[01:04] hi, when using admad's I18nRoute together with LocaleSelectorFilter, the locale gets reset by the dispatcher. any hints to avoid this? [01:11] i'll just copy and modify it [03:34] Did the interfaces file change from 14.04 to 16.04? [03:36] not drastically [03:36] I can't recall any changes, but that doesn't mean there weren't any.. [03:42] Polarcraft: anything specific you're after? [03:43] even for a more convoluted setup with bonds and bridges it actually seems exactly the same to me [03:43] certainly the same fort configuring a bunch of interfaces with a few pre-/post up/down commands [03:44] drab, well the /etc/network/interface file doesn't contain the nic generated. [03:44] Polarcraft: "the nic generated"? [03:44] you mean from install process? [03:45] is this ubuntu server 16.04 ibnstalled from server iso install media? [03:55] drab, let me rephrase that, when the box was installed by the server provider the /etc/network/interfaces file did not contain the nic "ens3". When on 14.04 it would have contained that nic in the /etc/network/interfaces. [03:58] Polarcraft: I see, so that problem is less about /etc/network/interfaces having changed and more about maybe the installer doing something different, altho that also seems unlikely, afaicremember my installs worked the same [03:58] Polarcraft: do you have anything in /etc/network/interfaces.d/ ? [03:58] that's the only change I've seen with newer lxc/cloud images [03:58] No that is blank. [03:58] I have no idea why this would have been changed though. [03:59] ok, then I got no clue, the installer should really have put it there, but again I think that's the problem, not a format change [03:59] I don't think it is, I'm more inclined toward an install bug of some kind specific to the interfaces. for example I've had all kinds of grief on these new servers with systemd "predictable" naming [04:00] I had to disable that for the installer to even work [04:00] Well I don't think I can change that as the host is sadly OVH. [04:02] I need to bridge the nic from "ens3" to a new one by the name of "ens3:0", but the way I used to do it I can't really do in this situation. [04:03] Polarcraft: is it possible your host set the system up to use systemd-networkd by default instead of ifupdown? [04:04] Well I don't know how OVH does it for 16.04 as I am still using 14.04 for my machine. [04:07] tarpman, they aren't doing systemd-networkd [04:27] tarpman, you learn something new everyday. Doing what I used to do worked perfectly. [05:29] EHLO! [05:29] Question [05:30] Does 16.04 Server autoupdate? [05:31] johefernan: If you tell it to during installation or afterwards [05:35] Dang, didn't remember [05:38] johefernan: grep "APT::Periodic::" /etc/apt/apt.conf.d/* [06:09] Good morning [10:05] smb: any opinion on how to ship Zesty with a working tor? [10:05] Disable apparmor? Or release note that apparmor needs disabling manually or something? [10:06] rbasak, I have not looked but the reverts I documented were in x an y. zesty may or may not be different [10:06] Ah [10:06] I'll check Zesty then. Thanks! [11:14] hi guys, what are the needed repo's for server deployment using your own internal mirror ? [11:14] do I need multiverse for an example ? [11:14] or universe ? [11:25] hi there [11:26] rizonz: you don't "need" multiverse unless you want to install something depending on it [11:26] i have to create a list of all packages together with their licenses (short) and the copytight file is not machine readable :( [11:26] rizonz: you should be able to live without for most regular cases [11:26] is there a way to get the licenses for each package installed? [11:28] rizonz: you can also go without universe IIRC, but that obviously limits you further [11:30] rizonz: why not go on one of your systems and apt-cache policy '*' and some awk/grep, you'll know where things come from [11:30] maybe there is an easier way to get that thou - not sure [11:31] Silicium: isn't that just what you look for http://askubuntu.com/questions/247757/how-do-you-find-the-licenses-for-everything-installed-on-your-system ? [11:32] i need the license [11:34] Silicium: which is in each of those files [11:35] Silicium: a few years there was an effort to make those somewhat machine readable which is mostly true now [11:35] Silicium: if you only need a shortname grep for it [11:35] cpaelzer: srsly? [11:35] i should grep for values i dont know? [11:36] Silicium: you should grep for the "License:" prefix they all have [11:36] no ubuntu 1404 the copyright files are not machine readable [11:41] Silicium: as I said - not all are [11:41] Silicium: but this gets you a good head start [11:41] Silicium: for i in /usr/share/doc/*; do printf "\n%20s: " "$(basename $i)"; awk '/^License:/ {print $2}' ${i}/copyright | sort | uniq | xargs ; done [11:43] cpaelzer: yeah I'm not sure [11:44] I canot find a decent mirrorlist which works [11:48] rizonz: my zesty lxc guest is fine with main only it seems [11:48] rizonz: apt-cache policy $(dpkg -l | awk '/^ii/ {gsub(":.*",""); print $2}' | xargs) | grep archive | uniq -c [11:48] 469 main no others [11:50] rizonz: and I can enable disable universe/multiverse without and effect other than restricting my potential packages that I can access [11:52] cpaelzer: but that cache is good enough, also when you reboot it ? [11:56] rizonz: sure, what fails in your cae and I can check? [11:57] rizonz: just upgraded two packages and it worked fine with the reduced pockets [11:57] cpaelzer: but you use a proxy then in your apt ? [11:57] of the clients [11:58] rizonz: I only wanted to confirm that a system can live fine with the reduced set, I haven't bothered to actually mirror something [11:59] rizonz: but IMHO it shouldn't mind where it comes from [11:59] rizonz: maybe there is the difference, my system only "asks" for main now and is working [12:00] rizonz: are you trying to set up a reduced mirror but the systems are unmodified and ask for main, universe, ... ? [12:00] rizonz: and that fetch fails then? [12:01] cpaelzer: mhh [12:01] I want to see what is in multiverse and universe, just some simple list [12:12] rizonz: you can tweak sometihng like this to your needs http://paste.ubuntu.com/24267442/ [12:18] cpaelzer: thannk, but there is no online reference ? [12:18] rizonz: you'd think there is one - and there might be but I don't know [12:19] rizonz: http://paste.ubuntu.com/24267506/ [12:19] rizonz: well, if you mean an online reference what a package is in in general (and not something specific to your system) that might be easy as I'm sure this is on LP somewhere [12:20] rizonz: https://launchpad.net/ubuntu/+source/ [12:20] cpaelzer: yeah will check out [12:20] rizonz: you will see the pocket per release in the table there [12:20] it's dofficult to decide what to mirror [12:21] at least I know I don't need i386 anymore :) [13:09] cpaelzer: I'm not sure the architecture selected should be the host architecture. It should be the architecture of the image selected. [13:09] (uvtool) [13:09] rbasak: does anybody ever really use it cross arch? [13:09] ok, then the image arch - fine with me [13:10] cpaelzer: since qemu can do cross arch I'm not sure we should rule it out [13:10] cpaelzer: if it's not too difficult I think? [13:10] rbasak: looking into it [13:10] cpaelzer: if --backing-image-file is ued, then we'd need a --image-arch option that could default to the host arch. [13:12] rbasak: well I'm only changing the default path and provide the templates - IFF it is overly complex for what we gain we might keep it as I proposed [13:12] rbasak: but let me look at it first [13:17] rbasak: ok looked at it, not that it would be impossible, but I really think that would make it more complex and error prone [13:17] rbasak: so far most handling on filters is kept to simplestreams for a reason [13:18] rbasak: to pick the arch from the image would need to understand those plus the case on backing image you mentioned [13:18] rbasak: I think even for cross arch usage the fact that there now willa ctually be arch-templates will help [13:18] rbasak: and one doing that might also add a --template command [13:18] cpaelzer: not really understanding the filters. Just picking up the arch metadata once an image is selected. [13:19] cpaelzer: how about as a compromise having an --image-arch option which defaults to the platform arch? [13:19] cpaelzer: and always use that instead of inspecting the image metadata. [13:19] cpaelzer: then at least it's overridable and separable internally from the platform arch. [13:19] s/platform/host/ [13:20] rbasak: yeah that sounds fine to me [13:20] rbasak: let me write something for that [13:20] Thanks [13:43] rbasak: pushed with the change [13:43] rbasak: I still would think I'd always just use --template if I need to override instead of --image-arch, but I'm fine to hear other peoples need [13:44] rbasak: so in terms of simplicity and less code I'd still vouch for my MR before this commit, but follow your decision once you can take a look at it === tekku is now known as tekk [13:53] zul: I expect you to hold me to account for completely disabling the unit tests for vmware-nsx - I raised https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1676712 to cover it [13:53] Launchpad bug 1676712 in vmware-nsx (Ubuntu) "random sqlite related failure during unit testing" [Medium,Triaged] [13:54] Ok, Xeon E5 running at 75-80 all the time, bad? Google tells me the maximum running temp is 73c [13:55] jamespage: heh [14:32] how can I easily find the dependencies on a package I installed? [14:43] cpaelzer: thanks for taking bug 1663671! The existing tests are overly complex IMHO. Feel free to cut them down massively. [14:43] bug 1663671 in squid3 (Ubuntu) "dep8 tests fail on infrastructure" [Medium,Triaged] https://launchpad.net/bugs/1663671 [14:56] rbasak: yeah I wanted to take a look to get it working there [14:56] rbasak: thanks for letting me know that dropping some might be acceptable [14:57] rbasak: I looked at another issue only to realize you posted the update I wanted to write, so I was assinging myself here just to take a look :-) [14:57] guys do we need universe/multiverse for webserver, nodeJS deployments and such ? [15:06] hi [15:07] how do ppl do failover for a nas hosting shared homedir? or you just dont' and use some other FS? [15:07] rizonz: FYI nodejs is universe [15:07] nodejs package [15:08] given that both nas boxes r on zfs I was thinking I could minimize the problem by taking a snap, moving that over to the other box, restore, then stop the first one, snap/restore one last time and move back to old primary [15:08] but it involves downtime during last sync, which that said should be pretty small [15:08] (writes volume isn't crazy) [15:10] cpaelzer: and multiverse ? [15:10] is there a lot of crap we mostly use ? [15:10] cpaelzer or maybe nacc: triaging bug 1676263. Looks like a regular HWE-related breakage. Do you know how we handle these right now? [15:10] bug 1676263 in iscsitarget (Ubuntu) "iscsitarget-dkms 1.4.20.3+svn502-2ubuntu4: iscsitarget kernel module failed to build" [Undecided,New] https://launchpad.net/bugs/1676263 [15:10] rizonz: as discussed earlier today that depends on tha packages you need [15:10] rbasak: there is a master bug to dup it to [15:11] cpaelzer: I know but I'm trying to catorize it some [15:12] rbasak: I don't have the number without searching, but nacc will have it available I hope [15:13] bug 1668808 perhaps? [15:13] bug 1668808 in iscsitarget (Ubuntu Xenial) "iscsitarget-dkms 1.4.20.3+svn502-2ubuntu4: iscsitarget kernel module failed to build [error: field ‘rx_hash’ has incomplete type]" [Undecided,Fix committed] https://launchpad.net/bugs/1668808 [15:14] There seem to be a number of different "master bugs" so I'm not entirely confident. I'll wait for nacc. [15:15] Thanks cpaelzer. Good to know there is an answer :) [15:16] rbasak: one sec [15:17] for trusty failures with 14.04.5: LP: #1612627 [15:17] Launchpad bug 1612627 in iscsitarget (Ubuntu Trusty) "iscsitarget-dkms 1.4.20.3+svn499-0ubuntu2.1 fails to build on linux-generic-lts-xenial kernel " [High,Fix committed] https://launchpad.net/bugs/1612627 [15:17] for xenial failures with 16.04.2: LP: #1668808 [15:17] Launchpad bug 1668808 in iscsitarget (Ubuntu Xenial) "iscsitarget-dkms 1.4.20.3+svn502-2ubuntu4: iscsitarget kernel module failed to build [error: field ‘rx_hash’ has incomplete type]" [Undecided,Fix committed] https://launchpad.net/bugs/1668808 [15:17] rbasak: cpaelzer --^ [15:18] nacc: thanks! [15:18] nacc, powersj, cpaelzer: I'm going to tag those master bugs "triage" to help triagers find them quickly. [15:19] rbasak: good idea, we should make our wiki state (and link) those and similar hints we develop for triaging over time [15:19] it might even do so already and I didn't check [15:24] jgrimm: [15:24] woops.. jgrimm: https://code.launchpad.net/~powersj/ubuntu-cdimage/server-zesty-ppc64el/+merge/321187 [15:25] * jgrimm looks [15:26] mhh it takes forever for apt-mirror to finish, I only download 64 bits 16.04, is apt-mirror that slow these days ? [15:28] powersj, cool, thanks [15:32] hi - question... I am migrating from vmware esxi to openstack, but that is not my question.... my question is that I have a raid 5 (3 hdd) and I am wondering if when I mount them I need to mount them all or only the one that is recognized? === akaWolf1 is now known as akaWolf [15:32] blacknred0: you don't mount individual disks in a raid5 [15:32] I guess that my confusion is that one of the drives show as vmware vmfs and another as linux lvm [15:33] nacc: do i use an utility like mdadm? [15:33] blacknred0: so it's not recognizing your raid5 at all? [15:33] blacknred0: how was your raid5 configured before? [15:33] nacc: doesn't seem like, no [15:34] You dont want to "migrate" your raid5, you want to recreate it in openstack [15:35] nacc: when i configured (which it was a long time ago and my memory doesn't fail me).... I literally let the dell utility handle it and when I boot esxi, I think it recognize the disk as one [15:35] Ussat: but doing that means that I will look my vdmk's, right? [15:35] look ? [15:36] oops... I will lose* [15:36] well, ya.....I mean, you are migrating them , right ? [15:37] Ussat: well... yes. I want to move away from esxi to openstack ... and my thought was that I could mount the raid and have openstack point to the vdmk's [15:38] The best way to do it is to create what you want in openstack and then migrate the data [15:38] but, if I am following your logic... you are saying that is not that simple [15:38] no, it is not, raid is never simple [15:38] reacreate what you want in openstack, raid5 and all, then migrate the data [15:38] Ussat: ok... you gave me something to think about [15:39] and have backups [15:39] I think that I will probably have to add more drives and create the data using openstack and then migrate from one drive block to the other [15:40] thanks nacc and Ussat ! [15:40] add more drives to the current raid ? no, just rea create your setup in openstack, then migrate the data [15:42] Ussat: but to create the setup I would need to mount the drive or drives, right? [15:42] or are you saying to move the vdmk's to an external hdd and configure the raid using openstack and then move the data back? [15:54] smoser: I'm trying an install of subiquity and getting cloud-init started and applying a default config when the installer boots, but once I reboot onto the final system (cloud-init is installed, and only has /var/lib/cloud/seed/user-data), cloud-init apparently doesn't run to apply my username/password and hostname for the installed system [15:54] any way I can debug this and figure out why it's not starting or not applying the seed? [15:54] rharper: ^ [15:55] cyphermox: zesty? you really want to populate /var/lib/cloud/seed/nocloud-net/{user-data,meta-data,network-config} [15:56] in /run/cloud-init/ds-identify.log will likely show that it didn't detect any user-data and the zesty behavior is to not run cloud-init; [15:57] blacknred0, I am saying create a brand spanking new openstack with your NEW raid5 avd copy the data over to it [16:01] o/ [16:01] smoser, wrong channel [16:01] \o/ [16:02] cyphermox, rharper is right. you have to provide meta-data and user-data for the nocloud seed to be considered valid. [16:03] rharper: yes, zesty [16:03] Ussat: got it :P .... thanks mate [16:03] there is a meta-data too, but not network-config AFAIK [16:03] the intent of network-config was for it to be disabled since it was version 1; maybe this has changed now? [16:03] network-config is optional; but you want a 'nocloud-net' dir under and put the files there [16:04] rharper: yeah, that exists [16:04] nocloud-(something), I didn't really notice [16:04] I'm about to boot the VM elsewhere to do look at the directory tree again [16:04] ok, well then a tar of /run/cloud-init /var/log/cloud-init* and /var/lib/cloud* and I can help debug [16:05] w.r.t network-config; you can't disable since you're rendering netplan IIUC [16:05] err, _can_ [16:30] rbasak: FYI http works on localhost now, ssl still fails but I'll no more get to that this evening [16:30] rbasak: on squid tests I mean [16:31] \o/ thanks! [19:25] hi [19:25] I need to make a cluster with an apps which was not design to be in cluster. Is it possible to make a cluster with it? [19:25] i use haproxy and keepalived [19:32] no [19:33] it's very unlikely you can cluster something that was not designed to cluster [19:33] you can use haproxy and keepalived, but that will NOT make a cluster, it will just make failover [19:33] if it was clustered, you wouldn't need keepalived or haproxy [19:33] I need failover at first place [19:33] but what is mandatory for failover? [19:34] dunno, that depends on you and your app [19:34] I've two nodes, and I need the unicity of writing to rabbitmq queue shared between two instance of the same binary [19:34] it could be as simple as just moving an ip [19:35] it could be as complex as moving filesystems, applications, vm's, ip's, nic ports, .... [19:35] what is a unicity? [19:35] I've been sais that my apps need to be "reentrant"? [19:37] the uniqueness of data written on disk [19:46] hello! In previous versions of Ubuntu, I had to delete /etc/udev/rules.d/70-persistent-net.rules when creating templates for VMs. Is this no longer necessary because of systemd/udev? The file does not exist. [19:47] right, it doesnt seem to be used anymore, although you can probably create the file manually === Isla_de_Muerte is now known as NwS [20:59] is it recommended to create virtual hosts on a regular users home directory or in the var/www directory? [21:32] damn I need to findout how large the 16.04 repo(s) are [21:34] rizonz: it's an upper limit of 1.1 TB, that's what mine covers for 12.04, 14.04, 16.04, 16.10, and devel [21:35] sarnold: ok, but what do you mirror then ? [21:35] everything [21:35] updates, security as well ? [21:35] 1.1Tb in total ? [21:36] I skip security, those are migrated to -updates every few hours [21:36] ok [21:36] so I should skip them as well [21:36] does it actually use security ? [21:36] on an apt-get upgrade ? [21:37] well, I should have said that I skip security.ubuntu.com; the -security pocket is on the main mirrors too [21:38] yes ok [21:38] here's where I started: https://wiki.ubuntu.com/Mirrors/Scripts -- if you strictly want only 16.04 there are probably better ways to get there; maybe debmirror or aptly [21:38] I do an apt-mirror [21:41] hrm, I don't see any references to the by-hash infrastructure in that script. it otherwise looks nice. [21:46] rizonz: du -hsc /var/spool/apt-mirror/mirror/us.archive.ubuntu.com/ -> 348G [21:47] that's for trusty and xenial, only main sources [21:47] both amd64 and i386 [21:47] drab: ok, but we need universe and multiverse as well I gues [21:47] *guess [21:48] yeah, they ar in there, I do only main for -src [21:48] but all for binary [21:48] actually,I'm forgetful it seems, we do -src for all too [21:49] ok, I do this atm: https://pastebin.com/qUgnH8KL [21:50] rizonz: http://dpaste.com/29N4ED7 [21:50] plus the src line above [21:51] drab: and that size is 384G ? [21:51] *348 [21:52] 360G, I du'ed the old mirror at first, the berkeley one we're using now is 360G [21:52] why is mine 500G and growing ? [21:52] do you run clean? [21:52] I'm using apt-mirror [21:52] not yet [21:52] and I hvae a clean job at the end [21:52] it's downloading [21:52] free's quite a bit each time [21:52] oh, I see [21:53] but does it put all first in skel ? [21:53] * rizonz forgot [21:53] I got not clue about that, been running this daily at 4am for the past year, don't think I've ever seen it go up to 500GB [21:53] rizonz: not ime [21:53] like it doesn't go up and then come down if that's what you're asking [21:54] mhh this is weird [21:55] my lines are not weird are they ? [21:55] ~ [21:55] :P [21:56] I don't think I've seen deb-amd64 before; I'm more familiar with deb [arch=amd64] [21:56] sarnold: that's what I'm using with apt-mirror, works fine [21:56] I think they are equivalent [21:56] okay [21:56] yeah me the same [21:56] rizonz: don't see anything wrong with your lines, they match my paste [21:57] compdoc, ah sorry I didn't know you replied. Thanks for the information! [21:57] oh, wait [21:58] oh, nm, that was accounted for too [21:58] I just noticed yuo had the installer lines too [22:00] drab: yap I need pxe boot [22:01] is there a faster way for du -h ? [22:02] not easily [22:02] df -h is way faster; if you have it on its own filesystem it'll be close [22:02] yeah true but centos is there as well [22:02] 589G [22:02] and growing [22:06] sarnold: is archive not bigger then normal ? [22:07] rizonz: every time I look at mine, it's somewhere between 1.01 and 1.11 TB [22:07] sarnold: for both of them ? [22:07] "both"? [22:08] yes you said precise and xenial [22:08] ah. I've got precise, trusty, xenial, yakkety, and devel on this [22:11] yes so I don't get it why mine is that large [22:14] sarnold: I'm looking for a clue [22:15] rizonz: makes sense; I just don't know what would make sense for a partial copy [22:15] sarnold: do you apt-mirror it or rsync ? [22:16] rsync [22:16] ok [22:16] sarnold: which script ? [22:17] I think apt-mirror is getting shit in [22:17] rizonz: I started with the bits on that wiki page ^^^ and modified for my local environment [22:19] sarnold: why do we use archive and not release ? [22:20] rizonz: for my use I wanted all ubuntu packages unpacked on my local system [22:21] including all updates [22:21] the question is, what should we use [22:23] sarnold: mhh ok, can you give an example of your script, I need to look as I don't see exactly how they exclude in a decent way [22:24] rizonz: http://paste.ubuntu.com/24270684/ [22:26] sarnold: but there you don't exclude security ? [22:26] rizonz: sorry, I mispoked earlier; I just don't bother to mirror security.ubuntu.com [22:27] sarnold: oh indeed, it's a different URL [23:00] sarnold: you use ubimirror or debmirror ? [23:04] rizonz: two-stage rsync: http://paste.ubuntu.com/24270684/ [23:06] sarnold: why that ? [23:06] rizonz: because it's the easiest way to get everything quickly [23:06] sarnold: ok [23:06] oh, otherwise it delete-after in the run ? [23:07] I wonder, how do I make sure I don't have crap but het the installer as well [23:08] oh men this sucks :P [23:08] centos and fedora were easier [23:15] sarnold: mhh rsync is actually kinda useless between i386 and amd64 [23:15] ? [23:15] like I said; I wanted _everything_ [23:15] yes true, do you still use 32bits then ? [23:15] if that isn't a good description of your problem, maybe rsync isn't the best tool :) [23:16] it is but not for Ubuntu repo :P [23:16] no, but my users do :) [23:16] so I wanted it easy available to unpack [23:16] get some other users [23:16] why do they use 32bits ?