[00:04] . [00:06] Hi all, I'm learning uvtool + ubuntu cloud images for my KVM machine. When creating a new guest with uvt-kvm create it ends up running two disks. Why is that? [00:06] thanks [00:07] inkata: #ubuntu is probably the best place for this. I'd help you, but unfortunately, I'm on a machine with no VM capabilities at the moment, so I can't confirm or deny or test. [00:07] (and I don't know the tools well enough) [00:07] tsimonq2: I think the docs send uvtool users to this channel :) [00:07] sarnold: oh? [00:07] tsimonq2: yeah, see the bottom of https://help.ubuntu.com/lts/serverguide/cloud-images-and-uvtool.html [00:08] then nevermind, stay around and wait for someone to answer, sorry about that inkata :) [00:10] inkata: a quick read through the source gives me the impression that it's probably creating a cloud-init datasource [00:10] inkata: you can find more information on cloud-init here https://cloudinit.readthedocs.io/en/latest/ though I've found it severely underdocumented how you actually -use- cloud-init :/ [00:10] tsimonq2: thanks man, as sarnold mentions, the docs point here. [00:11] sarnold: roger [00:21] sarnold: can't make heads or tails of cloud-init :-/ [00:21] inkata: you either? :( sigh. it looks so useful but I can't ever find anything that says "put this file and run there in order to use it with libvirt" or something similar... [00:22] sarnold: exactly lol [00:23] maybe it's a partition with a magic file on it or something [00:23] inkata: is it still visible in the resulting VMs? that might explain an awful lot about how to use the thing :) [00:25] sarnold: it's a disk, it shows on dmesg but it's not mounted. let me reboot the guest to try and see what's in it and report back [00:25] inkata: Ubuntu cloud images are undifferentiated. The second disk is a cloud-init datasource which differentiates it (eg. has your ssh public key on it). [00:26] It tells your instance what to do when it boots, essentially. [00:28] See http://cloudinit.readthedocs.io/en/latest/topics/examples.html for details of what it can do. With uvtool, you can pass this in using --user-data, to save having to mess with creating the datasource manually. [00:29] rbasak: do I pass a local file to uvtool? a plain file ? [00:32] inkata: you pass anything cloud-init will accept as userdata. That can be a shell script with an appropriate shebang, I believe, but usually it's in cloud-init's cloud-config format, which is yaml and covers pretty much anything you might need in a declarative format. Documentation at http://cloudinit.readthedocs.io/en/latest/topics/format.html [00:32] inkata: re-reading your question, yes - a file. [00:33] rbasak: /me gasps in awe that's quite useful.. let me try with a simple touch /tmp/newfile [00:34] smoser: IMHO, http://cloudinit.readthedocs.io/en/latest/topics/examples.html should have every example start with #cloud-config. Otherwise people in a hurry will miss that, which I think will cause mostly silent failure, right? [00:35] omg yes, if #cloud-config is required for some reason that absolutely should be in every single example :) [00:35] inkata: note that if using uvtool, if you specify --user-data then uvtool can no longer inject your ssh public key, since it uses cloud-init userdata to do that and you're overriding it. [00:35] rbasak, definitely true. [00:36] inkata: so in that case, you should write to ~ubuntu/.ssh/authorized_keys yourself, or make alternative arrangements. [00:36] inkata: if you use cloud-config then it's easier since there's a yaml key you can use to provide your ssh public key. [00:40] rbasak: cloud-init is a regular package I should install in my KVm host or in my regular desktop? [00:40] cloud-config* sorry [00:43] inkata: cloud-config is a format that is defined by cloud-init that it accepts - not a package. [00:44] The format is a YAML schema, essentially. It is straightforward to write and modify by example. [00:44] rbasak: roger that [00:45] inkata: look at http://cloudinit.readthedocs.io/en/latest/topics/examples.html, but stick "#cloud-config" at the top of any file you create so cloud-init knows it's getting cloud-config data. === kklimonda is now known as Guest67032 [01:51] rbasak: Re: #cloud-config file, how can I get the name of the new guest to be used in my file? it used to say the new_guest_name in the login screen, now it only says ubuntu [01:55] inkata: set the hostname and set manage_etc_hosts to localhost. [01:57] rbasak: but how to set the hostname dynamically in the cloud file? I need to catch uvt-kvm create THISHOSTNAME [01:59] inkata: you'll need to wrap uvt-kvm and create a temporary --user-data file. [01:59] inkata: if you override with --user-data, uvt-kvm cannot then do it for you. [02:00] rbasak: got it. This is getting to be more cumbersome than getting a fat server ISO and installing it from scratch lol [02:01] Really? :) [02:02] kinda yeah :-P [02:02] man the only thing less documented than how to use cloud-init is how to use preseeding :) [02:04] rbasak: I'm setting the hostname to blahblah and it works!, but ssh keys are borked.. working on it.. [02:09] Great! [02:09] * rbasak goes to bed now [02:11] rbasak: thanks man! [02:16] rbasak: sarnold it's aliiiiive!!! but needs mooar scripting than I anticipated or need right now. Thanks for your help guys, sleep tight! [02:16] inkata: great! :D [02:17] * inkata goes home === seg_ is now known as seg [04:26] I'm at Stanford making a drone startup with AI (deep learning + SLAM) and I'm looking for people who're great at linux and systems to help out and learn AI. Is anyone in the Bay Area and interested? === _ruben_ is now known as _ruben [09:12] Wanted to upgrade from 14 LTS to 16 LTS, and got an issue with my /boot/grub/menu.slt [09:13] it says the file changed, and the package maintainer's got a new version for this file. [09:13] here is my old menu.lst http://paste.debian.net/hidden/b02dfb5f/ [09:14] This is a server I'm only able to access through ssh so I'm not sure I'm goin to use that grub menu at all. [09:24] wrksx: this could be bug 1485685 if you're using a cloud image? [09:24] bug 1485685 in cloud-init (Ubuntu) "Users are prompted on upgrade of cloud images" [Medium,Triaged] https://launchpad.net/bugs/1485685 [09:28] rbasak, ubottu, how can I confirm I'm using a cloud image? [09:31] wrksx: how do you install the system? [09:32] rbasak, it was installed by my hoster. I just ran an upgrade [09:33] I suppose that mean it uses some kind of image anyway (since the hoster has to deploy a lot). But is that what you refer to as a cloud image I dunno [09:33] If you have an /etc/cloud/build.info then your system is based on a cloud image [09:33] If your hoster installed the system, perhaps they modified menu.lst. [09:34] rbasak, I have no /etc/cloud [09:35] rbasak, is that /boot/grub/menu.lst used to boot the server? [09:35] Then it seems likely to me that your host modified menu.lst. [09:35] Yes, depending on your host environment. [09:37] rbasak, two major diff between the two files, the hoster's one is referencing /initrd.img and /vmlinuz symlinks, and the package maintainer version is directly referncing inside /boot. [09:38] the other diff is that the hoster hass a biosdevname=0 on the kernel lines, which looks like important even if I don't know the purpose of it [09:44] Looking to chat to someone about conjure-up and if it supports spaces defined in MAAS? [09:49] stokachu: ^ [09:50] https://wiki.ubuntu.com/Booting says that "Grub will find the /boot/grub/menu.lst which configures its interactive menu". I don't understand why this should be used to boot a server since no one is there to see the menu at reboot time [09:52] We don't use that any more I don't think. [09:53] It's still commonly used though when a host wants to boot a kernel directly rather than by emulating a regular system boot. [09:53] rbasak, do you know a way to check what is used at boot time? [09:54] No, the guest can be fooled. Ask your host. [09:54] rbasak, oh damn. I was afraid of hearing that [09:54] You could eliminate some methods by confirming they're not configured on the guest I suppose. But I don't really have the time to go through that, sorry. [09:55] It's not clear to me if your menu.lst modifications were made by an Ubuntu installer, by an Ubuntu bug or by your host. [09:55] rbasak, that's alright thanks for the help, always appraciate it [09:55] Look into net.ifnames=0 - you might need this after upgrade to stop your NIC names changing. [09:56] Also the grub-legacy-ec2 package, which your host may (or may not) use to boot your kernel directly. [10:21] Andrew__: it uses juju underneath but there is no way to tell conjure-up to use a certain space [10:24] Ok thanks - was trying to see if I could do all the network definitions in MAAS instead of doing in advanced configuration [10:25] I have raised a issue around not being able to get into advanced configuration for cinder [10:25] Is there a way to target nodes in MAAS from conjure up? [10:26] Andrew__: re: cinder that should be fixed if you want to test ppa:conjure-up/daily-git [10:26] Andrew__: there isn't a way to target specific maas nodes via conjure-up unfortunately [10:27] Andrew__: the best you can do is set which node to bootstrap with using --bootstrap-to === dosaboy_ is now known as dosaboy [10:28] ok [10:39] Okay so I fucked up my system trying to go from LTS 14 to 16 [10:40] I was a bit careless because I thought there was no really important stuff on that box. [10:40] Now it's gone I feel I underestimated the cost of that loss. [10:41] System seemed not to boot [10:41] Using an option from the hoster I've been able to restart in 'rescue' mode [10:44] stokachu - all working in daily - thanks [10:46] stokachu - I am going through and setting all the networks in conjure-up advanced configuration is there any way to share common network configurations? === Ussat is now known as Cthulu-2016 [12:36] these docs seem to reference an installer (openstack-installer) that doesn't exist on 16.04: https://help.ubuntu.com/lts/clouddocs/installer/ [12:37] they're linked-to at the bottom of https://www.ubuntu.com/cloud/openstack [12:47] diddledan: we need to fix that, openstack-installer is 14.04 only, http://conjure-up.io is for Xenial and above [12:49] looks like these different docs (again linked on the openstack page) reference outdated juju commands https://help.ubuntu.com/lts/clouddocs/en/Installing-Juju.html [12:50] specifically `juju generate-config` isn't available on 16.04 from what I can discern [13:08] is it save to delete the content of /var/cache/apt/archies ? [14:01] I've got a execuatable call it `hello` doing ls it says `hello -> user/bin/hello*` , however I can't run it from the terminal by typing `hello` [14:02] oh wait [14:02] are links relative to the folder you `ls` [14:03] yes [14:04] anyway - the current directory isn't normally in PATH, and shouldn't be [14:07] Hi. I'm trying to setup an existing Ubuntu-Server 14.04.5 LTS instance with an attached RAID-1 array. I do this all the time on all sorts of other boxes with no issue. Here, though, try as I might, I can create the array, manually assemble is from shell -- all looks good -- [14:07] BUT, I can't manage to get it to assemble/start on boot. [14:07] AFTER boot, I can simply 'mdadm --assemble --scan', and all's good. But auto-assemble? [14:08] No dice .... Is there something unique, or not quite ready, in Trusty? [14:22] robert17: did you mdadm --detail --scan >> /etc/mdadm/mdadm.conf && update-initramfs -u # ? [14:22] RoyK: Yep. As always .... [14:23] "Quadruple-checked" superblocks, etc etc [14:23] check mdadm.conf if it has some ghosts [14:23] 'ghosts'? [14:23] old references to former arrays of the same name [14:23] Ah. It does not [14:24] No errors in boot log afaict. Just no startup. Kernel mod's loaded. Like I said, after startup, reassemble works perfectly. [14:25] Just a quiet, no-start on boot. [14:29] is the raid in any lvm config or in fstab? [14:32] Nope, not yet. [15:03] RoyK: appears to be unique to Trusty. Mine, anyway. Ubu16 works just fine. [15:32] robert17: I don't think the raid will be activated until it's actually used [15:33] 'used' ? as in ... it has to be mounted? [15:33] Works fine "everywhere else" ... [15:38] as in accessed [15:39] hm. Worth a try ... doing so now. [15:48] RoyK: Completely unneeded on Ubu 16 -- autoassembly on boot works with no access/mount, just mdadm.conf config. OTOH, on Ubu 14, no autoassembly on boot, UNTIL ... create an /etc/fstab mount for the array. Works now. [15:48] IIUC, that's going to be a problem if I want to boot from the array ... [15:49] It's got to assemble in order to GET to the /etc/fstab ... unless there's sufficient info in the initramfs [15:53] robert17: I've used raid1 for boot devices on most versions of debian/ubuntu, and never seen an issue with that [16:08] robert17: As I said, it usually assembles correctly, but isn't activated unless it's being touched by lvm or a filesystem or similar [16:10] RoyK: Do you by anychance need to use "dolvm" or "lvmwait=" in grub cmd line? [16:16] robert17: IIRC I've used lvwait once, but I don't remember the occation [16:35] RoyK: fwiw, setting "GRUB_CMDLINE_LINUX="dolvm lvmwait=/dev/mapper/VG_TEST-LV_TEST", where /dev/VG_TEST/LV_TEST is a VG+LV on PV=/dev/md1 seems to do the trick. No /etc/fstab mount or /etc/rc.local needed. Of course, that still requires that there's a mountable FS somewhere on the array ... [17:17] hello [17:18] does anyone know how to prevent a IF that is configured for DHCP from registering in DNS [17:19] thanks Royk [17:27] ?? === eagleeyes is now known as eagle === eagle is now known as Guest8335 === jelly-home is now known as jelly [18:15] What is needed for the broadcom WiFi to work? [18:24] anyone has compiled apache's mod_wsgi for python3 on xenial? [18:24] the version in the repo has a bug [18:34] madprops: what bug number? [18:35] https://bugzilla.redhat.com/show_bug.cgi?id=1345725 [18:35] bugzilla.redhat.com bug 1345725 in mod_wsgi "python3-mod_wsgi: site.addsitedir() fails if multiple directories in python-path" [Unspecified,New] [18:45] madprops: you could download the patch from the github issue; apt-get source the mod_wsgi package, apt-get build-depends for the mod_wsgi package, and try applying the patch [18:45] looks like our packages are quite a bit older than the patch on github, it might take some massaging, but it's worth a try [18:46] madprops: and if it works for you, you could post a debdiff to the debian bug and ubuntu bug and see if you can get it fixed for everyone else, too :) [19:03] Expansem [19:04] Guest8335: that's not a very good password.. hope it's not for anything important :) [19:05] Sarnold easy to say. [19:07] How can I find a kmod name? [19:08] Bcmwl-kernel-source looked to have built a module. [19:08] It appears it isn't loaded. [19:09] Guest8335: it looks like that generates a file named wl.ko [19:10] Guest8335: try running 'find /lib/ -name wl.ko -ls' and see if that finds anything [19:12] There was b43 it isn't a 43xx chip but it works on fedora. [19:20] sarnold: Whatever it built there's no recent wl .ko showing in /lib/ [19:21] When the package was installed it looked like it built against the running kernel. Running make in use SRC doesn't work. [19:25] Guest8335: how about 'dkms status'? [19:29] Sarnold status? [19:29] It was installed it looks like something to. Do with virtualbox. === dare is now known as udare [19:58] rbasak: hi hi [20:08] . [20:18] coreycb, hey I sycned nova 14.0.1 to proposed [20:18] jamespage, thanks [20:43] Probably a silly question, but upgrading a 14.04.05 MAAS(1.9) that controls openstack nodes to the latest HWE for continued support shouldn't affect anything with MAAS, right? [21:00] bc2946088: what do you mean by to the latest HWE ? [21:19] roaksoax: using the 16.04 Xenial HWE Stack on 14.04 === not_phunyguy is now known as phunyguy === duxklr is now known as jemurray === arooni_team_b is now known as arooni === tomaw is now known as 02HAAAAAB === karstensrage_ is now known as karstensrage === 02HAAAAAB is now known as tomaw === skeezix-hf is now known as Ofir === Ofir is now known as skeezix-hf [22:06] coreycb, release note looks good [22:18] I am trying to set up an openvpn server [22:18] oh it came to me [22:18] <3 [22:45] Hi anyone with experience setting up ubuntu servers on lunanode vps?