| rump | prepend_delivered_to | 00:02 |
|---|---|---|
| rump | prepend_delivered_header rather | 00:03 |
| lifeless | SpamapS: ping | 00:24 |
| SpamapS | lifeless: pong, howdy | 00:26 |
| lifeless | I've installed lxc | 00:26 |
| lifeless | but it didn't bring in cgroup-bin | 00:26 |
| lifeless | is that optional ? | 00:26 |
| lifeless | or a missing dep ? | 00:26 |
| lifeless | also, my /etc/lxc directory is empty | 00:27 |
| lifeless | (I have no basic.conf) | 00:27 |
| lifeless | SpamapS: ^ | 00:27 |
| SpamapS | hm | 00:29 |
| SpamapS | lifeless: /etc/lxc is in fact intentional | 00:30 |
| SpamapS | lifeless: but I think cgroup-bin seems like it should be in lxc's Depends or at least Recommends | 00:30 |
| SpamapS | as its not much use w/o cgroups | 00:31 |
| lifeless | SpamapS: so, can I have your basic.conf ? | 00:35 |
| SpamapS | lxc.network.type=veth | 00:37 |
| SpamapS | lxc.network.link=virbr0 | 00:37 |
| SpamapS | lxc.network.flags=up | 00:37 |
| SpamapS | Thats it | 00:37 |
| lifeless | I'm filing bugs on the missing dep | 00:38 |
| lifeless | does it need libvirt-bin too ? | 00:38 |
| SpamapS | hallyn: if you're around.. lifeless is working with LXC and curious why it doesn't recommend/depend on cgroup-bin (I am also curious about that) | 00:40 |
| SpamapS | lifeless: I believe hallyn is EDT or CDT US, so he may be past EOD today | 00:41 |
| lifeless | bug 800456 | 00:41 |
| uvirtbot | Launchpad bug 800456 in lxc "no dependency on cgroup-bin" [Undecided,New] https://launchpad.net/bugs/800456 | 00:41 |
| SpamapS | new baby at home tho so there's no telling when he's sleeping or awake. :) | 00:41 |
| * SpamapS confirms that | 00:41 | |
| SpamapS | lifeless: note that it may be related to bug 784093 | 00:42 |
| uvirtbot | Launchpad bug 784093 in lxc "lxc broken by cgroup-bin" [High,Confirmed] https://launchpad.net/bugs/784093 | 00:42 |
| lifeless | sure | 00:42 |
| SpamapS | I posted a workaround for that bug which I'm using now | 00:42 |
| lifeless | https://dev.launchpad.net/Running/LXC#preview is what I'm building up | 00:44 |
| === medberry is now known as med_out | ||
| RoAkSoAx | Daviey can look at the by end of week | 00:47 |
| lifeless | SpamapS: how do you tell lxc-create to use the local proxy ? | 00:47 |
| keyboardtalk | I cannot access my printer on ubuntu server 10.04 from other computers (windows) on network. How can I fix this? HP printer with CUPS | 00:52 |
| SpamapS | lifeless: it runs debootstrap, which respects the appropriate environment variables I think. If they're being stripped, add them to /etc/default/lxc | 00:54 |
| lifeless | it was sudo | 00:55 |
| SpamapS | lifeless: I use MIRROR=http://127.0.0.1:9999/ubuntu in mine.. which is the path to my local approx cache | 00:55 |
| lifeless | sudo http_proxy=... lxc-create... | 00:55 |
| lifeless | SpamapS: lxc-start: open /sys/fs/cgroup/cpu/lucid-test-lp/devices.deny : No such file or directory | 01:01 |
| lifeless | lxc-start: open /sys/fs/cgroup/cpu/lucid-test-lp/devices.deny : No such file or directory | 01:01 |
| lifeless | SpamapS: I have that workaround in place already | 01:01 |
| lifeless | ls /sys/fs/cgroup/cpu | 01:02 |
| lifeless | cgroup.clone_children cgroup.event_control cgroup.procs cpu.rt_period_us cpu.rt_runtime_us cpu.shares notify_on_release release_agent tasks | 01:02 |
| SpamapS | lifeless: did you restart cgred ? | 01:04 |
| lifeless | ah | 01:05 |
| lifeless | SpamapS: robertc@lifeless-64:~$ sudo service cgred restart | 01:05 |
| lifeless | cgred start/running, process 21711 | 01:05 |
| lifeless | robertc@lifeless-64:~$ sudo lxc-start -n lucid-test-lp | 01:05 |
| lifeless | lxc-start: open /sys/fs/cgroup/cpu/lucid-test-lp/devices.deny : No such file or directory | 01:05 |
| SpamapS | Argh! | 01:06 |
| lifeless | there is a devices.deny under the devices tree | 01:06 |
| lifeless | uhm, i think I managed to flub the copy-paste | 01:06 |
| SpamapS | double Argh! | 01:07 |
| * SpamapS knows how Charlie Brown feels when Lucy pulls back the football | 01:07 | |
| lifeless | ok, *now* have this: | 01:07 |
| lifeless | mount { | 01:07 |
| lifeless | cpu = /sys/fs//cgroup/cpu; | 01:07 |
| lifeless | cpuacct = /sys/fs/cgroup/cpu; | 01:07 |
| lifeless | devices = /sys/fs/cgroup/cpu; | 01:07 |
| lifeless | memory = /sys/fs/cgroup/cpu; | 01:07 |
| lifeless | } | 01:07 |
| lifeless | robertc@lifeless-64:~$ sudo service cgred restart | 01:07 |
| lifeless | cgred start/running, process 21766 | 01:07 |
| lifeless | robertc@lifeless-64:~$ sudo lxc-start -n lucid-test-lp | 01:07 |
| lifeless | lxc-start: open /sys/fs/cgroup/cpu/lucid-test-lp/devices.deny : No such file or directory | 01:07 |
| lifeless | robertc@lifeless-64:~$ ls /sys/fs/cgroup/cpu | 01:07 |
| lifeless | cgroup.clone_children cgroup.event_control cgroup.procs cpu.rt_period_us cpu.rt_runtime_us cpu.shares notify_on_release release_agent tasks | 01:08 |
| lifeless | SpamapS: cgconfig restart appears to be the magic | 01:10 |
| SpamapS | AH right | 01:10 |
| lifeless | this doesn't look brilliant; | 01:11 |
| lifeless | Ubuntu 10.04 LTS lucid-test-lp /dev/console | 01:11 |
| lifeless | lucid-test-lp login: init: ssh main process (45) terminated with status 255 | 01:11 |
| uvirtbot | New bug: #800468 in squid (main) "package squid 2.7.STABLE9-2.1ubuntu6 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/800468 | 01:12 |
| lifeless | SpamapS: also lxc-stop doesn't seem to stop t | 01:14 |
| lifeless | *it* | 01:14 |
| SpamapS | lifeless: that is actually normal, its caused by sshd starting before the network, and I believe it gets started later when the ifup finishes | 01:14 |
| SpamapS | heh.. impossible to tell because there's no syslog | 01:14 |
| SpamapS | doh | 01:14 |
| SpamapS | lifeless: installing rsyslog seems to have made sshd not die. | 01:16 |
| lifeless | SpamapS: interesting; as you say its cosmetic though | 01:17 |
| lifeless | so I'll note that its weird and move on; I can file a bug somewhere if you like | 01:17 |
| SpamapS | lifeless: lxc-stop actually does stop the container | 01:17 |
| SpamapS | it just says it can't | 01:17 |
| lifeless | sudo lxc-stop -n lucid-lp-test | 01:18 |
| lifeless | robertc@lifeless-64:~$ echo $? | 01:18 |
| lifeless | 0 | 01:18 |
| SpamapS | which is annoying | 01:18 |
| lifeless | I dispute this 'says it cannot' thing :) | 01:18 |
| lifeless | given I can ssh into the container after running that | 01:18 |
| SpamapS | Mine went away | 01:21 |
| SpamapS | lxc-stop: failed to stop 'lucid-test2': Operation not permitted | 01:21 |
| lifeless | mine hasn't :) | 01:21 |
| SpamapS | After saying that | 01:21 |
| hallyn | SpamapS: it doesn't work well with cgroup-bin right now, in fact | 01:21 |
| lifeless | should I file a bug ? | 01:21 |
| hallyn | needs some lubin | 01:21 |
| SpamapS | I'm able to halt it from inside too. | 01:21 |
| lifeless | SpamapS: I can poweroff internally | 01:21 |
| lifeless | just not -stop externally | 01:21 |
| SpamapS | lifeless: hallyn can probably answer your questions with an order of magnitude more authority than I can. :) | 01:22 |
| hallyn | i'm guessing your problems are due to cgred | 01:23 |
| hallyn | recommend just doing mount -t cgroup cgroup /sys/fs/cgroup and thart's that | 01:24 |
| hallyn | we do need to fix cgroup-bin interaction | 01:24 |
| SpamapS | hallyn: but it would be cooler if it worked with cgroup-bin. :) | 01:25 |
| lifeless | https://dev.launchpad.net/Running/LXC#preview is what I'm writing pu | 01:25 |
| hallyn | SpamapS: patches welcome :) | 01:25 |
| hallyn | SpamapS: might work on that at the sprint | 01:26 |
| ntoombs | hello | 01:26 |
| ntoombs | i have a question | 01:26 |
| ntoombs | i am recieving an error that looks like: | 01:26 |
| ntoombs | SIOCADDRT: Invalid argument | 01:27 |
| ntoombs | Failed to bring up eth1. | 01:27 |
| ntoombs | ...done. | 01:27 |
| ntoombs | after i try to restart networking using the sudo /etc/init.d/networking restart | 01:27 |
| SpamapS | ntoombs: thats not actually a valid way to "restart networking" | 01:27 |
| ntoombs | i have my network set up on static using the device eth1 | 01:28 |
| ntoombs | what is a valid way then? | 01:28 |
| ntoombs | i've done it before using dhcp | 01:28 |
| ntoombs | but static messed it up | 01:28 |
| SpamapS | For static it *should* work... | 01:28 |
| ntoombs | right but for some reason it's not | 01:29 |
| SpamapS | ntoombs: all that does is ifdown -a and ifup -a | 01:29 |
| ntoombs | i've been into my /etc/network/interfaces file and made sure everything is "correct" as far as i know | 01:29 |
| SpamapS | ntoombs: can you do 'ifdown eth1 ; ifup eth1' ? | 01:29 |
| ntoombs | i have before and i've gotten errors | 01:29 |
| ntoombs | i will post them | 01:29 |
| ntoombs | one sec | 01:29 |
| ntoombs | sudo ifdown eth1 | 01:30 |
| ntoombs | ifdown: interface eth1 not configured | 01:30 |
| ntoombs | sudo ifup eth1 | 01:30 |
| ntoombs | SIOCADDRT: No such process | 01:31 |
| ntoombs | Failed to bring up eth1 | 01:31 |
| qman__ | first check that eth1 exists in 'ifconfig -a' | 01:31 |
| ntoombs | i've looked in about every ubuntu and linux server forum there is out there and i still havn't fixed it | 01:31 |
| qman__ | those errors are usually hardware/driver level | 01:32 |
| ntoombs | i know it's there | 01:32 |
| ntoombs | i put it there | 01:32 |
| qman__ | yes, but does it show up in ifconfig -a? | 01:32 |
| ntoombs | :( don't tell me that | 01:32 |
| qman__ | configuring it and it showing up are different things | 01:32 |
| ntoombs | yes it does | 01:32 |
| qman__ | ok | 01:32 |
| qman__ | does setting it manually with ifconfig work? | 01:32 |
| qman__ | sudo ifconfig eth1 1.2.3.4/24 up | 01:33 |
| ntoombs | i've tried that too but i'll try again | 01:33 |
| qman__ | if it does work, it's likely an issue with the configuration or ifup | 01:34 |
| qman__ | if it doesn't work, try this first | 01:34 |
| ntoombs | i use the internal ip right? | 01:34 |
| qman__ | sudo ifconfig eth1 0.0.0.0 down | 01:34 |
| qman__ | sudo ifconfig eth1 0.0.0.0 up | 01:34 |
| qman__ | then configuring it with the IP you want | 01:34 |
| ntoombs | on doing the ifconfig eth1 1.2.3.4/24 up i got a right bracket pointing to my cursur | 01:36 |
| qman__ | that's a syntax error | 01:36 |
| qman__ | check that you didn't put in any \ or " or ' | 01:37 |
| qman__ | press ctrl+c to escape | 01:37 |
| ntoombs | yea i did it wrong | 01:37 |
| hallyn | lifeless: do you need cgroups-bin for some other reason, or was that just how you were getting cgroups mounted? | 01:37 |
| ntoombs | k this time no result | 01:37 |
| ntoombs | just a new line | 01:37 |
| qman__ | that means successful | 01:38 |
| qman__ | now 'ifconfig eth1' to verify | 01:38 |
| hallyn | lifeless: assuming this is on natty or oneiric, can you try without cgroups-bin, and just with fstab mounting cgroups? | 01:38 |
| hallyn | I suspect that'll allow lxc-stop to work | 01:38 |
| ntoombs | yea it's 11.04 natty narwhal | 01:38 |
| ntoombs | ifconfig eth1 gave me what it gave me before | 01:39 |
| lifeless | hallyn: just following SpamapS instructions :) | 01:39 |
| qman__ | if it shows the address you set, it worked and there's nothing wrong with your interface | 01:39 |
| qman__ | and the problem is either with your /etc/network/interfaces or something in ifup | 01:39 |
| ntoombs | well when i had my interface setup up on dhcp | 01:40 |
| qman__ | if it shows a different or no address, you probably have something else hooking it | 01:40 |
| qman__ | like dhcpcd | 01:40 |
| ntoombs | i could ping anything all day long | 01:40 |
| ntoombs | it showed the ip address that i set to it staticaly | 01:40 |
| qman__ | so the interface is fine | 01:40 |
| qman__ | next most likely is an error in /etc/network/interfaces, pastebin it if you can | 01:41 |
| ntoombs | yea and the netmask is 255.255.255.0 | 01:41 |
| ntoombs | ok one sec | 01:41 |
| ntoombs | i'll block out the ip | 01:41 |
| qman__ | don't | 01:41 |
| ntoombs | ok | 01:41 |
| qman__ | showing your IP isn't a security risk, everyone in here is publicly showing theirs | 01:41 |
| qman__ | by connecting to freenode at all, you are | 01:42 |
| qman__ | plus, if it's behind a NAT, it's meaningless to anyone but you anyway | 01:42 |
| ntoombs | http://pastebin.com/VhyRDawJ | 01:42 |
| ntoombs | k | 01:42 |
| ntoombs | i'm a networking noob | 01:43 |
| ntoombs | thank you so much for helping me :) | 01:43 |
| qman__ | looks fine | 01:43 |
| qman__ | try this | 01:43 |
| ntoombs | i can access my router with that gateway in my browser | 01:43 |
| qman__ | sudo ifconfig eth1 0.0.0.0 up; sudo ifup eth1 | 01:43 |
| ntoombs | same error i've been getting | 01:44 |
| ntoombs | SIOCADDRT: No such process | 01:44 |
| ntoombs | Failed to bring up eth1. | 01:44 |
| hallyn | lifeless: ok, cool, then just don't use it :) | 01:45 |
| hallyn | I just have | 01:45 |
| hallyn | cgroup/sys/fs/cgroupcgroupdefaults00 | 01:45 |
| ntoombs | i don't understand hallyn | 01:46 |
| hallyn | (in fstab) | 01:46 |
| hallyn | ntoombs: sorry wasn't directed at you | 01:46 |
| ntoombs | oh sorry | 01:46 |
| ntoombs | i guess i'm missing the other side of the conversation | 01:46 |
| hallyn | ntoombs: just having a very async conversation with lifeless :) | 01:47 |
| qman__ | ntoombs, try changing in your /etc/network/interfaces file | 01:47 |
| qman__ | so that 'address 1.2.3.4' is just like that, with only one space | 01:47 |
| qman__ | on all the lines | 01:47 |
| ntoombs | ok | 01:47 |
| ntoombs | done | 01:48 |
| qman__ | and try ifup again | 01:48 |
| ntoombs | same error | 01:49 |
| ntoombs | sudo ifconfig eth1 192.168.254.115 up; sudo ifup eth1 right? | 01:49 |
| qman__ | just sudo ifup eth1 | 01:49 |
| ntoombs | yea same error | 01:50 |
| qman__ | well | 01:51 |
| qman__ | unless your interfaces file is somehow different | 01:51 |
| ntoombs | the networking restart thing worked with dhcp | 01:51 |
| ntoombs | it doesn't work with static | 01:51 |
| ntoombs | does that count as different? | 01:52 |
| qman__ | paste/pastebin the result of 'sudo route -n' | 01:52 |
| qman__ | that error message means that it can't add a route because it's outside the valid range | 01:52 |
| keyboardtalk | How do I share a cups printer? | 01:53 |
| ntoombs | http://pastebin.com/i4qcmrzu | 01:53 |
| ntoombs | the 192.168.254.115 was assigned by dhcp | 01:53 |
| ntoombs | i didn't change it | 01:53 |
| ntoombs | route doesn't seem to recongnise my default gateway | 01:54 |
| qman__ | it isn't set | 01:54 |
| ntoombs | without -n it just shows an asterix | 01:54 |
| qman__ | try this | 01:54 |
| qman__ | http://pastebin.com/index/VhyRDawJ | 01:55 |
| qman__ | change your /etc/network/interfaces to that, and reboot | 01:55 |
| qman__ | assuming .15 is not already in use | 01:56 |
| qman__ | you can use another one if it is, just choose one outside the DHCP range | 01:56 |
| qman__ | which is probably 100-200 | 01:56 |
| ntoombs | nothing is different | 01:57 |
| ntoombs | and i'm working on a headless server from an ssh connection from my mac so if i reboot i'm afraid i won't be able to get back | 01:57 |
| ntoombs | i'll check the range | 01:57 |
| qman__ | well, you can't very well change network settings without disconnecting | 01:58 |
| qman__ | are you connected to a different interface? | 01:58 |
| ntoombs | it's from 1 - 253 :P | 01:58 |
| qman__ | well, you should limit that | 01:58 |
| ntoombs | you can if you use networking restart | 01:58 |
| qman__ | and then choose an address outside of it | 01:58 |
| qman__ | using 'networking restart' would, in fact, disconnect you if the settings changed | 01:59 |
| qman__ | if you did not get disconnected, it means the settings didn't change | 01:59 |
| ntoombs | yea it did last time | 01:59 |
| ntoombs | and i had to reconnect | 01:59 |
| ntoombs | i guess i just got lucky | 01:59 |
| ntoombs | ok i'll reboot | 01:59 |
| ntoombs | your pastebin file... what was different? | 01:59 |
| qman__ | I just added network and broadcast | 02:00 |
| qman__ | and removed the extra spaces | 02:00 |
| ntoombs | oh i had that earlier | 02:00 |
| ntoombs | but i didn't reboot | 02:00 |
| ntoombs | i'll try it | 02:00 |
| qman__ | what is most likely, is something is hooking your interface and preventing the change from being made | 02:00 |
| qman__ | a reboot would solve that issue | 02:00 |
| ntoombs | i don't see the network and broadcast in there | 02:00 |
| ntoombs | its network 192.168.254.0 | 02:01 |
| qman__ | ff | 02:01 |
| ntoombs | and broadcast 192.168.254.254? | 02:01 |
| qman__ | http://pastebin.com/muYERLhT | 02:01 |
| qman__ | sorry | 02:01 |
| ntoombs | no problem :) | 02:01 |
| ntoombs | and should i change the dhcp range in my router? | 02:02 |
| qman__ | yes | 02:02 |
| qman__ | if the router tries to hand out the IP you're using, it'll be troublesome | 02:02 |
| ntoombs | k i'm doing all that right now | 02:02 |
| ntoombs | brb | 02:02 |
| qman__ | so you should open a gap in that DHCP range, and use it for your statics | 02:02 |
| twb | Or simply instruct the DHCP server to statically allocate leases by MAC for specific IPs | 02:03 |
| ntoombs | should i unplug my router for changes to take effect? | 02:04 |
| qman__ | no | 02:04 |
| ntoombs | ok | 02:04 |
| qman__ | unless your router is a really poor quality one | 02:04 |
| ntoombs | lol | 02:04 |
| qman__ | it should apply once you click save | 02:04 |
| ntoombs | it's not that bad | 02:04 |
| ntoombs | it's applied but my desktop isn't assigned a local ip anymore | 02:04 |
| ntoombs | i'm on my laptop | 02:05 |
| ntoombs | but it's no problem for now | 02:05 |
| lifeless | hallyn: sorry, was on the phone for a bit there | 02:05 |
| ntoombs | i'll edit my interfaces file now | 02:05 |
| lifeless | hallyn: I'd rather not be telling folk to mount cgroups by hand in my docs - https://dev.launchpad.net/Running/LXC#preview | 02:05 |
| hallyn | lifeless: not even using fstab? | 02:05 |
| lifeless | hallyn: particularly as folk may have cgroup-bin installed for other reasons | 02:05 |
| lifeless | which would presumably clash in fun ways | 02:06 |
| hallyn | not-quite-sleeping baby in lap, would you midn talking tomorrow? | 02:06 |
| ntoombs | rebooting now | 02:06 |
| hallyn | lifeless: i think the real problem with the cgroups is that lxc-init doesn't start in cgroup:/ like it expects. (Since you seem to have all cgrousp compsited under same dir, that's the only thing that should be going wrong) | 02:06 |
| hallyn | lifeless: if this is high priority, i can take a crack at solving this tomorrow | 02:07 |
| hallyn | should be about a day's work, and like i say, it needs to be done sometime | 02:07 |
| lifeless | hallyn: well, I have a workaround (the new cgconfig from SpamapS) for folk using natty, which most lp devs are atm | 02:08 |
| ntoombs | qman__ looks like it's gonna time out and it's still not showing up on my router | 02:08 |
| lifeless | only a few are brave enough to be on oni yet | 02:08 |
| lifeless | hallyn: if we can get it sorted for oni release, that would be awesome | 02:09 |
| qman__ | ntoombs, it's not getting a DHCP lease, so it won't show up in the router | 02:09 |
| qman__ | try pinging the IP address you gave it | 02:09 |
| hallyn | lifeless: so it all works with SpamapS' config? | 02:09 |
| lifeless | hallyn: well, it doesn't shutdown on lxc-stop | 02:09 |
| lifeless | hallyn: but poweroff inside it appears to work | 02:09 |
| ntoombs | qman__: it would ping once then timeout | 02:09 |
| hallyn | lifeless: ok, cool, if you can get work done with that, that's great | 02:10 |
| lifeless | we certainly can; I like to report all friction I run into though. | 02:10 |
| qman__ | ntoombs, does it keep doing so? | 02:10 |
| ntoombs | qman__: dont i need it to show up on the router to do port forwarding and connect to the internet? | 02:10 |
| lifeless | hallyn: I'll file a separate bug about the lxc-stop not working | 02:10 |
| hallyn | lifeless: cool, i'll put cgroup-bin interaction high on todo list meanwhile | 02:10 |
| qman__ | as in, one response per ping? | 02:11 |
| hallyn | lifeless: ok, thanks | 02:11 |
| lifeless | hallyn: sweet, thanks | 02:11 |
| ntoombs | qman__: yes it times out every time after the first packet | 02:11 |
| lifeless | hallyn: one thing that would be -awesome- | 02:11 |
| lifeless | hallyn: would be a template that bind mounts /home and uses the same uids | 02:11 |
| qman__ | so if you stop and start a new ping, the first packet gets a response, but none after that? | 02:11 |
| ntoombs | qman__: i get one good packet then it times out every second after | 02:11 |
| ntoombs | qman__: correct | 02:11 |
| qman__ | that's either a hardware problem, a driver problem, or a really bizarre firewall problem | 02:12 |
| hallyn | lifeless: shouldn't be too bad. would you want just /etc/shadow entries for uids 1000+ copied over, or the whole thing? | 02:12 |
| lifeless | hallyn: let me describe my use case; then you can decide :) | 02:12 |
| qman__ | physically, how are you connected? | 02:13 |
| qman__ | direct to the router, or through a secondary switch? | 02:13 |
| lifeless | hallyn: I want to do LP development in a lxc container rather than a VM | 02:13 |
| hallyn | lifeless: the thing i've been using a lot (from ppa) is lxc cloning with LVM. Seems like something which could be good for lp dev | 02:13 |
| ntoombs | qman__: well the server is connected by ethernet direct to the router. and i connect via ssh from a terminal on my mac | 02:13 |
| lifeless | hallyn: so in the container I want to install rabbitmq, postgresql, apache etc | 02:13 |
| lifeless | hallyn: and I want my source code sitting in my /home/robertc | 02:13 |
| qman__ | mac is over ethernet or wifi? | 02:13 |
| lifeless | hallyn: where I can edit it as normal, commit with my bzr & gpg keys etc | 02:14 |
| keyboardtalk | How do I access a shared printer from Windows? | 02:14 |
| ntoombs | wifi | 02:14 |
| ntoombs | i can do eathernet though | 02:14 |
| qman__ | try plugging into ethernet, and trying | 02:14 |
| qman__ | it might be a wifi thing on the router | 02:14 |
| lifeless | hallyn: then run the tests within the lxc container, using the lxc postgresql, rabbit apache | 02:14 |
| ntoombs | ok | 02:14 |
| ntoombs | and again thanks a ton for helping :) | 02:14 |
| ntoombs | most people would have given up by now | 02:14 |
| ntoombs | this might disconnect me | 02:15 |
| qman__ | not a problem, some routers do some strange things in the name of 'security' | 02:15 |
| lifeless | hallyn: so inside the lxc it needs to write to me homedir without messing stuff up, but I don't care about service accounts | 02:15 |
| hallyn | lifeless: would you want the whole /home bind-mounted? just ~/src or something? | 02:15 |
| lifeless | hallyn: whole home/$me would be easiest to reason about I think | 02:16 |
| hallyn | 'without messing stuff up' meaning not changing owner uids, right? | 02:16 |
| lifeless | yeah | 02:16 |
| hallyn | ok | 02:16 |
| lifeless | oh for bonus points, | 02:16 |
| hallyn | lifeless: i think i'd use that myself :) would that be higher prio for you than cgroup-bin? | 02:16 |
| lifeless | definitely | 02:16 |
| lifeless | hallyn: bug 798476 is also a very interesting one for me | 02:17 |
| uvirtbot | Launchpad bug 798476 in lxc "it would be really useful to be able to make i386 lxc containers on a amd64 machine" [Undecided,New] https://launchpad.net/bugs/798476 | 02:17 |
| lifeless | hallyn: LP memory use is much lower on i386 | 02:17 |
| ntoombs | qman__: my router says i'm still connected wirelessly and my ping comes back the same. i have airport turned off co i can't connect to anything wirelessly | 02:17 |
| lifeless | hallyn: shall I file a bug about the developer bind-mounting template [option, whatever - i dunno how it all hangs together yet] | 02:17 |
| hallyn | lifeless: sure, that's good. (not sure hwo kosher it is to just add things to the blueprint right now) | 02:18 |
| qman__ | well, I don't know anything about macs, but you didn't even time out, so you'd have to have the same IP address | 02:18 |
| qman__ | which is unlikely if you switched to wired | 02:18 |
| hallyn | lifeless: hm, people *have* run i386 containers on amd64 | 02:18 |
| hallyn | but i can't remember who | 02:19 |
| lifeless | yeah | 02:19 |
| ntoombs | qman__: i really think i just have a dumb router | 02:19 |
| lifeless | its probably just a case of passing the arch flag to debootstrap | 02:19 |
| twb | hallyn: I do it | 02:19 |
| qman__ | ntoombs, entirely possible | 02:19 |
| lifeless | it *should* be trivial | 02:19 |
| twb | Actually, I did it only to see if I could; I don't in productioin | 02:19 |
| lifeless | twb: is it more than ^ | 02:19 |
| twb | http://paste.debian.net/120603/ | 02:19 |
| twb | Look in there for references to $arch | 02:20 |
| qman__ | if you think that's the case, I'd give the router a reboot then | 02:20 |
| lifeless | twb: thanks | 02:20 |
| ntoombs | sounds good | 02:20 |
| ntoombs | brb | 02:20 |
| twb | Looks like only issue is debootstrap | 02:20 |
| twb | And as a knock-on effect, which debootstrap cache I copy | 02:20 |
| hallyn | twb: nifty, thanks | 02:22 |
| lifeless | hallyn: bug 800482 | 02:24 |
| uvirtbot | Launchpad bug 800482 in lxc "a home/$me bind-mounted template would rock" [Undecided,New] https://launchpad.net/bugs/800482 | 02:24 |
| twb | Would it? | 02:24 |
| lifeless | it would | 02:24 |
| ntoombs_ | qman__: i'm back | 02:25 |
| qman__ | any luck? | 02:25 |
| ntoombs_ | qman__: now i'm getting errors like ping: sendto: host is down | 02:25 |
| ntoombs_ | and ping: sendto: no route to host | 02:25 |
| ntoombs_ | but now my router says i'm on ethernet :) | 02:26 |
| hallyn | lifeless: cool, thanks | 02:26 |
| qman__ | I'd try restarting the server again now that the router has rebooted, if you can reach the power button, just press it once quickly and it should power down | 02:27 |
| qman__ | then press again to boot up | 02:27 |
| qman__ | otherwise, if you can plug in a keyboard and press ctrl alt del | 02:27 |
| ntoombs_ | ok | 02:27 |
| keyboardtalk | How does one access a shared printer (CUPS) hosted on ubuntu 10.04 from Windows client? | 02:27 |
| ntoombs_ | quick press of a button did not a powerdown make | 02:28 |
| ntoombs_ | i could hold it but i'm assuming that is not advisable? | 02:28 |
| qman__ | generally no | 02:28 |
| qman__ | but if you're not running any software yet it shouldn't cause too much trouble | 02:28 |
| qman__ | what filesystem are you using? | 02:28 |
| ntoombs_ | ext4 journaled so it's good | 02:28 |
| qman__ | yeah | 02:29 |
| Feadin | Just did an apt-get update && apt-get upgrade and it wants to upgrade language-pack-en which now suddenly has LOTS of crazy dependencies, including firefox and many X libs... anyone noticed this? I'm using v11.04 | 02:29 |
| ntoombs_ | to change anything i have to swap the hard drive to my desktop | 02:29 |
| ntoombs_ | but we're just doing a reboot right? | 02:29 |
| qman__ | right | 02:29 |
| qman__ | if it still doesn't come online at all, you might have to do that though | 02:29 |
| twb | Feadin: no, but I know how to bypass the "helpful" locale packages and just generate the one you want | 02:29 |
| qman__ | may be that your router is filtering it out | 02:29 |
| ntoombs_ | whad do you suggest i change in my router? | 02:30 |
| twb | Feadin: uninstall those, then "locale-gen en_AU.UTF-8; update-locale LANG=en_AU.UTF-8", where "en" is your language and "AU" is your country. | 02:30 |
| ntoombs_ | its a 2WIRE 2701HG | 02:31 |
| qman__ | ah, that's the issue exactly then | 02:31 |
| Feadin | So you're suggesting we should bypass bugs and pretend that they don't exist? :) Just wanted to check if this is actually a bug or a new "feature" to simplify the maintainers jobs ;) | 02:31 |
| twb | Feadin: I'm suggesting you should do it the Debian way | 02:31 |
| twb | Basically because I don't like the "feature" | 02:31 |
| qman__ | sounds odd but 2wire DSL routers ONLY allow connectivity with DHCP-leased clients | 02:31 |
| qman__ | if you want to get around that limitation you need a better router | 02:31 |
| ntoombs_ | how about different firmware? | 02:32 |
| qman__ | theoretically possible but I don't know of any | 02:32 |
| ntoombs_ | m0n0wall? | 02:32 |
| qman__ | that firmware would allow it, but I don't know that it would run on your router | 02:32 |
| ntoombs_ | i was wanting to change it to that anyway | 02:32 |
| twb | Feadin: AFAICT ubuntu default behaviour is more or less to installing *every* english locale for every CLI package as "language-pack-en-base", and every english locale for every GUI package as "language-pack-en". | 02:32 |
| Feadin | I get it, but if I wanted to use Debian I'd be using Debian man :) Come on I'm not talking about personal preferences here. I'm just asking if anyone knows anything about this crazy dependencies which IMO shouldn't be on a server | 02:33 |
| qman__ | plus you need to make sure your ISP isn't doing some funny things with it | 02:33 |
| twb | Feadin: so in a default lucid server network install you lose about 200MB to thunderbird and firefox localization | 02:33 |
| qman__ | AT&T U-verse is notorious for this | 02:33 |
| twb | Feadin: shrug. Report a bug, then. | 02:33 |
| ntoombs_ | its actually windstream | 02:34 |
| qman__ | if the router allows DHCP reservations, that may be the easiest way to go | 02:34 |
| Feadin | Yeah I know I should report a bug, but only if it is actually a bug. I came here to ask if anyone knew anything about this issue in particular. | 02:34 |
| qman__ | Feadin, does it actually depend on them, or just recommend them? | 02:35 |
| Feadin | I mean, adding 144MB of multimedia dependencies for a metapackage on a bew server installation is not cool | 02:35 |
| twb | Feadin: if it's not a bug, the maintainer will just close the ticket with "not a bug" | 02:35 |
| ntoombs_ | where would i find dhcp reservations? | 02:36 |
| Feadin | It depends on them | 02:36 |
| twb | Feadin: IIRC language-pack-en shouldn't be installed, only -base | 02:36 |
| twb | Feadin: IIRC the preseeds on optical media installers are supposed to be set up to behave this way | 02:36 |
| ntoombs_ | qman__: would it work if i simply turned off dhcp? | 02:36 |
| qman__ | ntoombs_, that's a good question, it would have to be something like "reservation" or "static IP" or some other "computer settings" | 02:37 |
| ntoombs_ | like this forum post says how to do. http://www.dslreports.com/forum/r22286649-How-do-I-turn-off-DHCP-on-2Wire-2701HGG | 02:37 |
| Feadin | I installed these servers a couple of weeks ago selectiong only "OpenSSH" and leaving everything else on default, after that just added mysql-server and that's about it | 02:38 |
| qman__ | that might work, but be prepared to reset it if it doesn't | 02:38 |
| twb | Feadin: yes, I get it too, but I don't get a preseed file | 02:39 |
| qman__ | was the router provided by your ISP? | 02:39 |
| ntoombs_ | yes | 02:39 |
| twb | Feadin: I just manually uninstall those packages | 02:39 |
| qman__ | sometimes they hack them up | 02:39 |
| ntoombs_ | upon a $100 fee | 02:39 |
| qman__ | breaking basic functionality like this | 02:39 |
| qman__ | I need to go for a few minutes, I'll be back soon | 02:40 |
| ntoombs_ | ok | 02:40 |
| Feadin | Anyone has a default 11.04 server installation to try and do an "apt-get update && apt-get -s upgrade"? | 02:40 |
| Feadin | just want to be sure this is actually a bug and not my mistake | 02:41 |
| Gaming4JC | Hey guys, I have a VPS running and I did an update of MySQL. I bumped into this error now everytime the system starts I cannot get MySQL - ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2) | 02:59 |
| Gaming4JC | if I type... sudo /etc/init.d/mysql start | 02:59 |
| Gaming4JC | I can get it running again. But that's not much of an option since I need this to be stable and working 24/7 | 02:59 |
| Gaming4JC | I tried purge/reinstall of mysql, error persists. | 03:00 |
| Gaming4JC | ideas? :( | 03:00 |
| Jasonn | Gaming4JC: I can get you a vds real quick if you are looking for that | 03:00 |
| Gaming4JC | Jasonn: hmm? | 03:01 |
| Jasonn | check PM | 03:01 |
| === airtonix_ is now known as airtonix | ||
| === Jasonn is now known as JasonnAWAY | ||
| ntoombs | qman__: you there? | 04:51 |
| josePhoenix | Hi all | 06:14 |
| josePhoenix | I seem to have broken my server :D | 06:14 |
| josePhoenix | This is a great development in my evening | 06:14 |
| josePhoenix | Can I display the grub config from a grub prompt? | 06:15 |
| josePhoenix | I just ran a normal dist upgrade and it rebooted and then I got "error no such device" | 06:17 |
| lifeless | twb: is that template you hav esomething you're going to put fforward as a patch ? | 06:22 |
| lifeless | twb: it seems to be structured rather differnetly :) | 06:22 |
| josePhoenix | But really. What's my first thing to try when I'm at a grub prompt? | 06:24 |
| twb | enter? | 06:27 |
| lifeless | twb: is that template you hav esomething you're going to put fforward as a patch ? | 06:28 |
| lifeless | twb: it seems to be structured rather differnetly :) | 06:28 |
| josePhoenix | twb: well, it does go to a new line.. doesn't do much else | 06:29 |
| twb | lifeless: it's in-house shite | 06:29 |
| twb | lifeless: I'm happy to publish it but it's not exactly merge-ready | 06:29 |
| lifeless | yeah | 06:30 |
| lifeless | I think I have a minimal (fugly) patch to do i386, I'm just testing now | 06:31 |
| josePhoenix | Is there any way to list drives or see the grub configuration from the grub> prompt? | 06:33 |
| josePhoenix | or am I better off using a livecd and trying from there | 06:33 |
| lifeless | for drives just tab complete | 06:33 |
| lifeless | (TAB | 06:33 |
| josePhoenix | Doesn't seem to show anything | 06:34 |
| lifeless | thats unusual :) | 06:34 |
| josePhoenix | the worst thing to hear :P | 06:34 |
| josePhoenix | Well the error message is something like "unknown device" but from googling I've seen this mentioned in conjunction with LVM and with unused floppy drives being enabled | 06:35 |
| twb | josePhoenix: depending on the bash, you will have tab completion | 06:36 |
| twb | josePhoenix: (hd<TAB>, (hd0)/<TAB> etc | 06:36 |
| twb | hd0 = sda, hd0,1 = sda1 | 06:36 |
| twb | Oh, except I think in grub2 the numbering is from 0 for partitions as well :-/ | 06:37 |
| josePhoenix | ay de mi.... | 06:37 |
| twb | Note the parens around (hd0,1) are important | 06:37 |
| twb | If this is all too hard, just get a live CD and fix it that way | 06:37 |
| josePhoenix | hmm partition 0,1, ext2; partition 0,5: unknown filesystem | 06:37 |
| josePhoenix | But if the kernel + initrd was installed to a /boot partition it shouldn't have trouble reading it | 06:38 |
| twb | the latter might be swap | 06:38 |
| twb | Certainly it SHOULDN'T have trouble :-( | 06:38 |
| josePhoenix | The only thing that didn't go perfectly smoothly in this upgrade was that at one point the screen went blank and I hit enter to get it back without thinking | 06:40 |
| josePhoenix | and it chose N on the 'reboot now?' prompt and I had to start over | 06:40 |
| josePhoenix | but I thought ubuntu was too polished for that to fuck things up... | 06:40 |
| lifeless | so if its just got an old config | 06:44 |
| lifeless | subtracting 1 from the entries it was trying to boot with | 06:44 |
| lifeless | should make it all happy | 06:44 |
| josePhoenix | Could you elaborate on what that means? | 06:46 |
| lifeless | well | 06:46 |
| lifeless | your config should be present by default | 06:47 |
| lifeless | you can edit it in the grub prompt, or manually issue the commans | 06:47 |
| lifeless | https://help.ubuntu.com/community/Grub2#Command Line and Rescue Mode | 06:48 |
| lifeless | explore the fs's using tab complete to find the right partitions and then boot by hand | 06:48 |
| lifeless | if the stage2 loader couldn't be found thats more of a problem | 06:49 |
| lifeless | how far through is it getting ? | 06:49 |
| josePhoenix | It drops me at a grub> prompt after saying "unknown device" | 06:50 |
| josePhoenix | though that message only flashes on screen for an instant | 06:50 |
| josePhoenix | hmm does the grub prompt include a pager? I can't see the beginning of this config file -.- | 06:50 |
| lifeless | hah | 06:53 |
| lifeless | root@lucid-test-lp:~# arch | 06:53 |
| lifeless | x86_64 | 06:53 |
| lifeless | root@lucid-test-lp:~# dpkg --print-architecture | 06:53 |
| lifeless | i386 | 06:53 |
| josePhoenix | Man, I used to know all of this back when I used Gentoo. | 06:56 |
| airtonix | is it possible to provide an option in ~/.ssh/config to limit-rate (the scp has an argument -l that takes Kbits/s) | 06:56 |
| josePhoenix | Anyone know if ubuntu wants initargs? | 07:12 |
| josePhoenix | er | 07:12 |
| josePhoenix | init= bootarg? | 07:12 |
| josePhoenix | Is there a pager in the initrd shell? | 07:14 |
| twb | init=foo is not necessary by default | 07:14 |
| josePhoenix | Or any way I can see the top half of this file that scrolls offscreen? | 07:14 |
| twb | The busybox program provides a bunch of applets; which ones depends on how the team are feeling that week. | 07:15 |
| twb | Debian's busybox includes more (or less?) nowadays, but I think Ubuntu's is a 100kB smaller and containers less stuff :-( | 07:15 |
| josePhoenix | :[ | 07:15 |
| josePhoenix | more doesn't work and less is missing | 07:15 |
| twb | josePhoenix: if all else fails, you have to write to a file then use head and tail to get the page you want | 07:16 |
| twb | josePhoenix: I've seen that problem with more not working before, I think it's because the emergency environment's terminal capabilities are buggered – I never solved it myself | 07:16 |
| twb | re head/tail – or grep, sed, etc., of course | 07:16 |
| josePhoenix | okay, here is what I have figured out so far... I have a /boot at (hd0,1) that contains no vmlinuz/initrd.img but it does have vmlinuz-2.6.38-8-server + corresponding initrd | 07:18 |
| twb | The symlinks are in root, not boot, by default | 07:18 |
| josePhoenix | ah good to know | 07:19 |
| josePhoenix | there is also a partition at (hd0,5) which grub says is an 'unknown filesystem' but must be an LVM group because I know there's supposed to be a swap partition on here | 07:19 |
| josePhoenix | and there are only two primary partitions | 07:19 |
| josePhoenix | I think the problem is that I need to guess root= based on the startup output from the initrd stage | 07:19 |
| josePhoenix | or not | 07:21 |
| josePhoenix | I'm not sure how grub and kernel lines and LVM work | 07:21 |
| josePhoenix | well that's interesting. | 07:25 |
| josePhoenix | YAY IT BOOTED | 07:31 |
| keyboardtalk | I need some help enabling php5 on apache2. php code is not being parsed by the server | 07:37 |
| SpamapS | keyboardtalk: did you install libapache2-mod-php5 ? | 07:37 |
| keyboardtalk | yes | 07:37 |
| SpamapS | keyboardtalk: did you then run 'sudo a2enmod php5' ? | 07:42 |
| twb | SpamapS: doesn't it do that implicitly in the postisnt | 07:43 |
| slyboots_ | twb: Ah it should | 07:44 |
| keyboardtalk | SpamapS: yes | 07:44 |
| SpamapS | twb: not sure | 07:46 |
| SpamapS | would make sense | 07:46 |
| SpamapS | keyboardtalk: at this point you probably need to add a config file that adds the PHP handler | 07:49 |
| SpamapS | Would have hoped that happened automatically too but I guess not | 07:49 |
| twb | It *is* automatic provided he follows the conventions that the automatic scripts expect | 07:50 |
| twb | e.g. the mod_fcgi package sets up fast CGI only for *.fcgi files | 07:50 |
| SpamapS | been a while since I dropped it on a bare machine and tried it out. :-P | 07:51 |
| keyboardtalk | SpamapS: okay php seems to be working in /var/www, but not in other directories, namely /home/<user>/public_html/ | 07:54 |
| keyboardtalk | do php directories & files need certain permissions to work? | 07:55 |
| jmarsden | keyboardtalk: Read /etc/apache2/mods-enabled/php5.conf and note the deliberate turning off of PHP in user directories there, as a security precaution... | 07:58 |
| jmarsden | keyboardtalk: You can edit it if you want PHP in userdirs. | 07:58 |
| twb | I have a host running mod-php5, but I've locked it down as much as I could so it won't reflect the defaults | 07:58 |
| jmarsden | Or I think you could override it in .htaccess on a per-user basis | 07:59 |
| twb | jmarsden: assuming you leave .htaccess turned on :-) | 07:59 |
| jmarsden | Yes :) | 07:59 |
| SpamapS | keyboardtalk: if you want to run per-user php.. you'll want to look into something to keep them from attacking eachother. :) | 07:59 |
| luite | mod-php5 isn't terribly suitable for servers with multiple users anyway | 07:59 |
| * twb waves the "static HTML or GTFO" banner | 08:00 | |
| SpamapS | chrooted fastcgi .. lxc containers.. whole vms.. just don't turn on mod_php for user dirs | 08:00 |
| keyboardtalk | Okay, thank you very much | 08:00 |
| twb | I hate fastcgi | 08:00 |
| twb | I'd rather just have an app server daemon speak HTTP to the reverse proxy "real" web server | 08:00 |
| luite | hmm, fastcgi has been working fine for me (for php), but it does eat a lot of memory | 08:01 |
| greppy | I've been liking fcgi and/or suphp for a while :) | 08:01 |
| luite | hehe I can't really say that I like it, but it works (mod-fcgid, suexec and fastcgi) | 08:02 |
| twb | luite: are you running an httpd on the same host, or are you running fastcgi over a trusted network? | 08:02 |
| luite | same host | 08:02 |
| twb | Yeah see I do that but I don't like it | 08:03 |
| twb | <user> – <varnish> – <apache> – <fcgi app> | 08:03 |
| twb | I would prefer <user> – <varnish> – <http app> | 08:03 |
| keyboardtalk | Out of curiosity, what is the risk with enabling php for all users? | 08:03 |
| twb | keyboardtalk: that they'll use php | 08:03 |
| luite | keyboardtalk: all php scripts run under the same user id, so users can read each others php files | 08:04 |
| luite | keyboardtalk: (that's for mod-php for all users) | 08:04 |
| keyboardtalk | okay, good to know | 08:05 |
| SpamapS | twb: you can do that, just run your http app w/ mod_php ..why not? | 08:05 |
| twb | SpamapS: my app is written in C, not PHP | 08:05 |
| twb | At least this time it is | 08:05 |
| SpamapS | twb: In that case, libevent FTW | 08:05 |
| twb | SpamapS: how would that work? I am talking to upstream about it this week | 08:06 |
| SpamapS | full http server code built in.. :) | 08:06 |
| twb | I've only ever seen libevent used in rxvt | 08:06 |
| luite | twb: I like it that apache tries to automatically starts the fastcgi apps, fewer things to watch :) , but I use apache reverse proxy for my non-php web apps | 08:06 |
| luite | (written in haskell) | 08:06 |
| SpamapS | twb: recent versions of libevent have evhttp.h ... | 08:07 |
| SpamapS | http://monkey.org/~provos/libevent/doxygen-1.4.3/ | 08:07 |
| twb | luite: yeah, I am actually using apache not varnish, but that devalues my argument a little :-) | 08:07 |
| twb | SpamapS: neat | 08:07 |
| SpamapS | twb: you just register a callback per URI, and a default callback for dynamic URI's | 08:07 |
| twb | SpamapS: can I quote you (i.e. mail the scrollback to their list)? | 08:08 |
| twb | It'll save me paraphrasing | 08:08 |
| SpamapS | twb: sure, I'm just quoting from the docs | 08:08 |
| SpamapS | I played around with it a few months ago.. VERY easy to write an HTTP server | 08:09 |
| twb | I suppose it would be ironic if he switched to libevent and I couldn't get a new enough version into lucid | 08:13 |
| SpamapS | lucid is new enough | 08:14 |
| twb | Cool | 08:14 |
| * SpamapS shudders, recalling using CentOS 5's 4 year old libevent | 08:14 | |
| SpamapS | which is now.. 7 years old I think | 08:14 |
| twb | That's odd... urxvt appears to not use libevent in sid, but I thought it did | 08:15 |
| SpamapS | weird that a gui terminal would use it | 08:16 |
| SpamapS | I guess it makes sense.. but seems a bit heavy | 08:16 |
| twb | Ah, here http://software.schmorp.de/pkg/libev.html | 08:18 |
| twb | "A full-featured and high-performance (see benchmark) event loop that is loosely modelled after libevent, but without its limitations and bugs." | 08:18 |
| twb | Nice and objective there schmorp :-) | 08:18 |
| luite | the same lib that's used by the framework that I'm using :) ( http://snapframework.com/ ) | 08:21 |
| twb | luite: which, ev or event? | 08:21 |
| luite | ev | 08:21 |
| alex_21 | Hi all, | 08:21 |
| alex_21 | I am tring to clean everything off my server. | 08:22 |
| alex_21 | I want to return to a minimal installation. Is there any command to do this? | 08:22 |
| twb | alex_21: markauto most/all packages that are not Section: metapackages | 08:23 |
| twb | This is not a perfect rollback, tho | 08:23 |
| alex_21 | I want to completely reinstall everything, config and all without reinstalling Ubunut. | 08:24 |
| twb | You can't have that | 08:24 |
| twb | You can make it re-ask you most configuration options with "dpkg-reconfigure --all", but this is not the same. | 08:24 |
| twb | Nor it "aptitude reinstall ?installed" | 08:25 |
| twb | Feel free to try them anyway, once you have working backups. I think it'll be easier to do a full reinstall. | 08:25 |
| SpamapS | alex_21: for the future, this is why configuration management helps.. :) | 08:28 |
| === smb` is now known as smb | ||
| alex_21 | How can I reconfigure ldap. | 08:29 |
| alex_21 | ? | 08:29 |
| twb | Oh ldap | 08:40 |
| twb | you are totally fucked | 08:40 |
| twb | ldap HATES changes | 08:40 |
| twb | Especially if you're still using FDS or that other one. OpenLDAP is a little better | 08:40 |
| alex_21 | Allright guys. Thanks for the help. Reinstalling it is. | 08:51 |
| alex_21 | Good night. | 08:51 |
| nickmoeck | Is there a reputable PPA with mysql 5.5 packages for lucid? | 08:58 |
| soren | nickmoeck: I *seriously* doubt it. It's not even in Debian experimental or Ubuntu Oneiric yet. | 09:06 |
| twb | Packaging mysql isn't exactly trivial | 09:07 |
| nickmoeck | You're kidding, right? MySQL 5.5 was released in December! How are there no packages for it yet? | 09:11 |
| nickmoeck | surely it doesn't take more than 6 months to package | 09:11 |
| twb | nickmoeck: well, are you helping? | 09:14 |
| nickmoeck | twb: obviously not, I don't really know much about packaging. But I would think for something as important as mysql, packages for the latest version would be made within 6 months of the release | 09:16 |
| twb | Well, *I'm* not helping because the sooner mysql dies the sooner people will switch to sqlite and postgres, which are IMO infinitely better suited to toy and production usage respectively | 09:18 |
| twb | You can make a shit package in an afternoon, but getting it release-quality is difficult, especially since, as you say, so much gank relies on mysql and thus so many people will complain if you screw it up. | 09:19 |
| twb | Or it could just be that the guy responsible for packaging mysql for Ubuntu has gotten a job or a girlfriend in the last six months... | 09:19 |
| lifeless | twb: I've put up a patch for doing i386 lxc's | 09:19 |
| twb | lifeless: hmm? | 09:20 |
| lifeless | twb: would be interesting to know if it would suit your needs | 09:20 |
| lifeless | twb: bug 798476 | 09:20 |
| uvirtbot | Launchpad bug 798476 in lxc "it would be really useful to be able to make i386 lxc containers on a amd64 machine" [Low,Confirmed] https://launchpad.net/bugs/798476 | 09:20 |
| larsemil | old computer running md0 with lvm crashed. no i want to mount it on new hardware. It can find all volumegroups and logical volumes but it does not put them to /dev so that i can mount it. How do i import it? i am afraid vgimport does not work as it was not exported before old computer died. | 09:21 |
| twb | larsemil: lvm2 vgscan -ay? | 09:22 |
| twb | larsemil: something like that; last time I did it I RTFS'd the ramdisk to work out the details | 09:22 |
| larsemil | it finds it allright, but it does not make it mountable... | 09:24 |
| larsemil | twb: vgchange -ay was the solution | 09:25 |
| twb | good-o | 09:25 |
| larsemil | thanks for pointing me in the -ay direction! | 09:26 |
| twb | -ay means "bloody well do it" in LVM | 09:26 |
| phretor | any idea of why this error is occurring? http://paste.pocoo.org/show/416826/ "Exception during pm.DoInstall(): E:Internal Error, Could not perform immediate configuration (2) on util-linux" or how to solve it? | 09:56 |
| sarit | !paste | sarit | 11:05 |
| ubottu | sarit, please see my private message | 11:05 |
| uvirtbot | New bug: #800597 in postfix (main) "Latest postfix update breaks hash-based aliases.db (potentially other .db files too?)" [Undecided,New] https://launchpad.net/bugs/800597 | 11:22 |
| m3asmi | (98)Address already in use: make_sock: could not bind to address 0.0.0.0:80 !! | 11:37 |
| cloakable | m3asmi: There's something listening on port 80. Are you running a httpd already? | 12:01 |
| m3asmi | cloakble : I fix that problem | 12:02 |
| m3asmi | I ask about the : The requested URL /www was not found on this server. | 12:03 |
| uvirtbot | New bug: #800340 in ipxe (universe) "[MIR] ipxe" [Undecided,New] https://launchpad.net/bugs/800340 | 12:20 |
| uvirtbot | New bug: #800656 in etckeeper (main) "Update Manager can not proceed if there is a lock on /etc" [Undecided,New] https://launchpad.net/bugs/800656 | 13:37 |
| sommer | morning all | 13:54 |
| === med_out is now known as medberry | ||
| === oubiwann` is now known as oubiwann | ||
| === Ursinha is now known as Ursinha-lunch | ||
| uvirtbot | New bug: #800543 in dbconfig-common (main) "Installing zabbix-frontend-php fails in noninteractive mode" [Undecided,New] https://launchpad.net/bugs/800543 | 15:13 |
| hallyn | oh man, ppa build started yesterday morning still hasn't started | 15:23 |
| === jjohansen is now known as jj-afk | ||
| ScottK | Heya sommer. | 15:30 |
| ^^rcaskey | I've got an old server i'd like to get onoboard with automatic security updates. Just dist-upgraded to lucid, is there just a package i need to add or reconfigure? | 15:34 |
| zul | hallyn: around? | 15:42 |
| hallyn | zul: yeah, what's up? | 15:43 |
| zul | hallyn: we said we can drop the 9022-allows-lxc-containers-with-lxcguest.patch right? | 15:43 |
| hallyn | zul: are you doing 0.9.2? | 15:44 |
| zul | hallyn: yeah if we dont it needs to be rediffed | 15:45 |
| hallyn | you're pulling from sid, or upstream? | 15:45 |
| zul | sid | 15:45 |
| hallyn | zul: fetching. But in any case, so long as it has the patch from upstream adding the LXC_UUID, then yes | 15:46 |
| zul | hallyn: ok ill drop it then | 15:46 |
| hallyn | though, hm | 15:46 |
| hallyn | zul: we'll have to make sure then that the lxcguest package for natty has the LIBVIRT_LXC_UUID fix | 15:46 |
| zul | you mean for oneiric | 15:47 |
| hallyn | zul: no | 15:48 |
| hallyn | zul: lxcguest runs in the guest, so natty guests on oneiric hosts will need to do the right thing | 15:48 |
| zul | hallyn: oh right | 15:49 |
| hallyn | zul: can you take care of SRU'ing that? (i'ts not in natty, just checked) | 15:49 |
| zul | hallyn: yeah i can probably do that | 15:49 |
| hallyn | zul: cool, thanks. Should just be able to cp debian/*.upstart from the oneiric version | 15:49 |
| hallyn | i'm getting my ass kicked by the lxc template stuff (trying to consolidate the templates into one) | 15:49 |
| zul | hallyn: heh sucks to be you when is a new version of lxc suppose to be out | 15:51 |
| hallyn | whenever dlezcano feels there's a new feature | 15:51 |
| hallyn | zul: heh, found my biggest problem - missing ':' after option in getopt. doh | 15:59 |
| hallyn | new template is gonna rock | 15:59 |
| zul | hallyn: doh! | 15:59 |
| RoAkSoAx | Daviey: when are we looking into having cobbler in main? | 16:01 |
| RoAkSoAx | kirkland: ^^ | 16:02 |
| Daviey | RoAkSoAx: We'll check on it start of next week i think | 16:02 |
| Monotoko | hey guys...I'm following this tutorial: https://help.ubuntu.com/8.04/serverguide/C/postfix.html and when I try to connect I get kicked out by this: http://pastebin.com/WQ3e99Dq, what should be at "private/auth-client"? | 16:02 |
| kirkland | RoAkSoAx: personally, i have no strong need for cobbler to be in main | 16:03 |
| kirkland | RoAkSoAx: but i think Daviey and your team will probably need it so | 16:03 |
| RoAkSoAx | Daviey: ok cool! | 16:04 |
| RoAkSoAx | kirkland: thought orchestra was gonna have to be in Main by this cycle | 16:04 |
| kirkland | RoAkSoAx: that's up to you guys | 16:05 |
| Monotoko | :( | 16:05 |
| RoAkSoAx | kirkland: ok ;) | 16:05 |
| === koolhead11|afk is now known as koolhead11 | ||
| lynxman | ping Daviey | 16:22 |
| uvirtbot | New bug: #800744 in samba (main) "package samba 2:3.5.8~dfsg-1ubuntu2.2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/800744 | 16:22 |
| w00 | Isn't that cute.. install munin wants to also put x11-common on my system ;/ | 16:23 |
| Doonz | has anyone set up openvpn server in ubuntu 10.04 | 16:24 |
| Riz | Doonz: yup | 16:26 |
| SpamapS | lynxman: generally nick: ping is more reliable.. some people (like me) don't have their name highlighted in all instances.. just when its the first word | 16:32 |
| SpamapS | Doonz: https://help.ubuntu.com/10.04/serverguide/C/openvpn.html | 16:32 |
| lynxman | SpamapS: will have that in mind, he normally replies though :) | 16:33 |
| Doonz | 1 sec damn work is getting in the way | 16:36 |
| medberry | SpamapS, ah, I've never heard of that setting/behavior. Probably comes in handy if you have a ubiquitous nick. | 16:42 |
| SpamapS | medberry: its the default in irssi | 16:42 |
| pmatulis | can one install a kvm guest running the devel release with vmbuilder? | 16:47 |
| Doonz | SpamapS: ive tried that guide a few times and it didnt work | 16:57 |
| Doonz | https://help.ubuntu.com/10.04/serverguide/C/network-configuration.html#bridging Ive tried following that guide but it wont bring up the bridged connection | 16:57 |
| ScottK | Doonz: Please file a bug with details against the ubuntu-docs project. | 16:58 |
| Doonz | its not a bug. just a stupid user trying to do something he doesnt know anything about | 16:58 |
| Doonz | :( | 16:58 |
| Doonz | https://help.ubuntu.com/10.04/serverguide/C/openvpn.html <-- im going to retry this guide and ill pastebin the errors | 16:59 |
| SpamapS | pmatulis: should be doable | 17:04 |
| pmatulis | SpamapS: do you know how? i mean, is it done with hacking? | 17:05 |
| pmatulis | SpamapS: b/c suite=oneiric doesn't work if host is natty | 17:05 |
| Daviey | lynxman: sorry, was otp | 17:06 |
| SpamapS | pmatulis: not sure. I just use iso's + virt-manager. :-P | 17:06 |
| pmatulis | SpamapS: ok | 17:06 |
| lynxman | Daviey: no problem | 17:06 |
| Doonz | well its official im an idiot | 17:14 |
| Doonz | so anyone wanna set up openvpn server for me | 17:15 |
| === jj-afk is now known as jjohansen | ||
| === koolhead11 is now known as koolhead11| | ||
| Doonz | *sigh* | 17:43 |
| Doonz | heres what i get after i follown the openvpn guide http://pastebin.com/RUq5Mtup | 17:44 |
| maxo | I'm using Ubuntu Enterprise Cloud (Eucalyptus). If I terminate an instance, any configuration made is lost. How can one get round this problem? | 18:17 |
| lynxman | RoAkSoAx: ping | 18:19 |
| RoAkSoAx | lynxman: pong | 18:22 |
| fowlduck | We have an Intel Xeon E5645 and we're not seeing the 2 threads per core in lscpu, cat /proc/cpuinfo, etc, despite the processor being capable of hyperthreading. Is there a way to determine what features are supported by the kernel? I'd like to check if hyperthreading is enabled | 18:22 |
| fowlduck | btw, `uname -a` # => Linux 358015-domain 2.6.38-8-server #42-Ubuntu SMP Mon Apr 11 03:49:04 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux | 18:22 |
| fowlduck | we're using a bare-metal host and we're trying to prove that the issue isn't with the OS (or fixing it if it is), so I can't exactly check the BIOS or anything | 18:25 |
| amero | i have a question about socks proxy server. for a socks server to be able to accept and forward incoming connections back to clients behind the proxy, what does it need to have? | 18:34 |
| SpamapS | fowlduck: if the CPU can do it, it should be in the flags in /proc/cpuinfo | 18:43 |
| SpamapS | fowlduck: it may still be disabled in the BIOS | 18:43 |
| SpamapS | fowlduck: I'd say 90% of workloads see no real benefit to HT | 18:43 |
| fowlduck | SpamapS: thanks for getting back to me, I got some help from the guys in #ubuntu-kernel on the problem | 18:44 |
| fowlduck | You're right, too. It seems to be disabled in the BIOS | 18:45 |
| fowlduck | despite the minimal performance improvements, we'd like to get the potential benefits if its available | 18:45 |
| === Ursinha-lunch is now known as Ursinha | ||
| RoAkSoAx | SpamapS: should postinst/postrm scripts call "service XYW restart" or similar? or should that be done with invoke.rc-d | 19:13 |
| RoAkSoAx | invoke-rc.d* | 19:13 |
| SpamapS | invoke-rc.d is the *only* way allowed | 19:13 |
| RoAkSoAx | SpamapS: that's what I thought! Thanks | 19:14 |
| rump | does rsyslogd miss anything besides apache on a default install? | 19:17 |
| pmatulis | i didn't know apache was part of the default install | 19:24 |
| RoyK | pmatulis: it isn't | 19:31 |
| pmatulis | RoyK: rump is mixed up | 19:32 |
| uvirtbot | New bug: #800824 in cloud-init (main) "cloud-init-nonet times out in lxc" [High,Confirmed] https://launchpad.net/bugs/800824 | 19:36 |
| vooze | i just switched from debian til ubuntu-server (or trying to) but for some reason, the ubuntu 11.04 server CD cant see my main harddrive.. debian can just fine.. any ideas? | 19:51 |
| uvirtbot | New bug: #800833 in amavisd-new (main) "package amavisd-new-postfix 1:2.6.5-0ubuntu2 failed to install/upgrade: il sottoprocesso vecchio script di post-installation ha restituito lo stato di errore 1" [Undecided,New] https://launchpad.net/bugs/800833 | 19:56 |
| RoyK | vooze: sound strange - but just a thought - if you plan to setup a server, why don't you use an LTS release such as 10.04? | 19:57 |
| RoyK | 11.04 will only be supported for 18 months | 19:57 |
| RoyK | LTS server releases are supported for a full 5 years from release | 19:57 |
| Alfafa | Hi I haven't followed this channel. But you are aware that a aptitude full-upgrade today will pull in a lot of irrelevant packages? like firefox,synaptic etc..gui programs? | 19:58 |
| Alfafa | I also found a forum post from one experiencing the same http://ubuntuforums.org/showthread.php?t=1788159 | 19:59 |
| RoyK | Alfafa: sometimes dependencies change - but really - if you don't have any graphical stuff installed, an apt-get dist-upgrade (I guess that's like aptitude full-upgrade) really shouldn't add much new | 19:59 |
| RoyK | Alfafa: which version? | 19:59 |
| Alfafa | It is 11.04 | 19:59 |
| RoyK | 11.04 has a few dependency issues - I know | 20:00 |
| Alfafa | apt-get dist-upgrade does the same..pulling in gui stuff | 20:00 |
| * RoyK sticks to LTS releases for servers | 20:00 | |
| Alfafa | Ok..I would just make sure someone knew :-) | 20:00 |
| vooze | RoyK: yeah, trying that now.. burning atm. | 20:00 |
| ScottK | Alfafa: No. No it doesn't. You added something. | 20:00 |
| RoyK | Alfafa: report a bug | 20:00 |
| ScottK | I've upgraded servers from 10.10 to 11.04 just fine. | 20:01 |
| RoyK | ScottK: the dependencies in 11.04 are rather patchy | 20:01 |
| * w00 same | 20:01 | |
| ScottK | RoyK: Point to bugs please. | 20:01 |
| RoyK | I've just been testing on a test VM, and had lots of wierd issues | 20:01 |
| Alfafa | I didn't add something I ran a perfectly normal aptitude full-upgrade yesterday and today the same pull in a lot of gui packages not needed | 20:01 |
| ScottK | Alfafa: What does apt-get upgrade do? | 20:01 |
| RoyK | ScottK: so I just ditched the VM and let it go | 20:02 |
| ScottK | OK. | 20:02 |
| Alfafa | apt-get upgrade should work much better | 20:02 |
| Alfafa | but you wouldn't get kernel updates i believe by using that | 20:02 |
| RoyK | ScottK: upgrade to the latest version of packages within the distro version (say, 10.04.1), but it won't upgrade to 10.04.2 - dist-upgrade would | 20:02 |
| ScottK | Alfafa: OK. How about apt-get dist-upgrade? | 20:02 |
| ScottK | RoyK: No. That's wrong. | 20:03 |
| RoyK | sure? | 20:03 |
| ScottK | Yes. | 20:03 |
| ScottK | 10.04.2 is nothing more than all the updates released for 10.04 on a certain date. | 20:03 |
| RoyK | hm.. I thought it was a new subversion, slightly newer packages etc | 20:03 |
| ScottK | The difference between upgrade and dist-upgrade is that dist-upgrade will add/remove packages if needed and upgrade won't. | 20:03 |
| RoyK | never mind | 20:03 |
| ScottK | Generally in stable releases that just comes up with kernels that break ABI and need new package names. | 20:04 |
| RoyK | what's ABI? | 20:04 |
| ScottK | Application Binary Interface. | 20:05 |
| RoyK | k | 20:05 |
| ScottK | If the binary interface changes there are packages that need to be rebuilt against the newer ABI so they change the package name. | 20:05 |
| RoyK | I'm running 2.6.35 on my Lucid servers atm - haven't seen any issues there | 20:05 |
| ScottK | That means a kernel upgrade (if it breaks ABI) needs dist-upgrade. | 20:05 |
| RoyK | ah - IC | 20:05 |
| RoyK | thanks for the info | 20:05 |
| RoyK | I generally only use dist-upgrade, though | 20:06 |
| RoyK | guess that should be safe, then | 20:06 |
| Alfafa | I think it may be a new upgradedable package today which have a rather weird recommends. I can look at the packages it will upgrade to see which one of them pulls in the 57 new(mostly gui packages) packages | 20:06 |
| ScottK | Lucid was released with 2.6.32.21.22 and has 2.6.32.32.38 now. | 20:06 |
| ScottK | Every time the 4th number changes it's an ABI break and there's new binary packages. | 20:06 |
| RoyK | ScottK: there's a maverick backport of 2.6.35-25 in the repos | 20:07 |
| RoyK | ScottK: and that's recommended because of some bugs not fixable in 2.6.32 | 20:07 |
| RoyK | (last I checked) | 20:07 |
| ScottK | It's got a completely different set of package names. | 20:07 |
| RoyK | dist-upgrade won't pull it, though, you'll have to apt-get install it | 20:07 |
| ScottK | I think it's recommended only if you're having issues with the normal one. | 20:07 |
| ScottK | My LTS servers are running the regular 2.6.32 with no problems. YMMV. | 20:08 |
| RoyK | well, I just updated my boxes - all 20-odd servers are running 2.6.35 now without issues | 20:08 |
| Alfafa | ok..i think the bug is the recommends for language-pack-en-base which recommends firefox-locale-en which then depends on firefox maybe? but I am not sure..i believe aptitude full-upgrade sometimes install the recommended packages which apt-get upgrade doesn't? | 20:10 |
| ScottK | Perhaps. | 20:11 |
| ScottK | Let me look into it. | 20:11 |
| RoyK | vooze: any luck? | 20:11 |
| uvirtbot | New bug: #800845 in munin (main) "plugin postfix_mailvolume invalid output when log file missing" [Undecided,New] https://launchpad.net/bugs/800845 | 20:12 |
| ScottK | Alfafa: I don't think that's part of the standard server install. | 20:13 |
| Alfafa | ScottK: Hmm..ok I will look into why it is there then | 20:15 |
| uvirtbot | New bug: #800856 in cloud-init (main) "resizefs module causes problems on LXC containers" [Undecided,New] https://launchpad.net/bugs/800856 | 20:21 |
| Alfafa | apt-get -s install apt-rdepends | 20:22 |
| Alfafa | sorry | 20:22 |
| === raubvogel1 is now known as raubvogel | ||
| lifeless | hallyn: SpamapS: thanks for your aid - this is what I have now - https://dev.launchpad.net/Running/LXC#preview | 20:26 |
| Alfafa | ScottK: Ok. Couldn't find any reasonable rdepends...so removed the language-pack* maybe it has been put in some time earlier (it has been upgraded to 11.04 and not reinstalled) | 20:29 |
| Alfafa | ScottK: Sorry for wasting your time | 20:30 |
| ScottK | No problem. | 20:30 |
| ScottK | Sometimes this is how you find out. | 20:30 |
| Alfafa | ScottK: I will just answer the other ones forum post..then somebody else can find it ;-) | 20:30 |
| RoyK | anyone around that knows a good L2 network discovery tool? | 20:42 |
| === lullabud is now known as warzauwynn | ||
| lifeless | SpamapS: hallyn: next fail point - fuse fails to install in the lxc | 21:11 |
| Daviey | lifeless: wow, that is a suprise.. fails to dpkg unpack/configure? | 21:13 |
| lifeless | Setting up fuse-utils (2.8.1-1.1ubuntu2) ... | 21:13 |
| lifeless | creating fuse group... | 21:13 |
| lifeless | mknod: `fuse-': Operation not permitted | 21:13 |
| lifeless | makedev fuse c 10 229 root root 0660: failed | 21:13 |
| lifeless | chgrp: cannot access `/dev/fuse': No such file or directory | 21:13 |
| SpamapS | lifeless: yeah you have to mknod outside of lxc IIRC | 21:14 |
| lifeless | SpamapS: where should I file the bug ? | 21:14 |
| SpamapS | hallyn: ^^ ? | 21:15 |
| lifeless | devpts fs doesn't help | 21:15 |
| lifeless | http://jtrancas.wordpress.com/2011/02/09/fuse-filesystems-lxc-container/ perhaps | 21:17 |
| Daviey | lifeless: looks like an lxc issue then... with the template. | 21:19 |
| lifeless | adding that stuff to the config seems to be enough | 21:21 |
| lifeless | bug 800886 | 21:23 |
| uvirtbot | Launchpad bug 800886 in lxc "lucid fuse-utils fails to install in lxc container" [Undecided,New] https://launchpad.net/bugs/800886 | 21:23 |
| RoyK | bug 1 | 21:23 |
| Daviey | timeout \o/ | 21:23 |
| RoyK | hm... | 21:23 |
| uvirtbot | Launchpad bug 1 in ubuntu "Microsoft has a majority market share" [Critical,In progress] https://launchpad.net/bugs/1 | 21:23 |
| Daviey | i'm wrong! | 21:23 |
| Daviey | thanks lifeless | 21:24 |
| RoyK | this pool is FAST http://paste.ubuntu.com/630972/ :D | 21:25 |
| hallyn | lifeless: SpamapS: no bug. edit the container's config (/var/lib/lxc/<container>/config) and add a devices whitelist eception for /dev/fuse | 21:26 |
| Daviey | hallyn: can we not ship that? | 21:27 |
| hallyn | Daviey: we shouldn't. | 21:27 |
| Daviey | hallyn: oh? | 21:27 |
| lifeless | hallyn: why not ? | 21:27 |
| hallyn | bc we pretend containers aren't 100% unsafe as is :) | 21:27 |
| lifeless | hallyn: could we ship an option | 21:27 |
| lifeless | hallyn: so folk don't go off of random internet sites like I just did. | 21:28 |
| hallyn | lifeless: right, just add the option to the config | 21:28 |
| hallyn | lifeless: no problem, the lp bug shoudl steer people :) | 21:28 |
| hallyn | where is that link you pasted before | 21:28 |
| lifeless | http://jtrancas.wordpress.com/2011/02/09/fuse-filesystems-lxc-container/ | 21:29 |
| hallyn | ah there it is | 21:29 |
| lifeless | seems to be for udev | 21:29 |
| lifeless | so perhaps not relevant to lucid | 21:29 |
| hallyn | lifeless: thx :) | 21:29 |
| Daviey | hallyn: I wonder if handling the fuse failure could be better handled in the fuse package giving reasoning for failure, and suggesting uncommenting it on the config? | 21:29 |
| Daviey | (ship it commented out?) | 21:29 |
| hallyn | Daviey: could be | 21:29 |
| lifeless | whats the security issue with this? | 21:30 |
| hallyn | presumably it allows the conatiner to load arbitrary fuse filesystems? and abuse any holes which might exist in the /dev/fuse api | 21:30 |
| hallyn | security issue is just 'a big blog tying into the kernel' | 21:31 |
| lifeless | but fs's are contained | 21:31 |
| hallyn | maybe i'm being unreasonably, i dunno | 21:31 |
| lifeless | if fuse is unsafe | 21:31 |
| hallyn | contained by what? | 21:31 |
| lifeless | I thought they were namespaced ? | 21:31 |
| hallyn | that's not the angle i'm worried about, | 21:32 |
| hallyn | it's that the kernel is calling out to functions in userspace, | 21:32 |
| hallyn | and the container is writing those functions | 21:32 |
| Daviey | I love the fact that fuse is a userspace implementation for filesystems... and yet, we are still blocked on the kernel module :) | 21:32 |
| lifeless | right... | 21:32 |
| lifeless | but that doesn't seem intrinsically insecure | 21:32 |
| hallyn | it's intrinsically fragile | 21:32 |
| lifeless | as long as the kernel doesn't trust the userspace implementation blindly | 21:32 |
| hallyn | lifeless: it's the same reason why we want to sandbox containers so that any fo the 100 newest syscalls can't be used :) | 21:33 |
| * lifeless would like to not need fuse, but is installing gnome inside the lxc container :( | 21:33 | |
| Daviey | lifeless: you sir, are a crazy man. | 21:33 |
| hallyn | lifeless: well again, containers aren't secure to begin with, | 21:33 |
| lifeless | Daviey: windmill tests | 21:33 |
| lifeless | Daviey: for LP | 21:33 |
| hallyn | lifeless: so i don't want to make it too hard. But is adding that entry to the config too much? | 21:33 |
| Daviey | lifeless: interesting! | 21:33 |
| lifeless | Daviey: https://dev.launchpad.net/Running/LXC | 21:34 |
| Daviey | hallyn: it seems a discoverability issue | 21:34 |
| lifeless | hallyn: well, the less it Just Works | 21:34 |
| hallyn | ypu | 21:34 |
| hallyn | yup | 21:34 |
| lifeless | hallyn: the harder it is to use. | 21:34 |
| lifeless | hallyn: so I accept that if there is an issue, we shouldn't do it. | 21:34 |
| hallyn | Daviey: so being a discoverability issue, i'd lean toward fiing it in fuse-utils | 21:34 |
| lifeless | hallyn: but I'd rather concrete 'we need to do X to fix it' objections rather than 'it might break'. | 21:34 |
| hallyn | lifeless: the issue will become more meaningful when we have user namespaces. I just don't want toopen all the doors in the meantime so that then user namespaces will be worthless | 21:35 |
| lifeless | hallyn: wearing my obstinate user hat rather firmly. | 21:35 |
| lifeless | hallyn: anyhow - that wiki page I just linked documents doing the home bindmount | 21:35 |
| Daviey | lifeless: So are you thinking of doing this to integrate with the tarmac pre land testing? | 21:35 |
| lifeless | hallyn: the one bit I'm missing is user mirroring | 21:35 |
| hallyn | lifeless: maybe we should as on #ubuntu-hardened how safe they feel with /dev/fuse API. I know little | 21:35 |
| lifeless | Daviey: we don't use tarmac :) - and no. | 21:35 |
| hallyn | lifeless: you can use the lxc-ubuntu template I linked to in the bug | 21:35 |
| lifeless | Daviey: doing it to have less memory footprint than the lucid VM I currently use. | 21:36 |
| hallyn | lifeless: let me post some debs actually to people.canonical.com, one sec | 21:36 |
| Daviey | lifeless: oh right | 21:36 |
| hallyn | lifeless: http://people.canonical.com/~serge/lxc_0.7.4-0ubuntu7.1clone3_amd64.deb | 21:37 |
| hallyn | lifeless: with that deb, you can do 'lxc-create -t natty -n natty -f /etc/lxc.conf -- -b lifeless | 21:38 |
| hallyn | and user lifeless will be bind-mounted and created in the container | 21:38 |
| lifeless | wicked | 21:38 |
| hallyn | it also supports '-a i386' | 21:38 |
| lifeless | nice | 21:38 |
| lifeless | that will be in oneiric? | 21:38 |
| hallyn | yeah, unless it hits resistence upstream (unlikely for this) | 21:38 |
| hallyn | lifeless: actually yo uneed to add '-r natty', as the default container is lucid | 21:39 |
| lifeless | hallyn: well, I want lucid :) | 21:40 |
| hallyn | lifeless: excellent :) | 21:40 |
| lifeless | hallyn: launchpad.net deploys on lucid | 21:40 |
| hallyn | lifeless: then until lp gets aroudn to compiling the ubuntu-virt ppa's lxc, you'll need to add '-x' to trim the container (as it can't install the lxcguest package) | 21:40 |
| lifeless | lxcguest ? | 21:41 |
| hallyn | oh yay, it built | 21:41 |
| Daviey | hallyn: Hmm, i don't think lifeless wants to deploy in lxc... just devel | 21:41 |
| hallyn | lifeless: yes, lxcguest works around some blotches where upstart can't deal with not beign on hardwrae | 21:41 |
| hallyn | Daviey: so? | 21:41 |
| lifeless | hallyn: so I should install that in my lucid guest ? | 21:41 |
| lifeless | $ apt-cache show lxcguest | 21:42 |
| lifeless | W: Unable to locate package lxcguest | 21:42 |
| lifeless | E: No packages found | 21:42 |
| hallyn | lifeless: only if you want it to be safe across udev upgrades | 21:42 |
| hallyn | right, it comes from ppa:ubuntu-virt/ppa. It's in natty and oneiric, but not lucid and maverick | 21:42 |
| lifeless | hallyn: is it needed for lucid? | 21:42 |
| Daviey | hallyn: Ahh.. i missunderstood "then until lp gets aroudn to compiling the ubuntu-virt ppa's lxc | 21:42 |
| hallyn | not if you create a trimmed container (which the lxc-lucid template did, and which the new template does if you add -x) | 21:42 |
| hallyn | Daviey: oh yeah, it just took over 24 hrs for it to get around to it | 21:43 |
| hallyn | lifeless: are these 'run and delete' contaienrs? | 21:43 |
| lifeless | hardware testing have solten all the builders | 21:43 |
| lifeless | hallyn: nope | 21:43 |
| hallyn | you're going to keep themaround and upgrade? | 21:43 |
| lifeless | hallyn: we install a GB of dependencies | 21:43 |
| lifeless | so i doubt folk will treat them as transient things | 21:43 |
| hallyn | then yeah, use lxcguest | 21:43 |
| Daviey | hallyn: You know, having someone on the launchpad team that wants to use a package from a PPA oddly increases it's build time :) | 21:44 |
| lifeless | let me know when its available for lucid? | 21:44 |
| hallyn | it does things like bind-mount an empty fiel over /lib/init/fstab, so that if mountall/udev gets updated it still boots | 21:44 |
| Daviey | (err, decreases) | 21:44 |
| hallyn | lifeless: it's available now in the ubuntu-virt ppa | 21:44 |
| hallyn | lifeless: it must have just compiled | 21:44 |
| hallyn | lifeless: I assume there is no way to add that package into the official lucid archive at this late date | 21:45 |
| lifeless | hallyn: sure you can, its not trivial but it is doable | 21:45 |
| lifeless | lxcguest wants to futz with /etc/init/console.conf | 21:45 |
| hallyn | lifeless: that file only exists bc the lxc-lucid template created it | 21:46 |
| lifeless | ok, so I | 21:46 |
| hallyn | so yes, the idea is the template doesn't change the container, you just run lxcguest in a plain VM, then you can boot it either in kvm or a container | 21:46 |
| hallyn | no changes | 21:46 |
| Daviey | hallyn: backports baby! | 21:47 |
| hallyn | Daviey: well, it's in ubuntu-virt ppa... | 21:47 |
| hallyn | i thought that was a reasonably official place to put it | 21:47 |
| lifeless | https://dev.launchpad.net/Running/LXC#preview now with lxcguest info | 21:47 |
| * Daviey watchs hallyn stir the beehive of PPA's vs Backports. | 21:47 | |
| hallyn | lifeless: I just want to get my lxc-lvmcreate script back itno my pacakge, then I will push it to ppa:serge-hallyn/lxc-natty and blog about how to use it to do the arch and home bind mounting | 21:49 |
| lifeless | hallyn: let me know as things improve and become accessible, I'll happily update this wiki page | 21:50 |
| hallyn | lifeless: great,thanks | 21:51 |
| hallyn | jdstrand: kees: does one of you have any input on the safety and sanity of the api used with /dev/fuse? | 21:51 |
| hallyn | (regarding bug 800886) | 21:51 |
| uvirtbot | Launchpad bug 800886 in lxc "lucid fuse-utils fails to install in lxc container" [Undecided,Confirmed] https://launchpad.net/bugs/800886 | 21:51 |
| hallyn | if you gusy tell me it's safe, I'll just add it to the whitelist int he template | 21:52 |
| * jdstrand does not otoh, but I am going to point mdeslaur at this too (who might have more up to date info) ^ | 21:52 | |
| hallyn | cool, thanks. yeah i just meant top of head. if research is required i'll go read the code. | 21:53 |
| kees | hallyn: my understanding is that the fuse perms are checked at the kernel level. | 21:54 |
| kees | hallyn: I haven't though much about how that might behave with a container, though. | 21:54 |
| kees | in _theory_, it should be fine, but I've never tested it | 21:54 |
| hallyn | kees: oh i wasn't even thinking about uid translations over namespaces. (that'll be my job :) | 21:55 |
| hallyn | kees: I'm wondering how... 'rich' the API over /dev/fuse is. Do you trust people who connect to it? | 21:55 |
| hallyn | kees: or does it have a small, ilmited API that I can sort of trust | 21:56 |
| hallyn | by 'do you trust people' i meant 'do you *have* to trust people' | 21:56 |
| kees | hallyn: well, it's designed for non-root users, but I haven't spent any time auditing it. | 22:02 |
| hallyn | kees: so it's world writeable usually? | 22:02 |
| hallyn | so it is | 22:03 |
| kees | hallyn: yes | 22:03 |
| hallyn | so why would i worry about it | 22:03 |
| hallyn | kees: thx :) | 22:03 |
| kees | heh, np | 22:03 |
| hallyn | of course, | 22:03 |
| hallyn | i assume th kernel code checks uids. but again that just means its up to me to clean that up when time comes | 22:03 |
| hallyn | neaty | 22:03 |
| hallyn | neato, even | 22:03 |
| lifeless | \o/ | 22:05 |
| hallyn | gah, i *still* haven't asked for lxc upload perms | 22:22 |
| hallyn | zul: can you help a brother out? :) | 22:22 |
| hallyn | Daviey: still there? | 22:24 |
| hallyn | Daviey: could you push http://people.canonical.com/~serge/lxc_0.7.4-0ubuntu11/lxc_0.7.4-0ubuntu11.dsc for me? | 22:26 |
| huats | hallyn, is it just uploading it ? | 22:29 |
| huats | or do you need anything special on it ? | 22:29 |
| huats | (well you mean to ubuntu just to be sure) | 22:29 |
| huats | (I am asking if the great Daviey is not around) | 22:30 |
| hallyn | huats: yeah, just uploading it | 22:32 |
| * SpamapS shouts "CHOO CHOOO" after riding the sl train | 23:30 | |
| === lullabud is now known as warzauwynn | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!