| zul | axisys: http://www.ubuntu.com/usn | 00:01 |
|---|---|---|
| axisys | zul: cool! thanks a lot | 00:02 |
| axisys | zul: wait! i was looking for a command to see if there is any security patch available | 00:03 |
| axisys | running locally | 00:03 |
| hallyn | kees: is there a libseccomp function for loading the output of seccomp_export_bpf(fd)? | 01:28 |
| === Aaton_off is now known as Aaton | ||
| === n0ts_off is now known as n0ts | ||
| === slide23 is now known as slide | ||
| === Aaton is now known as Aaton_off | ||
| stgraber | hallyn: clear_config_item seems to fail on lxc.cgroup.devices.allow here | 02:52 |
| stgraber | hallyn: and get_config_item is return lxc.network.0.ipv4 sorted somehow, so my tests fail as clearing + setting two of them + getting and comparing give them back in the reverse order :) | 02:53 |
| stgraber | (I worked around that one by having the python code consider them "sets", so by definition unsorted, fixing the comparisons) | 02:53 |
| stgraber | hallyn: pushed the python changes, will push a batch of PPA builds now and implement the network stuff in python tomorrow (easy handling of interfaces) | 03:01 |
| === cpg is now known as cpg|away | ||
| hallyn | stgraber: yes, you can only clear_config_item(lxc.cgroup). | 03:42 |
| hallyn | stgraber: as for ipv4 sorting, i can't imagine what would cause that! | 03:44 |
| hallyn | good night | 03:48 |
| === n0ts is now known as n0ts_off | ||
| === n0ts_off is now known as n0ts | ||
| redactd | hi, is there a way to set postfix so that server to server comms has optional or no tls but client to server enforces tls? my main.cf appears to only allow it to be on or off globally | 06:52 |
| === almaisan-away is now known as al-maisan | ||
| === n0ts is now known as n0ts_off | ||
| === n0ts_off is now known as n0ts | ||
| === cpg|away is now known as cpg | ||
| === Lcawte|Away is now known as Lcawte | ||
| jamespage | SpamapS, fastcgi upstream located - mailing list now active again.... | 07:08 |
| === al-maisan is now known as almaisan-away | ||
| wizonesolutions | If I'm setting up virtualization (libvirt + KVM) on a 12.04 Server install, and I'm using it to run a couple large VMs and basically split the box in two, is virt-install alone appropriate or should I try vmbuilder? | 07:41 |
| wizonesolutions | not sure if vmbuilder is aimed at people who are going to package up the VMs or what | 07:41 |
| wizonesolutions | there seem to be a few options for creating VMs | 07:41 |
| Degot | Hi. In FreeBSD I'm using unionfs to mount shared folder (/data/software/) into user's home folder (/home/user/software/). users are jailed in their homes via ftp. How to do the same in ubuntu server 12.04 ? | 07:58 |
| === zyga-afk is now known as zyga | ||
| demonspork | Degot: I don't know a ton about it, but I have encountered ftpd that can do the home directory jail in their settings | 08:02 |
| demonspork | Degot: but it has been a few years since I was digging deeply into setting up FTP stuff | 08:02 |
| Degot | demonspork, I can jail. It's not a problem. The problem is how to mount folder into the jail from the outside | 08:03 |
| Degot | demonspork, FYI - mount --bind /home/user/software /data/software | 08:22 |
| mardraum | wizonesolutions: virt-install gives you everything you need | 08:41 |
| RoyK | iirc someone in here was talking about zfs outside of Oracle's control was outdated etc etc - http://blog.vx.sk/archives/35-Novinky-v-open-source-ZFS.html ;) | 09:21 |
| === almaisan-away is now known as al-maisan | ||
| reisi | does sudo -u logcheck logcheck -to (test, output to stdout) work for anyone in ubuntu server 12.04 LTS? | 09:28 |
| Exopaladin | So glad they re-introduced aclmode | 09:28 |
| RoyK | Exopaladin: what's that? | 09:31 |
| Exopaladin | It basically enforces ACLs and stops dumb apps like rsync blowing away your ACLs | 09:31 |
| RoyK | heh | 09:33 |
| Degot | Hi, again. I have the Vbox VM with VRDE enabled. Need to stream video of the vm's desktop. Idea is : vm-rde -> rdp client -> x264 (zero-latency) -> stream server (ffmpeg?). Any ideas hot to implement this ? | 09:53 |
| jamespage | wizonesolutions, just use virt-install | 09:56 |
| jamespage | vmbuilder does not get much maintenance love these days... | 09:56 |
| === ninjak_ is now known as ninjak | ||
| RoyK | wizonesolutions: I use virt-manager - nice little gui | 10:02 |
| uvirtbot | New bug: #1026857 in bind9 "Installing bind9 as per HACKING.txt leaves a running bind daemon on developer's machines" [Undecided,New] https://launchpad.net/bugs/1026857 | 10:11 |
| === matsubara is now known as matsubara-afk | ||
| === al-maisan is now known as almaisan-away | ||
| === ninjak_ is now known as ninjak | ||
| zul | good morning | 12:11 |
| === cpg is now known as cpg|away | ||
| jamespage | morning zul | 12:17 |
| zul | hey jamespage | 12:18 |
| jamespage | zul: its Friday! | 12:18 |
| zul | jamespage: yeah i have a wicked headache going on right now :( | 12:19 |
| * jamespage stops shouting | 12:22 | |
| stgraber | hallyn: oh much trouble would it be for you to allow clearing these lxc.cgroup keys allowing multiple values? (cgroup.devices.allow/cgroup.devices.deny are the two that I know but there may be more) | 13:21 |
| stgraber | hallyn: having to clear the whole lxc.cgroup tree and re-adding everything would require quite a bit hack on the python side (querying all the keys, all the values, storing them, clearing, then setting them all back again) | 13:22 |
| hallyn | stgraber: the thing about devices.allow is that even set_item and get_item won't be the same | 13:22 |
| stgraber | how so? | 13:22 |
| hallyn | stgraber: i can add some special cases without too much trouble, but want to make sur eit sconsistent | 13:22 |
| hallyn | because you can't get_item(cgroup.devices.allow). try 'cat /sys/fs/cgroup/devices/devices.allow' :) | 13:23 |
| stgraber | well, get_config_item(cgroup.devices.allow works here ;) | 13:23 |
| hallyn | uh. oh. yeah i'm confusing it with the real cgroup | 13:23 |
| stgraber | >>> test.get_config_item("lxc.cgroup.devices.allow") | 13:24 |
| stgraber | ['c *:* m', 'b *:* m', 'c 1:3 rwm', 'c 1:5 rwm', 'c 5:1 rwm', 'c 5:0 rwm', 'c 1:9 rwm', 'c 1:8 rwm', 'c 136:* rwm', 'c 5:2 rwm', 'c 254:0 rwm', 'c 10:229 rwm', 'c 10:200 rwm', 'c 1:7 rwm', 'c 10:228 rwm', 'c 10:232 rwm'] | 13:24 |
| stgraber | I'm trying to make it so that test.set_config_item(key, test.get_config_item(key)) usually works | 13:24 |
| stgraber | (as that's what I think people will expect) | 13:24 |
| hallyn | hm is that how i did that | 13:25 |
| hallyn | all right yeah i can do that for cgroup | 13:26 |
| hallyn | hm, i hope my 'bzr pull' did the right thing | 13:28 |
| hallyn | no it unapplied all patches | 13:29 |
| hallyn | whatever (rm -rf and re-fetch worked. bzr trying to mess with us) | 13:30 |
| hallyn | stgraber: please try that (just pushed). | 13:35 |
| hallyn | zul: ok, regarding tftpd-hpa. It doesn't start when my laptop starts up, because my wireless isn't up yet at runlevel 2. | 13:36 |
| hallyn | zul: I can trivially reproduce with a container with empty network namespace | 13:36 |
| hallyn | zul: is this a bug, or expected behavior? | 13:36 |
| hallyn | i'd say it's a bug... | 13:36 |
| hallyn | will file it, just wanna make sure i won't get yelled at :) | 13:37 |
| zul | id say its a bug, but you can modify the upstart script to start on the wireless cant you? | 13:37 |
| zul | hallyn: you are going to get yelled at anyways :) | 13:37 |
| stgraber | hallyn: testing | 13:37 |
| hallyn | zul: s'ok, i might be going deaf, yell away :) | 13:37 |
| hallyn | ok will file | 13:37 |
| hallyn | zul: well not jsut wireless. but i'm sure there's a way to say 'wait until a nic is up'. I just forget what it is | 13:38 |
| hallyn | SpamapS would remember of course | 13:38 |
| zul | hallyn: and i might be giong blind :) | 13:39 |
| hallyn | zul: bug 1027081 | 13:42 |
| uvirtbot | Launchpad bug 1027081 in tftp-hpa "fails if network is not yet configured at runlevel 2" [Medium,New] https://launchpad.net/bugs/1027081 | 13:42 |
| hallyn | biab | 13:42 |
| stgraber | hallyn: can you think of any reason why bind-mounting a tmpfs would create a separate tmpfs instance (empty)? | 13:48 |
| stgraber | hallyn: that's affecting pstolowski. He has /tmp as a tmpfs, sets up lxc to bind mount /tmp/.X11-unix, the bind-mount shows up in /proc/mounts in the container but the dir is empty | 13:49 |
| uvirtbot | New bug: #1027081 in tftp-hpa (main) "fails if network is not yet configured at runlevel 2" [Medium,New] https://launchpad.net/bugs/1027081 | 13:50 |
| stgraber | hallyn: hmm, at least part of the problem is that something in the boot sequence wipes /tmp | 13:59 |
| stgraber | hallyn: another problem I noticed here is that lxc-clone doesn't change the path to the fstab | 14:00 |
| jhobbs | a/wg 9 | 14:00 |
| stgraber | hallyn: so even though I had a proper fstab, the mount still wouldn't happen as it was looking at the source container's fstab... | 14:00 |
| SpamapS | zul: the way to say "wait until a nic is up" is "runlevel [2345]" | 14:03 |
| SpamapS | hallyn: ^^ | 14:03 |
| zul | SpamapS: ah ok | 14:03 |
| SpamapS | because its not "a nic" you want | 14:03 |
| SpamapS | its "all nics" | 14:03 |
| SpamapS | if you want a specific interface.. you want network-interface-up IFACE=xxx | 14:04 |
| SpamapS | hallyn: sorry reading the backscroll now... | 14:05 |
| SpamapS | hallyn: so it fails if its brought up w/o any interfaces? | 14:06 |
| hallyn | SpamapS: yes | 14:07 |
| hallyn | SpamapS: and i'm not sure what the state of my laptop is when it starts. it boots in about 8 seconds so it's certainly nto waiting for failsafe | 14:08 |
| hallyn | but since it's serving my other laptop i'm not going to reboot to test right now | 14:08 |
| hallyn | stgraber: eh what? lxc-clone doesn't change the path to /var/lib/lxc/<newcontainer>/fstab in the config?? | 14:08 |
| hallyn | i'm convinced it used to | 14:08 |
| hallyn | yeah still does, at least in precise | 14:09 |
| hallyn | GAH | 14:09 |
| hallyn | that guy who made conflicting changes upstream, didn't keep that | 14:10 |
| SpamapS | hallyn: yeah, transient interfaces like wifi are a problem for server stuff in general. :p | 14:10 |
| hallyn | stgraber: rhanna whoever that is | 14:11 |
| hallyn | SpamapS: yes, i recall it coming up for somethign else. don't remember what though | 14:11 |
| uvirtbot | New bug: #1017972 in libunwind (universe) "[MIR] google-perftools, libunwind" [Medium,Incomplete] https://launchpad.net/bugs/1017972 | 14:11 |
| hallyn | SpamapS: any guidance ? or is it complicated enough to discuss at next uds? | 14:11 |
| hallyn | stgraber: do you have other changes queued for quantal? | 14:12 |
| stgraber | hallyn: I might have an apparmor change, hold on a sec, checking | 14:12 |
| SpamapS | hallyn: its a bit of a corner case.. but its basically the reason there's a need for a way to define states that you want for your job, rather than events | 14:12 |
| hallyn | SpamapS: HUSH! | 14:12 |
| hallyn | :) | 14:12 |
| hallyn | SpamapS: maybe since in.tftpd binds to 0.0.0.0 it really wants to restart any time a new nic comes up | 14:13 |
| SpamapS | hallyn: if it binds to 0.0.0.0 it should never need to be restarted | 14:13 |
| hallyn | stgraber: lemme first send the fix upstream. GRRR. | 14:13 |
| SpamapS | hallyn: its only stuff that wants to bind to specific IPs that are problematic | 14:13 |
| stgraber | hallyn: http://paste.ubuntu.com/1101870/ is the change I have here (not commited or anything), would be good if you could include it | 14:14 |
| stgraber | hallyn: basically my nesting apparmor profile was working great to run containers, not so much for creating them ;) | 14:15 |
| hallyn | stgraber: boy that tosses security right out the window | 14:15 |
| stgraber | hallyn: sure but that's kind of the point of the lxc-default-with-nesting profile isn't it? :) | 14:16 |
| hallyn | i guess we really need jjohansen's upcoming 'deny @{PROC} /sysrq-trigger w' | 14:16 |
| hallyn | yeah | 14:16 |
| hallyn | stgraber: ok, will include that, thanks | 14:17 |
| hallyn | btw, dlezcano has pulled my github tree, but is still testing the results (or something) | 14:17 |
| hallyn | so we may just get a release soon | 14:17 |
| hallyn | SpamapS: well i woul dhave thought that bind(0.0.0.0) would just bind to lo, then pick up eth0 automatically. but it doesn't. so i toss my assumptions aside | 14:17 |
| stgraber | hallyn: cool, that should reduce the number of patches by 50 or so ;) | 14:18 |
| SpamapS | hallyn: perhaps it does more than bind to 0.0.0.0 | 14:18 |
| hallyn | SpamapS: yeah, i guess, makes sense | 14:19 |
| hallyn | SpamapS: like phone home! :) | 14:19 |
| SpamapS | hallyn: for instance, named sometimes takes 0.0.0.0 to mean "bind to all the addresses" not "bind to 0.0.0.0" | 14:19 |
| hallyn | stgraber: wait, i just found the line that is supposed to change the fstab location. are you sure it's not working? | 14:19 |
| stgraber | hallyn: http://paste.ubuntu.com/1101882/ | 14:20 |
| stgraber | hallyn: that's when running my PPA packages (lxc-api-and-python branch), but I don't think lxc-clone is different from that in quantal (shouldn't be at least) | 14:21 |
| hallyn | stgraber: no it isn't, i see it here. i'll figure out why... | 14:23 |
| hallyn | stgraber: found it. our template inserts 'lxc.mount =' (two spaces before the =), and rhanna's changes to lxc-clone hardcoded a check with one space | 14:30 |
| hallyn | stgraber: testing http://people.canonical.com/~serge/lxc-clone.debdiff | 14:40 |
| hallyn | stgraber: ok, pushed. | 14:45 |
| === n0ts is now known as n0ts_off | ||
| glosoli | Does all Ubuntu 12.04 has outdated kernel 2,x ? | 14:48 |
| patdk-wk | heh? | 14:49 |
| patdk-wk | they all come with 3.2 | 14:49 |
| glosoli | hmmm | 14:49 |
| glosoli | what the heck.. | 14:49 |
| glosoli | I bought a server, and now for curiosity typed uname -a and ..... | 14:49 |
| glosoli | Linux 273170 2.6.32-308.el5.028stab099.3 #1 SMP Wed Mar 7 15:56:00 MSK 2012 x86_64 x86_64 x86_64 GNU/Linux | 14:49 |
| stgraber | is that a physical server or a container? | 14:50 |
| glosoli | probably container | 14:50 |
| mardraum | looks like it | 14:50 |
| stgraber | then that makes sense | 14:50 |
| hallyn | intersting | 14:50 |
| stgraber | 12.04 running on a RHEL6 kernel would give you that output | 14:50 |
| glosoli | stgraber: aaa | 14:50 |
| glosoli | thanks | 14:50 |
| mardraum | "el5" | 14:51 |
| stgraber | or RHEL5 indeed :) | 14:51 |
| stgraber | hmm, 2.6.32 RHEL5 that seems odd, I thought they had a 2.6.18 in that one? (but I don't pretend to know what the RH guys are doing) | 14:51 |
| mardraum | yeah I think you are right | 14:52 |
| mardraum | it's something nasty regardless :p | 14:53 |
| stgraber | sounds like it's: http://wiki.openvz.org/Download/kernel/rhel5/028stab099.3 | 14:53 |
| glosoli | Interesting | 14:53 |
| stgraber | hallyn: http://paste.ubuntu.com/1101935/ apparently clear stopped returning False, but that's about all it did ;) | 14:58 |
| hallyn | stgraber: sigh | 15:10 |
| hallyn | stgraber: ok, i see. a few mins to push. | 15:26 |
| === n0ts_off is now known as n0ts | ||
| jamespage | zul, where would you like me to put the openvswitch1.6.1 packages? I've tested them locally and they look OK? | 15:28 |
| zul | do you want me to review them or? | 15:28 |
| obelus | Hey - I'm going to be migrating my Ubuntu server from a VM to a physical box, and I'm looking for a way to get a list of all my manually installed packages. I've searched for a solution, but none of them seem to work, and the list of default packages provided for the desktop version in the .manifest file doesn't seem to exist for the server version of Ubuntu | 15:31 |
| obelus | Any pointers? | 15:31 |
| hallyn | stgraber: pushed, works for me under python | 15:33 |
| stgraber | yay | 15:33 |
| hallyn | stgraber: (i was thrown by my own adding 'lxc.cgroup' to the get_item(lxc.cgroup) results :) | 15:34 |
| uvirtbot | New bug: #1027130 in clamav (main) "package clamav-base (not installed) failed to install/upgrade: el subproceso instalado el script post-installation devolvió el código de salida de error 1" [Undecided,New] https://launchpad.net/bugs/1027130 | 15:41 |
| uvirtbot | New bug: #1024086 in python-tx-tftp (universe) "[MIR] python-tx-tftp" [High,New] https://launchpad.net/bugs/1024086 | 15:46 |
| hallyn | hm, usb-creator-gtk under fvwm (in vncserver) is complaining about not talking to udisk | 15:48 |
| stgraber | hallyn: confirmed "test.set_config_item("lxc.cgroup.devices.allow", test.get_config_item("lxc.cgroup.devices.allow"))" works fine now! | 15:50 |
| hallyn | stgraber: \o/ | 15:52 |
| stgraber | hallyn: btw, get_config_item in python is full of magic to try and figure out whether lists should be accepted or not and checking that the value is set properly and reverts to the old value if something goes wrong. Seems to be working well in my tests so far ;) | 15:53 |
| hallyn | stgraber: is the magic a workaround for soething lacking in the c lib? (i may have to take a look) | 15:54 |
| stgraber | hallyn: not really, there isn't much the C library could do to make it work better. The only way to know if a key contains a list is to check its value, which doesn't work when its unset (obviously). So set_config_item in python will always save the current value, clear the key (in case it's a list), set the new value(s), then check that the new value matches the input, if it doesn't, it reverts the value and returns False | 16:00 |
| hallyn | ok | 16:01 |
| obelus | I guess if I don't find a solution I can just install a base copy of the server and compare a package list from that. Probably faster than trying to find a way to do it in my current install, though I would rather a more 'proper' solution | 16:03 |
| === imbrandon is now known as Guest41572 | ||
| === Guest41572 is now known as imbrandon | ||
| === imbrandon is now known as imbrandon_ | ||
| === imbrandon_ is now known as imbrandon | ||
| uvirtbot | New bug: #1027154 in maas "need way to get system based on mac address" [Undecided,New] https://launchpad.net/bugs/1027154 | 16:41 |
| kyle__ | xpost from ubuntu, but hey. For those running labs of ubuntu boxes, with 12.04 is it more streightforward to authenticate off of LDAP or a samba PDC? In my case all user-files are shared via samba, for various reasons. I have no physical windows boxes (nor do I intend to) | 17:14 |
| === n0ts is now known as n0ts_off | ||
| jtran | anyone know much about xtables-addons? i'm trying to get ip accounting working | 17:24 |
| jtran | get_handle_usage failed: Can't get handle usage information from kernel | 17:24 |
| === jdstrand_ is now known as jdstrand | ||
| RNAndyB | I recently set up an ubuntu server with a 1TB disk. I have /home mounted on a partition of 898Gb. DF is showing 6.9G of usage but there are only a handful of small files in my home directory that du totals to about 500k. Also LSOF doesnt show any large files being open under /home. Could anyone point me in the direction of discovering what the 6.9G is being used for? | 17:29 |
| hallyn | SpamapS: lp:~serge-hallyn/ubuntu/quantal/pulseaudio/pa-upstart updated, seems to test ok | 17:59 |
| * hallyn lunch | 17:59 | |
| SpamapS | hallyn: last niggle. You do not need the script/end script on the exec line | 18:02 |
| SpamapS | hallyn: be careful though, that may mean you can drop from 'expect daemon' to 'expect fork' | 18:03 |
| Phibs | I'm trying to cobbler/orchestra 12.04 server ISO, it finds my preseed file but each time it asks me for a cdrom, any help would be greatly appreciated. | 18:27 |
| Phibs | anyone? | 18:34 |
| IdleOne | patience is requires | 18:36 |
| IdleOne | required also | 18:37 |
| Phibs | indeed :) | 18:37 |
| IdleOne | Phibs: you might try asking on askubuntu.com or the forums | 18:37 |
| IdleOne | both great support resources | 18:37 |
| Phibs | okie | 18:37 |
| Phibs | thanks | 18:37 |
| stgraber | hallyn: "apt-get install --reinstall initscripts" <- is that good enough a testcase for the initscripts bug? | 18:51 |
| Phibs | Every time it boots it doesn't try the network (but it does DL the preseed file), it tries the cdrom and fails | 18:51 |
| hallyn | stgraber: i don't think so, i think you also need to debootstrap --dsecondstage with the new deb, as that has a different /dev contents | 18:52 |
| hallyn | SpamapS: will just killing the pulseaudio daemon and watching it start back up be enough confirmation that 'expect daemon' is fine? | 18:55 |
| stgraber | hallyn: well, that's going to be difficult to do outside of quantal... the --reinstall definitely triggers the bug here (and explains why some people keep reporting it on precise) | 18:55 |
| hallyn | stgraber: https://wiki.ubuntu.com/SergeHallyn_debootstrap ... i show I originally was reproducing it | 19:00 |
| Phibs | looks like it's not detecting the network drivers | 19:00 |
| hallyn | I'm probably being dense, but don't grok why that'll be hard to do outside of quantal | 19:00 |
| hallyn | because of -updates? | 19:00 |
| wizonesolutions | mardraum: jamespage: Thanks, I will go with virt-install. RoyK: Is it a text-based GUI? I'm not putting X on this box.. | 19:02 |
| stgraber | hallyn: don't you need them to match the version number of these in precise for that trick to work? | 19:02 |
| stgraber | hallyn: testing with quantal is easy as debootstrap will use the release pocket and just work. For -proposed, you'd need to have debootstrap grab from release + updates + proposed pocket to test | 19:03 |
| stgraber | anyway, the fix doesn't work, so I need to dig more | 19:03 |
| stgraber | might be that the fix works for a clean debootstrap, but it's going to break any existing container on upgrade | 19:03 |
| hallyn | stgraber: yeah i guess i hand-built a version with the same package version | 19:03 |
| hallyn | that's no good :) | 19:04 |
| stgraber | for me the debootstrap issue isn't the worst really, that one we workaround already :) the problem is that currently any update of initscripts in a container will fail, so the first initscripts SRU needs to fix that before we can upgrade to quantal or even push another initscripts SRU | 19:04 |
| hallyn | SpamapS: bzr+ssh://bazaar.launchpad.net/~serge-hallyn/ubuntu/quantal/pulseaudio/pa-upstart/ seems to still work (killing pulseaudio, it gets restarted) | 19:05 |
| hallyn | stgraber: i thought we only worked around it in precise, not quantal | 19:05 |
| hallyn | maybe you changed that :) | 19:05 |
| hallyn | hm, yeah, i guess it's handled for all cases. | 19:06 |
| SpamapS | hallyn: stop/start, not kill | 19:06 |
| SpamapS | hallyn: killing the daemon will not reload the job config | 19:07 |
| SpamapS | hallyn: and be careful | 19:07 |
| SpamapS | hallyn: if its wrong, upstart will have a phantom entry for the pid that has already died | 19:07 |
| hallyn | SpamapS: stop/start worked too 9as did restart) | 19:08 |
| stgraber | hallyn: hmm, actually, what was the workaround in lxc again? I'm wondering if we didn't indeed regress in quantal which would explain why the fix doesn't quite work here. | 19:08 |
| hallyn | stgraber: /usr/lib/lxc/templates/lxc-ubuntu, check for /dev/shm | 19:09 |
| stgraber | hallyn: I'd expect the new .postinst to work fine when /dev/shm is a symlink to /run/shm and I thought that's what we did in lxc, but checking my templates, /dev/shm is a directory... | 19:09 |
| hallyn | stgraber: on precise, we're doing http://paste.ubuntu.com/1102286/ | 19:10 |
| hallyn | now my q containers on p actually don't have /dev/shm at all | 19:10 |
| SpamapS | hallyn: alright sweet, I'm about to head to lunch, but I'll most likely upload it after lunch | 19:11 |
| hallyn | SpamapS: thx | 19:11 |
| stgraber | hallyn: yeah, but that code depends on /run/shm existing, which isn't the case in the templates here (as /run/shm is created at boot time) | 19:11 |
| stgraber | hallyn: confirmed that all my machines with a clean precise lxc + recent precise cache don't have /run/shm in the cache and so /dev/shm is a directory instead of a symlink | 19:13 |
| stgraber | so the workaround doesn't actually work | 19:13 |
| hallyn | it definately used to | 19:14 |
| stgraber | maybe something was nice enough to mkdir -p /run/shm at some point and that disappeared at some point before release | 19:14 |
| hallyn | yeah... | 19:14 |
| === cpg|away is now known as cpg | ||
| stgraber | hallyn: so sounds like the SRU might be correct to fix the debootstrap/chroot case but we need some more code to fix the containers before hitting that part of the code, otherwise it's hitting the bind_mount code which gets rejected by apparmor | 19:22 |
| RoyK | wizonesolutions: it's X-based, but you don't need to start X on the server to make X work over ssh | 19:24 |
| uvirtbot | New bug: #1027219 in cobbler (universe) "[Precise] Default install assumes MaaS" [Undecided,New] https://launchpad.net/bugs/1027219 | 19:31 |
| hallyn | stgraber: hrmph. we should probably allow bind mounts | 19:37 |
| hallyn | bind mounting /proc to /fakerproc won't bypass the /proc/sysrq-trigger checks | 19:38 |
| stgraber | hallyn: you sure? I'm pretty sure it'll bypass the check | 19:39 |
| * stgraber tests | 19:40 | |
| stgraber | root@python-lxc:~/python-lxc/mapping# echo h > /proc/sysrq-trigger | 19:41 |
| stgraber | -bash: /proc/sysrq-trigger: Permission denied | 19:41 |
| stgraber | root@python-lxc:~/python-lxc/mapping# mount --bind /proc/ /mnt | 19:41 |
| stgraber | root@python-lxc:~/python-lxc/mapping# echo h > /mntsysrq-trigger | 19:41 |
| stgraber | hallyn: so nope, we can't allow them :) | 19:41 |
| hallyn | stgraber: /mntsysrq-trigger or /mnt/sysrq-trigger? | 19:47 |
| hallyn | (just making sure) | 19:47 |
| stgraber | oh, typo, though the result is the same | 19:49 |
| stgraber | root@python-lxc:~/python-lxc/mapping# echo h > /proc/sysrq-trigger | 19:49 |
| stgraber | -bash: /proc/sysrq-trigger: Permission denied | 19:49 |
| stgraber | root@python-lxc:~/python-lxc/mapping# echo h > /mnt/sysrq-trigger | 19:49 |
| stgraber | root@python-lxc:~/python-lxc/mapping# | 19:49 |
| hallyn | stgraber: (i don't have a nested profile handy so couldn't easily check :) | 19:49 |
| hallyn | stgraber: drat. i must mis-recall what jjohansen said | 19:49 |
| stgraber | me neither, I just added a "mount," line to my default profile | 19:49 |
| Phibs | got it to install, now only issue is its not setting the network config after using dhcp to install | 19:50 |
| stgraber | well, that container had the nested profile, but I couldn't remember exactly where it was allowed to bind mount, so I just allowed all mounts for the test ;) | 19:50 |
| hallyn | stgraber: yeah, confirmed... (just *had* to check myself, bc i really was hoping) | 19:52 |
| hallyn | stgraber: I bet this is a result of attach_disconnected, actually | 19:53 |
| Phibs | anyone know how to make the installer wipe out existing partitions in preseed ? | 20:11 |
| guntbert | Phibs: I let it create the complete partitions afresh | 20:15 |
| Phibs | guntbert: how do I clear | 20:15 |
| guntbert | Phibs: please see https://help.ubuntu.com/12.04/installation-guide/i386/preseed-contents.html#preseed-partman | 20:17 |
| Phibs | thanks | 20:17 |
| guntbert | Phibs: no problem, there are several example preseed files to be found, if you want you can look at what I have found: http://delicious.com/guntbert/preseed | 20:22 |
| Phibs | yeah still having broken probs :( | 20:23 |
| Phibs | and i'm using cobbler / orch so might be diff | 20:23 |
| Phibs | getting this msg | 20:25 |
| Phibs | http://screencast.com/t/8ywVouva0z | 20:25 |
| Phibs | anyone know how to get cobbler/orchestra to set the static IP as configured for the system, and to have it check in with cobbler/orch when done? | 20:38 |
| guntbert | Phibs: d-i partman-lvm/confirm boolean true *should* help (they say on the above mentioned page) | 20:38 |
| Phibs | guntbert: yeah that ws the issue | 20:38 |
| Phibs | thanks for that ref | 20:38 |
| Phibs | just a few more hoops and i'll be much happier ;0 | 20:38 |
| guntbert | Phibs: I'm in the middle of a similar project: configure ~20 lab pc for running as VM hosts, I don't intend to use orchestra, cobbler is enough for me, but apt-cacher turns out to be very useful (I can leave the software-sources unchanged) | 20:42 |
| Phibs | guntbert: orchestra is cobbler ;0 | 20:44 |
| Phibs | with a .css file ;0 | 20:44 |
| guntbert | Phibs: if I read correctly cobbler is part of orchestra | 20:45 |
| Phibs | could be more than just cobbler | 20:46 |
| Phibs | the screenshot is exactly the same ;0 | 20:46 |
| Phibs | http://3.bp.blogspot.com/-j8Fs7iROCMo/Tqm4PW-_9ZI/AAAAAAAAEfE/Gf3AvSu5gc4/s1600/cobbler-add.png | 20:46 |
| guntbert | Phibs: as I am about to go on vacation I will delay those things for a few days :) | 20:52 |
| Phibs | ;0 | 20:53 |
| hallyn | ahs3: yay, netcf 2.0 was just released | 20:54 |
| hallyn | i'll work on the debian package monday | 20:54 |
| Phibs | ah ok, the cobbler part of orchestra is the provis server, it has 3 other parts | 20:54 |
| ahs3 | hallyn: ah, cool. holler when ready | 20:56 |
| hallyn | ahs3: will do | 20:57 |
| === bitmonk_ is now known as bitmonk | ||
| === cpg is now known as cpg|away | ||
| uvirtbot | New bug: #1027259 in samba (main) "postgresql-common E: Sub-process /usr/bin/dpkg returned an error code (1)" [Undecided,New] https://launchpad.net/bugs/1027259 | 21:21 |
| halvors | I want to setup a wifi accesspoint on my server, but i can't get it working using the "interfaces" file in "/etc/network". Here is what i got for now: http://pastebin.com/hWz3bDhd | 21:25 |
| stgraber | hallyn: rebasing the branch on ubuntu:lxc, please don't commit to it in the next 30min or so (time required for bzr to rebase with all the quilt stuff ;)) | 21:33 |
| halvors | What happend to the opendchub package? | 21:36 |
| hallyn | stgraber: ok | 21:38 |
| hggdh | zul: Tan wants a response from you :-) | 21:47 |
| Phibs | anyone know how to get the static ip info from cobbler into /etc/network/interfaces post install? | 21:49 |
| stgraber | hallyn: rebased btw (had some problems that I had to fix with an extra commit ...). You'll need to pull with --overwrite next time. | 22:26 |
| hallyn | stgraber: I rm -rf and re-branch every time :) | 22:27 |
| hallyn | s'ok, I did 'bzr init-repo lxc', and do it inside lxc/, so it's all cached locally | 22:27 |
| stgraber | :) | 22:28 |
| hallyn | man, i'd log off if only this one build test would FINISH | 22:28 |
| stgraber | I should probably spend some time fixing bzr-builddeb to stop being annoying instead of working around it :) | 22:28 |
| hallyn | hm, i forget why i don't do bzr bd | 22:29 |
| hallyn | but i don't. i always debian/rules build && fakeroot debian/rules binary | 22:29 |
| hallyn | oh. i guess it's bc most times i ahve syntax errors and i want to see them, fix them and type 'make' | 22:29 |
| hallyn | iow, cause i suck | 22:29 |
| stgraber | I'm using it to build my packages, works fine, the problem is really when merging/pulling/rebasing that it's trying to be clever with quilt and failing really badly | 22:29 |
| hallyn | oh, yes, like i said, when i pulled your tree this morning (into my tree from yesterday, just 'bzr pull'), it left me with all quilt patches unapplied | 22:30 |
| hallyn | even though it wasn't like that in yoru tree - when i did rm -rf and re-branch, it was fine | 22:30 |
| hallyn | all right i'm off \o | 22:33 |
| Neal_ | How can I make a user have access to file owned by another user? | 22:39 |
| Neal_ | Like file f is owned by asdf:asdf but I want user neal to have access to it | 22:40 |
| jtran | anyone know anythng about xtables and IPACCOUNT? | 22:52 |
| MoleMan | Neal_: either allow all users access, or create a group to add the user to, and chgroup <groupname> the file | 23:11 |
| wizonesolutions | RoyK: Hmm, if I have a hard time with the command line I might give that a try. I've never used X over SSH before, though I've heard/read about it being possible. | 23:45 |
| === guampa_ is now known as guampa | ||
| === cpg|away is now known as cpg | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
