[00:09] The keystone charms is giving me an "install-error" in the juju gui, and "agent-state: install-error" in the juju cli. Is there a way I can look at a log or debug this to see why it failed? I'm using MaaS underneath Juju (and actually MaaS in running on top of KVM machines for testing). === freeflying_away is now known as freeflying [00:15] Azendale: juju debug-hooks [00:22] bigjools: I just tried that, now I'm ssh'd in to the box. Is there a log somewhere of Juju's attempts to configure the machine? [00:23] Azendale: I'm no expert but look in /var/log/juju on the bootstrap machine [00:24] bigjools: ok, thanks for the advice, I'll try it [00:36] are there 32-bit and 64-bit version of the php5 deb package? [00:36] 'm trying "apt-get install php5:i386" but it can't find the package [02:11] hey guys [02:11] I keep getting this really annoying email from PHP: [02:11] PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php5/20090626/mcrypt.so' - /usr/lib/php5/20090626/mcrypt.so: cannot open shared object file: No such file or directory in Unknown on line 0 [02:12] Get heaps of emails with this same message. Has anyone come across it? [02:16] jotterbot1234, Do you have php5-mcrypt installed? [02:18] yes [02:18] Apparently it is a known issue for Debian [02:18] Working off this: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=662090 [02:18] Debian bug 662090 in php5 "php5: Update problem with package php5" [Normal,Fixed] [02:19] Are you running mod_php or php-fpm? [02:19] mod_php i believe [02:20] Can i get you to do a service apache2 restart ? [02:20] just the standard LAMP install I thought. It's amazon EC2 though so I didn't set it up [02:20] sure, one moment [02:21] actually [02:23] I think I've pegged it. Wasn't installed on one of the other VMs. There are a few running behind a load balancer. [02:23] so frustrating! I need config management like puppet or something running! nevermind [02:23] On an unrelated note... can someone update the /topic ? 13.10's the in-dev version, not 12.10 [02:24] jotterbot1234, Trust me, the moment you say puppet, you'll be wanting to do everything. If you go that route, do as little as possible with it. [02:26] Can you elaborate on what you mean? Are you saying that is a poor choice to run config management? I have been looking at a few. Especially salt stack vs puppet [02:27] No, I'm saying once you get a little taste of puppet, you'll want to do ALL THE THINGS IN PUPPET, which results in lost efficiency, and if you're like me, never getting the project done. [02:27] haha i see! [02:28] I thought it has been running long enough now that they have all templates shared so you don't have to reinvent the wheel so much [02:29] but I will heed your warning when looking over it! [02:29] Yeah, for the most part, but you have to configure each service separately. [02:29] I was at the point where I was installing oh-my-zsh via puppet before I put it down and forgot about it. [02:32] I just need something for some basic config/settings management. And then something like munki for software installs. I am trying to come up with a linux(specifically ubuntu) backend for OSX clients for a project === freeflying is now known as freeflying_away === thumper is now known as thumper-afk [04:38] Hi, I am going to make a backup server in a domain with windows servers and windows clients. Could you please tell me from where should I start? (I prefer linux backup server) === freeflying_away is now known as freeflying [06:25] o/ [06:25] nfs help here please (linux) [06:25] i have two directories listed in exports. I can only mount the one that is listed first [06:26] it seem the exportfs only exports the first entry. [06:45] where is exports? [06:50] it is located in /etc/exports [06:56] why isn't there a 13.04 server iso? [06:57] why isn't there a 13.04 server iso for 32-bit*? [07:02] gartral: "i386" [07:11] mardraum: that option doesn't appear on the site [07:11] and i've looked for 10 minutes [07:18] gartral: http://releases.ubuntu.com/13.04/ near the end you will find ubuntu-13.04-server-i386.iso === thumper-afk is now known as thumper [12:08] If I have multiple interfaces, the first one using dhcp and the others dhcp or static, what is the best way to define the default route manually? [12:08] I just do not want to have eth0 as default route [12:09] Best answer I found so far was to have reject lines in dhclient.conf [12:10] raub: there are many ways to do this of course. I would go with the dhclient route too, since logically you want DHCP but do not want to set the default route via DHCP. So configuring dhclient to not use the default gateway supplied by the DHCP server seems like the right place to do that. [12:11] raub: I'm not sure about the reject lines. I've done it in the past by writing a trivial /etc/dhcp3/dhclient-enter-hooks.d/ script that checks $interface, sets new_routers to empty and copies old_domain_name_servers to new_domain_name_servers. === gary_poster|away is now known as gary_poster [12:43] Can anyone help installing Ubuntu server w/ CD? [12:43] Booting from CD takes to GRUB command entry screen [12:43] Not the menu entry screen [12:43] As such, I don't know what to type to make it start. [12:44] alias_bij: sounds like the cd is bad [12:44] alias_bij: it should take you to the boot menu [12:44] ikonia: ok, thanks; I'll try another CD. [12:45] alias_bij: also check the image you are burning with the checksums [12:45] make sure it's valid [12:45] ikonia: good point; i'll verify that too. thank you. [13:03] hello when i'm logged in in my ubuntu system tab autocompleet does not work at all [13:03] how should i fix this [13:13] smoser: fyi, libvirt's filesystem binding thing uses qemu's "virtfs" thing, which is some kind of network filesystem (plan 9?). So I don't think it's as bad a hack as you thnk. [13:13] (ie. it's properly doing things at fs layer, not block layer) [13:13] i think its a hack [13:13] not for sharing filesystem data. i think thats fine. [13:13] I can see how it might be useful for debugging. Depends on the use case that Tim wants, I think. [13:13] i think juju insisting on looking directly (or reading from) the filesystem of the thing that its using is a hack. [13:14] I'm undecided. I'd like to understand the use case better. [13:14] and my suggestion of solving that problem generally still stands. [13:14] ie, cifs or nfsv4 would be better. [13:15] The nice thing about qemu's built in support is that I don't have to run or configure any additional daemon. [13:15] (I wasn't aware that it existed until today) [13:15] utlemming had been using that a bit. [13:15] it does work. [13:15] and i think we even at some point added the plan 9 into thei cloudimage for that purpose. [13:15] the modules. [13:16] but for juju's purposes i think its a hack. for debugging its fine. but even then, i really think better to set it up generically using tcp-ip. [13:16] all the cool kids are using tcp-ip these days. [13:17] rbasak, this is exaclty the same argument i've made at ODS many times. [13:17] taking advantage of a hypervisor specific "host<->guest" communication layer means that all hypervisors have to implement it. [13:18] (well, not it specifically, but some form of it). [13:18] Sure, I do see your point of view. I'd just prefer to reserve judgement until I understand the use case. [13:19] For the debugging case, I think a one-command-option request to map the fs across would be useful. I'm not sure how to do that without set up on the host any other way. [13:19] yeah... the issue is that that one-command-option becomes precedence. [13:19] However, even this way it's a bit of a pain from an apparmor perspective. At least in Precise, aa-virt-helper doesn't appear to understand the filesystem "binding" definitions. [13:19] and the next hypervisor does it the same way [13:20] (which is exactly what is happening here) [13:20] and hten you have to know: if i'm on lxc, then i look here, if i'm on kvm, then i look here .... [13:20] So I'm not sure it's worth implementing right now, at least until I understand juju's need better. [13:24] Why is the dhclient trying to connect to a bootp server? [13:24] From netstat, udp 0 0 0.0.0.0:68 0.0.0.0:* 805/dhclient3 [13:24] raub: dhcp is a super-subset of bootp [13:26] jrwren: Aha === Guest8446 is now known as wedgwood [13:42] Hey guys, I'm running ubuntu 10.10 on an old server, the apt repos are down and I don't have the update manager, what's are my upgrade options here? [13:45] goles: 10.10 isn't supported anymore [13:46] !eolupgrades [13:46] End-Of-Life is the time when security updates and support for an Ubuntu release stop, see https://wiki.ubuntu.com/Releases for more information. Looking to upgrade from an EOL release? See https://help.ubuntu.com/community/EOLUpgrades [13:49] RoyK: I see that :), I was wondering what could I do [13:49] see eolupgrades above [13:54] RoyK: It's weird, I'm following the instructions but I'm failing to install the update-manager-core, etc. https://gist.github.com/Goles/06ea4f67c3c7e33b099a [13:56] RoyK: that page must be quite outdated... I mean the latest instructions are for Intrepid to Karmic [14:12] goles: Basically the same, but here is the AU link: http://askubuntu.com/questions/91815/how-to-install-software-or-upgrade-from-old-unsupported-release [14:12] Note that you also have to „convert” the security.... repos. [14:28] hi [14:29] I am looking to script install mysql [14:29] what is the best way to silently set a random password? [14:30] what would the default password be if I apt-get install -yq ? [14:31] http://stackoverflow.com/questions/7739645/install-mysql-on-ubuntu-without-password-prompt [14:31] my answer ^^ [15:54] stgraber: hm, lxc in saucy isn't working well for me at all... [15:54] hallyn_: ah? what's the problem? [15:54] well (1) pstore being in container's fstab is causing it to fail - which isn't supposed to happen [15:55] and (2) there seems to be a problem with lxc monitord startup [15:55] (bug 1225308 was opened for that) [15:55] Launchpad bug 1225308 in lxc "Segmentation fault with "lxc-ls --fancy"" [Critical,Triaged] https://launchpad.net/bugs/1225308 [15:56] hallyn_: hmm, let me create a new clean container here [15:56] hallyn_: lxc-list (alias to lxc-ls -f) sure works fine here [15:56] i created a new saucy container on my precise host. in there is where i'm having trouble [16:01] hm. i did NOT think we required 'optional' in fstab entries to make them optional. [16:02] lxc-stop -k is also broken [16:02] oh, that's nested cgroup stuff [16:03] * hallyn_ shifts to a kvm instance [16:39] stgraber: oh, uh. duh. we can't just move the lxc-monitor location like that can we? /me does a little test [16:42] that is, we can't just move the monitor to /run, breaks already-running containers. [16:45] hallyn_: right, I'd just mark any bug resulting from that as won't fix and asking the user to just reboot their system [16:46] drat. [16:46] hallyn_: that'd be unacceptable for an SRU but in the dev release I don't think we should maintain migration code [16:46] i guess [16:47] ok i'm going to push a patch to git and saucy to add 'optional' to the pstore fstab entry [16:54] hallyn_: why did we need that? for kernels that don't have /sys/fs/pstore? [16:54] yes [16:55] ok. When you cherry-pick to saucy, try to keep the same format for debian/patches. So far I've been doing simple git format-patch, renumbered the patch and dumped it into debian/patches/ [16:55] so i guess realistically it shouldn't hit people running the saucy package - if they do r->s upgrade they should reboot before starting a new container [16:55] that way the file name matches the commit title and the format makes it clear that it's a straight cherry-pick from upstream and can be dropped with the next milstone [16:56] ok, makes sense [16:56] right. Someone updating from raring to saucy is also likely to get hit by monitor API changes anyway, so rebooting is usually a good idea :) [16:57] im' worried about my precise-based containers host that hosts services (using lxc ppa) [16:57] guess we'll see. maybe i should reboot it today before i travel. [16:58] got a feelign it's going to be unhappy [16:58] well, so long as the containers don't crash as they'd in the past when the monitor API would change, all that'll break are the tools on the host [16:58] true [16:58] and you can fix that by doing a clean shutdown+reboot of all the containers === frojnd_ is now known as frojnd [16:59] all right, back to real work then [17:25] Hi all. Anyone on ubuntu-server 12.04 and using libvirt hooks? I cannot get my libvirt to execute the scripts in /etc/libvirt/hooks. This bug https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/891472 mentions something wrong with the apparmor profile. But in 12.04 libvirt 0.9.8-2ubuntu17.10 the hooks directory is mentioned in the apparmor profile. [17:25] Launchpad bug 891472 in libvirt "apparmor profile for libvirt does not allow hooks to be executed" [Low,Fix released] [17:27] I new to this channel, I assume uvirtbot just graps interesting links (like a bug tracker link) and quotes current status or so... [17:30] jonascj: yeah, the bot tries its best to be useful, but... :) [17:31] jonascj: check dmesg or /var/log/audit/audit.log or dmesg logs to see if there are apparmor DENIED messages for your hooks [17:39] sarnold: thank you for your reponse. I do not have a /var/log/audit/audit.log and "sudo grep audit /var/log/kern.log", "sudo grep hook /var/log/syslog" etc. yields nothing. [17:40] but I guess that means apparmor is not blocking anything [17:40] jonascj: DENIED might be more informative, the specific execution or access being denied might not show up with 'hook'. (I'm less sure about 'audit', it's been a while since I've run without auditd..) [17:42] sarnold: DENIED shows up in auditd too. Most of the message is the same, its mostly just the preamble that is different [17:42] sarnold: I can just install auditd if all it does it redirect apparmor messages [17:42] jjohansen: I was curious if 'audit' shows up in dmesg :) hehe [17:44] jonascj: auditd also does a lot of PAM auditing; it does enough extra things that I'd recommend installing it when you've got some time to read the documentation. (not that the docs are great, but it does more than just redirect apparmor messages.) [17:44] sarnold: so audit messages show up in dmesg if you are not running auditd. If you are running auditd they won't show up in dmesg [17:48] so maybe it is not app armor after all. === FunnyLookinHat_ is now known as FunnyLookinHat [18:24] hmm, I've found out that one "predefined" script /etc/libvirt/hooks/daemon works (if I create it and for example echo to stderr). But it is as if it does not scan the hooks-directory. [19:14] I have a daemon (nsqd) that for which i need to write an upstart script [19:14] it takes a bunch of cli options, including some of them repeated [19:14] it does not have a config file [19:15] any suggestions on how to best implement my own config file that will be read by the upstart config and then utilized in executing the start command? [19:50] adam_g/roaksoax: https://code.launchpad.net/~zulcss/python-keystoneclient/mox/+merge/185915 [19:53] hi guys, is it possible to creat a raid1 of two drives after ubuntu server is installed ? of so how ? (ubuntu is running on another drive) [19:58] anything is possible [20:10] kyze: Want that RAID1 to contain the actual OS install (the root partition, etc), or will it for a sepdata data storage filesystem? Also, is the drive Ubuntu is running from any of the two drives you want to create the RAID1 of? [20:10] andol no ubuntu is running off its own volume, its 2 new drives that shall run in raid 1 [20:12] kyze: Ahh, then there is no need to be creative. Assuming you want a software raid, just use mdadm to build a RAID1 pair of those two new drives. [20:13] would i need to format the drives beforehand ? is there a possibility to get exisiting data directly into the raid or do they have to be clean ? [20:16] kyze: So, you already have some data on those two disks? Well, unless you are sure on what you are doing the easiest and safest is really to temporary copy that data somewhere else, create the raid set, and copy the data back. [20:18] andol one drive has some data on it yeah, but if its not safe im going to get it on some external beforehand [20:18] The alternative would be copying all the data two one of the two drivees, creating a degraded raid1 on the other disk, create a new filsystem ontop of the degraded raid1, copy all the data from the other disk onto the degraded raid1 fs, then finally make cleanoun the data disk and join it to the now raid1. [20:19] Given that you might also want to be experementing until you get everything right, copying it to the external drive is probably a good idea. [20:26] andol thanks for your help, im definitely going to back it up to the external then maybe fool around a bit [20:38] smoser, still around? [20:38] adam_g, here. [20:38] smoser, regarding qemu, kvm +12.04.. [20:39] smoser, you think we should try to add compat. to the ubuntu version or do it in cloud archive? [20:42] adam_g, i dont know. i said that just as an idea [20:42] it would certainly make it easier. [20:43] it seems like for this particular fix that it can't do much harm [20:43] the only thing is getting the device permissions right. [20:43] adam_g, right? [20:43] hallyn_, what do you think about trying to add some compatibility fallback to the qemu package so that the kvm_* module gets modprobe'd by maintainer script if the kernel does not support auto loading it [20:43] as if we jus tmodprobe, will udev hook everythign up correctly perm wise on precise ? [20:44] smoser, i *think* the previous behavior was to have the package modprobe it and udev took care of the ACLs [20:45] i'm sure that i'm going to eat these words, but a commented line that does: [20:47] grep "^kvm_" /proc/modules || { modprobe kvm_intel || modprobe kvm_amd64 || : } >/dev/null 2>&1 [20:47] seems like it can't do much harm [20:48] yz. [20:48] ya... i'd like hallyn_'s input though, since he's more familiar with the packages. i'm not sure where that would even go now. === _thumper_ is now known as thumper [21:02] zul, also, looks like a new version of alembic is required for ceilometer now https://launchpadlibrarian.net/150433252/buildlog_ubuntu-precise-i386.ceilometer_1%3A2013.2%2Bgit201309161630%7Eprecise-0ubuntu1_FAILEDTOBUILD.txt.gz [21:06] adam_g: ack [21:31] I"d rather not make it uglier than it was originally. We can just re-insert whatever was there in quantal or raring [21:31] it'll "fail" in container, but we just || true it (which is why I removed it - it was getting ugly for - at the time -= zero benefit) [21:32] adam_g: if it's still causing you pain, let's just re-add it. Please open a bug against qemu. [21:33] hallyn_, ok [21:35] https://bugs.launchpad.net/cloud-archive/+bug/1218459 [21:35] Launchpad bug 1218459 in qemu "qemu no longer loads kvm_* module on older kernels (12.04)" [Undecided,New] [21:37] hallyn_: hey, could it be that we acquire a full lock when creating a container? apparently creating a container and running lxc-list at the same time makes lxc-list wait for the create to finish, doesn't seem ideal :) [21:43] stgraber: lxc_container_new() sees that create is ongoing and should return false [21:43] so I should expect lxc-list to just ignore it and move along [21:43] I'll have to test with current upstream to confirm, the box I'm on may be on an older version [21:43] now maybe lxc_contaienr_new() is misbehaving... [21:44] stgraber: it doesn't hang for me actually [21:44] stgraber: however, lxc-list is not doing --fancy fo rme [21:44] i have to do lxc-list --fancy :) [21:46] Hello, I am trying to install ubuntu with raid onto a dell 755. With USB, I have tried downloading the 32and 64 bit alternate install disks for 12.04 and I have tried two different USB sticks. (installed with lili)Under all circumstances, after booting to usb, choosing any of the options causes the computer to reset. [21:47] Is lili bad for this and I should try another usb installer? [21:47] adam_g: ok, thanks. [21:53] Is there anything special about calling a script using /etc/network/interface's pre-up vs calling it from command line? [21:54] Oh, NVM. The IP is not known yet. === thumper is now known as thumper-afk [22:08] sona11111111: I believe best these days is to just dd the disk images to a usb stick [22:11] sona11111111, heh, not sure, I have never had an issue with ubuntu and dell 755's (I have >30 of them) [22:12] also, try to use the usb jacks on the bad, and not the front [22:12] back === bigjools_ is now known as bigjools === acrocity_ is now known as acrocity [23:05] adam_g: http://people.canonical.com/~serge/qemu-upstart.debdiff [23:06] hallyn_, sweet === thumper-afk is now known as thumper [23:09] adam_g: will upload once i do a trivial local test that i didn't do something stupid [23:10] adam_g: (if I don't, since i'll be running to dinner soon, pls shout at me tomorrow) [23:10] hallyn_, great. thanks a lot [23:27] I got a 12.04 server, with a RAID 1 on SSDs, that is in need for some TRIM. I need kernel 3.8 for this. Can I run kernel 3.8 with 12.04? Or should I update to the 13.04 (non LTS)-version? [23:31] MarkusDBX: you can install a 3.8 kernel intended for use with 12.04 LTS, see the "Ubuntu kernel 3.8" section here, https://wiki.ubuntu.com/PrecisePangolin/ReleaseNotes/UbuntuDesktop [23:35] sarnold: Interesting. Should I expect as solid performance as with my current 3.2.0-29? [23:36] MarkusDBX: yes, I believe so, though note that the support length for that kernel may be shorter than the kernel you're currently on -- you may need to upgrade that kernel to a newer version when 13.04 leaves maintenance; it'll be replaced with either 13.10 or 14.04 based HWE stack when the time comes... [23:38] I plan to upgrade to kernel 3.8, but to stay on 12.04, until 14.04 is released. Then I will probably just reinstall with that version. [23:47] MarkusDBX: ah, good, that should work fine, see [23:48] MarkusDBX: see "Version Matrix" https://wiki.ubuntu.com/Kernel/Dev/ExtendedStable for details, 3.8.x will EOL in august 2014, that should give you plenty of time to upgrade to 14.04 LTS when the time comes :) [23:49] sarnold: ah great. My plan might work in that case. Thank you