is actually argument to option -f, which takes a filename. if filename is -, tar uses stdin/stdout instead of file. '.' is the directory that tar will archive. '|' is unix pipe [10:51] afuentes told this a few lines earlier, actually [10:59] an archive should always be in a format .tar right ? correct ? [11:03] zetui: a tar archive will have a .tar or if it is also gzipped it would could be tar.gz or .tgz or if it is just compressed it could be .tar.Z [11:03] there are multiple archive formats [11:03] zetui, if u mean that a tar file should end in .tar, yes, its preferable but its not requerided [11:22] ok [11:37] hey all, i have a problem with getting a java cgi script to run with apache2 (yes i know cgi is bad, its part of a uni lab), the shell script contains "java -Xms16m Hello" and does not seem to run when accessed from url as cgi [11:37] it runs well when tested on cli, both as me and user apache [11:37] ive tried instead running a python script or echoing out the header which works [11:38] but the java program never seems to run (have it write to a file just to see, which it doesnt) [11:38] but apache does not report what is the problem, i only get "malformed header" which is quite true since the program never ran [11:45] ibatron: can u run a hello world [11:46] yea [11:46] ibatron: also i remember setting the correct environment varible. its been a while i used CGI ;-( [11:46] ibatron: anything apache error logs says ? [11:46] yea i know was surprised to see us being asked to use it but guess they want to teach us to feel pain [11:47] ibatron: tail -f /var/log/httpd/access and error will give some clue [11:47] ibatron: that sad ;-( [11:47] in my cgi script, if i echo content type it works fine [11:47] ibatron: is ur prof a sadist ;-( [11:47] apache error reports malformed header [11:47] but the header printed is fine [11:48] i know the java app never even exuctes, because if it did it would write a file [11:48] it seems like it yes :S [11:48] ill check what access says [11:49] access just says 500 (internal server error) which is because of the malformed header, which in turn is due to the java program never running [11:49] ive tried running sudo -u apache ./hello.cgi and that executes just fine [11:50] ibatron: are you setting the path in your cgi script? if not, try giving the full path to the java interpreter [11:50] yea using the full path to java [11:51] I'm not sure it should be needed anymore, but I used to have to set JAVA_HOME to the absolute path to the binary (eg, /usr/lib/jvm/java-6-openjdk/jre/bin/java on mine, not the /usr/bin symlink) [11:52] hrm it might be a path error but not the binary [11:52] i am using java Hello [11:52] it might be that it cant find Hello? [11:53] ibatron: could be. [11:53] can i specify the path to Hello somehow? [11:53] usually running all my java apps in eclipse so a bit new to cli running java [11:54] like java /var/www/Hello.class --run-main or something [11:55] hrm maybe thats what that $CLASSPATH is all about... [11:58] what flavour of linux would be best for an internet startup? [11:59] dnjaramba: ubuntu [11:59] * greppy looks at the channel name [12:00] i'd prefer something with cherry in it, possibly with a hint of vanilla [12:00] pmatulis: anyone using it out in the wild? pardon me asking [12:00] dnjaramba: is anyone using ubuntu? [12:01] dnjaramba: yes, many people are using ubuntu in the marketplace [12:01] we're probably talking tens if not hundreds of thousands of people here, i'd guess [12:01] greppy: tried CLASSPATH=/var/www/html/ then export CLASSPATH but still same issue [12:02] pmatulis: i think am trying to ask how sturdy the server version is, in terms of performance and security [12:03] dnjaramba: very good [12:03] wikimedia run some fairly trafficed sites off it [12:03] dnjaramba, i worked for a year at an internet startup and we had server infrastructure in amazon ec2, and all of our servers ran ubuntu server 11.04 lts. not a single OS-related problem occurred [12:03] dnjaramba, we didn't have more than about a dozen instances, but we had real production use without any problems, including a major launch [12:04] dnjaramba: next april will be the next LTS release (security and major bug updates for 5 years). it is more geared towards the corporate world [12:04] dnjaramba, also my hobby project has about 3.5 million page loads per month and it runs on a single dell poweredge running ubuntu server [12:04] !lts | dnjaramba [12:04] dnjaramba: LTS means Long Term Support. LTS versions of Ubuntu will be supported for 3 years on the desktop, and 5 years on the server. The current LTS version of Ubuntu is !Lucid (Lucid Lynx 10.04) [12:04] dnjaramba: I use the LTS versions for mail/web/shell servers [12:05] hmm, can someone correct that factoid to state that beginning with 12.04 desktop is supported for 5 years also? [12:06] thanks guys!! [12:06] dnjaramba: see you around [12:08] i already know alot about ubuntu for desktops, was just wondering about the server version because i wanna use ubuntu for a gig [12:08] dnjaramba: time to come onboard then. what kind of gig? [12:10] pmatilus: sort of like salesforce for schools [12:10] dnjaramba: database backend i presume? [12:10] pmatilus: yeah [12:11] dnjaramba: what database? [12:12] i am currently looking at MongoDB, neo4j and of course MySql [12:16] i am passing log level debug to syslog, but i can't see where or which console it is logging the message to [12:19] df -h [12:19] oops, sorry [12:26] greetzz === jason is now known as Guest54142 === Guest54142 is now known as jasef === lborda is now known as lborda_otp [13:28] hi... how i can know if i'm running ubuntu server or desktop uname -a and lsb_release are too similar [13:31] dender: you can cat the /etc/lsb-release [13:31] <_ruben> the most obvious difference is the lack/presence of X [13:34] dender: I also use this in a few scripts [13:35] dpkg --get-selections | grep linux-image | grep -v deinstall [13:35] I found it at: http://askubuntu.com/questions/12562/how-to-check-if-ubuntu-desktop-or-server-is-installed [13:35] that seems awful heavy to check if its server/desktop! [13:35] agreed [13:35] dpkg -l ubuntu-desktop? [13:35] `uname -a | grep server` .. if $? == 0, its a server [13:36] kiall, unlikely to work [13:36] patdk-lap: works every time .. [13:36] there are too many kernels other than -server [13:36] kiall, none of my servers have a -server kernel [13:36] normally -virtual :) [13:36] all the server ubuntu server kernels have -server? [13:36] dooh [13:36] virtual [13:37] yup [13:37] dunno if it's still the case, but for 32bit it used to be -general-pae [13:37] oh -generic-pae [13:38] but I also use -generic-pae on desktops too [13:39] I ran into the same regex pattern search headaches for desktop/server/virtual writing puppet modules [13:39] `uname -a| grep -P '(server|virtual)'` ;) [13:39] kiall, still doesn't work for 32bit servers :) [13:39] -_- [13:39] lucid 32bit uses -generic-pae [13:39] there *has* to be a better way than dpkg .. [13:39] facter provides the "operatingsystemrelease => 3.0.0-12-virtual" [13:40] oh ya, you also have -xen too :) [13:41] Kiall: it does seem like a basic need someone would have solved a while ago [13:45] let us know if you find a more efficient way [13:46] see if openssh-server is installed :) [13:47] ;) [13:57] I am running 10.04, is it possible to easily install the 2.6.35 kernel? [13:57] or even more recent kernels? [13:58] storrgie: yes, the .35 kernels are now in the LTS repo [13:58] you just have to manually select them for install [13:59] I've found 10.04 LTS handles higher IO workloads a little better with the .35 kernel [14:05] ninjix, how do I do this, my apologies I'm a little concerned about doing this on my system [14:05] I dont want to screw it up [14:05] install it, reboot [14:05] storrgie: just a moment while I log into one of my LTS boxes [14:05] if it didn't work, reboot with the old kernel :) [14:06] I have a problem with virsh and thought perhaps someone here knowns about it. when I destroy vms they disappear on one of my servers and I don't know how to fix it [14:06] so instead of saying domain1 is shutoff it just shows nothing [14:07] ninjix, thanks! [14:08] ninjix, just checking your history? :D [14:08] patdk-wk, I'm doing this on a remote system, I would have to drive about 200 miles to fix it if it doesnt work [14:08] Its in a locked closet [14:10] well, either drive if it doesn't work [14:10] or setup grub to failover on reboot [14:10] if ninjix got it working fine, I'll just take his advice [14:11] storrgie: do you have console access at all? Serial or Virtual? [14:14] ssh [14:14] I'll caution you on switching kernel families remotely [14:14] without access to the console [14:15] Sure, understood. I'm guessing you have done this on some of your remote machines [14:15] Ubuntu and Debian do an awesome job of making this easy but it doesn't alway go according to plan [14:16] This machine is quite vanilla, I think the only weird I have is a single mdadm volume [14:16] otherwise, I'm installed to a basic storage device [14:16] everything should be pretty vanilla [14:16] I find that on average about 1/25 machines need console attention for some reason after switching kernels [14:16] Do you have all of your packages updated? [14:16] ifyeah [14:16] yeah*, just did a dist-upgrade [14:16] ok [14:17] did it pickup an updated .32 kernel? [14:18] 2.6.32-34 [14:18] agd@minerva:~$ uname -r [14:18] 2.6.32-34-generic-pae [14:18] is a restart request pending? ls /var/run and look for a restart request file [14:18] nope I restarted after install [14:18] im going to restart again though, just in case [14:19] If it's restarting remotely then you *should* be OK [14:20] as soon as it comes back up I'll be ready [14:20] I rebooted it last evening also [14:20] ok, up [14:20] you've got a choice of also using the newer "Natty" kernel build with 2.6.38 [14:21] you said that 2.6.35 is in the lts tree? [14:21] aptitude search linux-image should give you the list of available kernel packages [14:22] LTS has .32, .35, .38 kernel families [14:23] I think I'll go for 2.6.35 [14:23] however I'm making this choice arbitrarily [14:23] I do need 2.6.35 or greater because I'm trying to do usb audio 2.0 [14:23] im worried that I may update my kernel and alsa may need update also [14:24] this a desktop or server? [14:25] server [14:25] I literally use it for smb, mdadm and mpd/alsa [14:26] just got a new dac (schiit bifrost) at this location, they want to play music from the thing via usb 2.0. [14:26] I don't do much with audio devices on servers [14:26] well, I can shoot for 2.6.35 and see if it works [14:26] do I literally just do a: [14:27] sudo apt-get install linux-image-2.6.35-30-generic-pae [14:27] yes [14:27] and everything else works out? [14:27] How can I change the motd so its always the same thing? [14:27] grub regens and such? [14:27] keeps getting regenerated [14:28] yup [14:28] eblow, fix your update-motd script [14:28] lets roll the dice! [14:29] have to thank all of the people that have worked so hard over the years making it so easy. :) [14:30] patdk-wk: I installed update-motd [14:30] now what [14:31] Its not changed anything [14:31] heh [14:31] rebooting! [14:32] Godspeed, man [14:33] elb0w, try man update-motd, and you probably want to UNINSTALL it [14:33] stgraber: could you push http://people.canonical.com/~serge/lxc.debdiff (for oneiric-proposed, fix two SRU bugs for xlyz) [14:37] ninjix, working! [14:37] ninjix, also, alsa picked my usb dac right up [14:37] ninjix, thank you so very much! [14:37] storrgie: excellent [14:48] Suppose I have several PEM X509 certificates and RSA private keys in a directory, and past owners haven't been all that tidy.. is there an openssl or other command that will check whether a given certificate and key match? [15:09] is there anyway I can download a source package from like precise? without having to update all my apt sources? [15:10] hallyn: looking [15:14] patdk-wk: You could try https://launchpad.net/ubuntu/precise/ , using the search. [15:14] ya, that isn't really a command line option :) [15:15] There are command-line web browsers :) [15:16] I was hoping for something like, apt-get source -t percise xxx [15:16] cause I want the source from the other system, but I don't want to screw up my updates [15:38] patdk-wk: did you read the man page for apt-get? [15:40] http://francisaltomare.com/upvotes/ [15:40] lol [15:41] elb0w: go to #ubuntu-offtopic for that stuff [15:41] o [15:41] wrong chan [15:41] my bad [15:53] hello, using qemu for linux kvms, is there a qemu-specific module that is needed? (separate from kvm) [15:54] when trying to start a vm I get an error about the supported vm types: error: internal error unable to start guest: Supported machines are: === SDCoeyer_ is now known as Zhann [16:16] hi all, when I configure ethernet interface with dhcp, who writes /etc/resolv.conf file? [16:27] filo1234: dhclient-script [16:38] Question .. iptables does not have a daemon .. upstart or init.d script. Is this part of the base package? Built into the OS ? [16:41] pdtpatrick1: I think that talks pretty much with the kernel [16:44] why would iptables need a daemon? [16:44] or are you thinking of the rhel iptables startup script to enable default firewall rules? [16:46] right something along those lines [16:46] albeit .. was just curious [16:52] anybody know how i can get cobbler to kickstart with the local squid proxy instead of gb.archive.ubuntu.com? [16:53] Is it possible for temporary downtime (let's say recurring 20 minute gaps in service) to be caused by DNS problems? [16:53] sure [16:54] A server has been unreachable today intermittently by people, whereas AWS, didn't notice any problems [16:54] but how should we know? [16:54] patdk-wk who is that to? [16:54] you know, almost anything can do that :) [16:54] patdk-wk is that directed at me? [16:54] this is the internet, no one said 100% of it is up and running 100% of the time [16:55] jamiemill, na, just to the internet irc channel ghost [16:55] SpamapS : do you happen to be here? [16:56] patdk-wk OK. I just wondered because I'm not sure how DNS works, I know it is cached so I thought it strange that the availability was intermittent. But still I suspet DNS is where the problem lies because a few times recently my browser said "name not resolved" or something, and then magically came back a few mins later [16:56] But I wonder how I can detect and prove a DNS problem [16:56] jamiemill, and the DNS in question is? [16:56] 123-reg [16:56] that isn't a dns, that is random numbers/letters [16:57] fqdn? [16:57] using a CNAME to map to a amazon load balancer [16:57] genii-around: I read right now your answer.. thanks, but can I modify that script for to set a couple of DNS like google dns for example? [16:57] don't tell me your depending on dns search to map hostnames to full names? [16:57] patdk-wk ah sorry, yes it's a XXX.XXX.com which is a CNAME pointing at AWS elastic load balancer [16:58] sorry, if your going X out everything I ask, or not give it, I can't help you [16:58] genii-around: or what I can modify for my goal? :) [16:58] if you dns name is so super secret, maybe you should use it on the internet? [16:58] patdk-wk: calm down, man :) [16:59] if i'm using an external hard disk to act as backup storage space, would it be prudent to throw additional linuxswap onto the external disk to supplement the swap on the drive which the system runs off of? [16:59] i ask because i've got most of my swap and RAM already used [16:59] EvilResistance, what does free -m, say? [17:00] filo1234: Instead of modifying the system scripts, why not instead add post-up directives to your interfaces file? [17:00] patdk-wk: it was hypothetical. its got about 5 VMs on the server [17:00] VPSes [17:00] hence the usage of RAM and SWAP [17:00] :/ [17:00]