[01:56] <coreycb> jamespage, something's causing haproxy to fail to install from trusty-mitaka staging.  http://paste.ubuntu.com/15822925/
[01:57] <coreycb> seems to be related to the new lua5.3, http://launchpadlibrarian.net/253502335/lua5.3_5.3.1-1_5.3.1-1ubuntu1.diff.gz
[03:02] <Capprentice> Hi! Im having problem with the e1000e driver on Ubuntu 14.04 . It crashed under load. I have enabled IP o\dodinge1000e 0000:00:19.0 enp0s25: Reset adapter unexpectedly [  197.382071] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[08:11] <Capprentice> My server has 16 GB RAM, Core i5 CPU Intel, 4x2 TB HDD. What will be the optimal settings for sysctl for Squid Cache
[08:12] <ikonia> it doesn't work like that
[08:12] <ikonia> you need to look at what you're machine is doing, how/what you're using squid,
[08:12] <ikonia> you'll probably see very little benifit from changing sysctl
[09:09] <jamespage> frickler, new ceph rc for today - will also include that maintainer script noise fix
[09:17] <jamespage> coreycb, that change got reverted to lua - tests ok now
[10:47] <codepython777> anyone here?
[10:57] <Capprentice> hi
[11:05] <Seveas> Capprentice: you'll want to maximize your available tcp port range, tune max file descriptors for what's usable, reduce swappiness and maybe tune some tcp parameters for http traffic (lower timeouts, no keepalive etc)
[11:09] <codepython777> does anyone use a PPA here to distibute their software across machines?
[11:17] <Seveas> I do
[11:18] <codepython777> Hi Seveas
[11:18] <codepython777> do you use launchpad to do that? Or can that be done without a launchpad account?
[11:18] <Seveas> launchpad
[11:19] <codepython777> so there is no way to do this without a launchpad account?
[11:19] <codepython777> Also, what is a good resource to follow, to get started ?
[11:19] <codepython777> is there a way to say you only support certain distributions? (like say 15.10)
[11:20] <Seveas> any good debian packaging tutorial will work
[11:20] <Seveas> and you'll need to upload each package to each distribution
[11:21] <Seveas> so only supporting 15.10 == only uploading to 15.10
[11:21] <codepython777> i see
[11:21] <codepython777> so debian uses the same mechanism for packages i guess?
[11:21] <Seveas> a similar mechanism
[11:21] <Seveas> you can build repos without launchpad as well, but you said 'ppa', so I was assuming you meant launchpad's repo functionality
[11:22] <Seveas> if you don't want to use launchpad, you can use reprepro or other tools to take a collection of packages and turn them into a repo.
[11:22] <codepython777> if i build repos without launchpad, is it easy to host it? to let users add it and run updates?
[11:22] <Seveas> you just need a webserver :)
[11:23] <codepython777> that i have :) nginx
[11:23] <codepython777> where are the docs to create the files for distribution? :)
[11:24] <Seveas> man reprepro
[11:24] <Seveas> and a debian packaging tutorial. Not sure where to find one these days
[12:47] <coreycb> ddellav_, nova 12.0.3 is out, want to include that in your SRU?
[12:54] <ddellav_> coreycb sure
[12:54] <coreycb> ddellav_, thanks
[12:55] <vassie> Hello, is it possible to install Xenial without LXD? Personally I have no need for it as I want to use Docker
[12:55] <patdk-wk> I have never installed lxd with xenial
[12:55] <patdk-wk> so it must be
[12:56] <vassie> patdk-wk: I have done a clean install of Xenial beta 2 and LXD is installed by default
[12:56] <patdk-wk> using what install options?
[12:56] <patdk-wk> default, install everything?
[12:56] <sdeziel> vassie: Docker can run inside LXD mind you
[12:57] <rbasak> vassie: it shouldn't cause any harm by being there. If it does, then please file a bug.
[12:57] <rbasak> Can you just remove it?
[12:58] <ogra_> it isnt a recommends
[12:58] <ogra_> so it will pull out the meta package with it
[12:58]  * patdk-wk only ever installs ubuntu-minimal
[12:59] <ifohancroft> hello everyone. where do i find the package for kmod-nls-iso8859-1
[13:00] <ifohancroft> i need it for a server but no internet connection there or anything, i am in recovery mode so i want to download it here and transfer it there via usb key or something and install it manually
[13:06] <patdk-wk> ifohancroft, in the normal kernel package
[13:06] <patdk-wk> linux-image-......
[13:07] <ifohancroft> thanks, patdk-lap. so if i want just the module i pretty much need to reinstall the kernel?
[13:07] <patdk-wk> yes
[13:07] <patdk-wk> it's part of the kernel
[13:08] <patdk-wk> so if it's broken, just boot using a different installed kernel
[13:08] <patdk-wk> you should have 2 atleast, if not more
[13:08] <ifohancroft> i was thinking about that, however in the grub menu i only see one and it's debug version so i guess they may have been wiped out or something or is it just not showing in the options?
[13:09] <patdk-wk> newer grub menu only shows the current one
[13:09] <patdk-wk> there is another option at the end
[13:09] <patdk-wk> like a show all option
[13:09] <patdk-wk> that will show the other versions
[13:09] <ifohancroft> oh, ty, i will look for that
[13:09] <patdk-wk> older grubs, show all of them by default
[13:09] <patdk-wk> or rather, older ubuntu grub config makers :)
[13:20] <sdeziel> vassie: this fresh post is really apropos: https://www.stgraber.org/2016/04/13/lxd-2-0-docker-in-lxd-712/
[13:34] <jamespage> coreycb, just waiting for the magnum binaryies to publish and I;ll promote proposed->updates
[13:35] <coreycb> jamespage, ok gnocchi is on it's way too.. just noticed it has an i386 test failure
[13:36] <jamespage> coreycb, yah
[13:36] <jamespage> I might be tempted to hit the button of despair for that first...
[13:46] <jamespage> coreycb, ok promoting proposed -> updates
[14:00] <coreycb> jamespage,  I think pandas is stuck in proposed due to sunpy autopkgtest regressions, which appear to mostly be due to the version of numpy we have.  numpy 1.11.0 could use a sync but isn't just bug fixes.
[14:01] <coreycb> jamespage, not sure what to do, it's slightly late in the cycle
[14:01] <pmatulis> hallyn: for the cloud-init fix, so nova controllers need to have their package updated?
[14:03] <hallyn> pmatulis: i thought it was just the cloud images on which the instances were based.  smoser ^ ?
[14:04] <smoser> pmatulis, which is this ?
[14:05] <pmatulis> https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1569974
[14:10] <vassie> rbasak: I've tried to remove lxd but it will also remove ubuntu-server
[14:13] <rbasak> stgraber_: ^ perhaps the lxd seed should be a recommends and not a depends?
[14:14] <rbasak> kirkland: ^
[14:15] <vassie> Would be nice if it was an option via tasksel
[14:15] <vassie> I want to use Docker and not LXD
[14:16] <vassie> LXD has created a network interface that I don't need/want
[14:16] <rbasak> You can. Just use Docker.
[14:16] <vassie> rbasak: I know, but it would be nice to remove LXD as they kind of both do the same thing, I only need one, not both
[14:17] <teward> vassie: removing lxd will remove ubuntu-server and i think ubuntu-standard metapackages, which I think are critical to make sure you get the right upgrade paths
[14:17] <rbasak> Ubuntu ships vi and nano. Both do the same thing. I only need one, not both.
[14:17] <teward> s/and ubuntu-standard//
[14:18] <teward> vassie: use Docker, leave lxd alone.  but leaving it installed won't hurt anything
[14:18] <teward> (removing it may remove some of those metapackages which I think are 'good to have' to ensure proper upgrade paths and package set changes)
[14:19] <rbasak> Options via tasksel don't really work any more. An increasing number of Ubuntu server users use cloud images where tasksel isn't an option.
[14:19] <rbasak> (also it's quicker since you don't need to run an installer)
[14:33] <pmatulis> smoser: ?
[14:36] <kirkland> rbasak: sure, I guess
[14:36] <kirkland> rbasak: do you want to take care of that?
[14:42] <rbasak> kirkland: will do, thanks.
[14:44] <saftblandarn_> Hi, I'm having problems to connect with my server with putty. It was working fine until I restarted putty
[14:45] <saftblandarn_> I've tried to restart my server and "sudo service ssh restart"
[14:57] <kirkland> rbasak: actually
[14:57] <kirkland> rbasak: what, if any, reprocussions are there for doing so?
[14:57] <rbasak> kirkland: AFAIK, only that you can remove it without being told it's bad.
[14:57] <kirkland> rbasak: right -- what, if any, images do we build that *don't* install recommends?
[14:58] <kirkland> rbasak: because we really, really do want lxd just about everywhere
[14:58] <kirkland> rbasak: we can even run lxd in lxd
[14:58] <kirkland> rbasak: I'm a little nervous, actually, about doing this today, and seeing images tomorrow, that don't have lxd in them any more
[14:59] <rbasak> kirkland: we can confirm with People Who Know if you prefer. I was taking the approach that doing it sooner rather than later means more time to detect fallout.
[14:59] <rbasak> I'm surprised that lxd isn't seeded on desktop?
[15:00] <rbasak> It's in supported-misc-servers, cloud-image and server.
[15:00] <rbasak> Is smoser around?
[15:00] <stgraber> I just did the seed change based on scrollback
[15:00] <stgraber> snapd is a recommend too and that seems to have appeared everywhere
[15:00] <smoser> rbasak, here.
[15:01] <rbasak> smoser: any impact to dropping lxd to a recommend that you can think of? Since you were looking into this kind of thing.
[15:01] <rbasak> (in the seeds)
[15:01] <stgraber> IIRC the only images that were building without recommends were the phone images (not sure if they still do)
[15:02] <smoser> snappy might build without recommends.
[15:02] <smoser> but it could have lxd explicitly if it needs it. or otherwise handled there.
[15:02] <stgraber> ah could be, but we don't want lxd on those either :)
[15:02] <stgraber> there is a lxd snap for that
[15:02] <smoser> the one difference that i think is notable is that recommends rather than depends means:
[15:02] <smoser>  apt-get --purge lxd
[15:03] <smoser> does *not* tell you: you're getting rid of ubuntu-server
[15:03] <stgraber> yeah, that's precisely what we want to allow
[15:03] <smoser> i dont know.
[15:03] <smoser> do we ?
[15:03] <smoser> i absollutely feel that way for ubuntu-snappy
[15:03] <stgraber> well, that's what rbasak and kirkland were discussing anyway
[15:04] <smoser> well, the ubuntu-meta would need to be re-done also
[15:04] <smoser> updated
[15:04] <stgraber> sure, I just did the seed change for now, then saw the discussion here so paused before uploading a new ubuntu-meta
[15:04] <rbasak> Based on this discussion I think we're good.
[15:04] <smoser> i thin its fine. stgraber you touched both server and cloud-imge ?
[15:05] <stgraber> lxd doesn't do anything on the system until you start to use it, so I'm not sure why you'd remove it, but if allowing folks to remove it make some of those lengthy LP discussions go away, then it doesn't really hurt to let them
[15:05] <stgraber> smoser: yup
[15:05] <kirkland> okay
[15:07] <smoser> i never really knew about (recommends) in a seed before.
[15:08] <smoser> before probably a year ago. i'd might have used it more
[15:08] <stgraber> it's pretty widely used for desktop stuff I think
[15:09] <patdk-wk> not used for stuff like vlan/ifenslave/mdadm?
[15:10] <patdk-wk> though, it has annoyed me I cannot setup a vlan in the installer to get network access lately
[15:11] <sdeziel> bonding and vlan support in the installer would be handy
[17:33] <pulsar12> i am struggling with pppd daemon which is trying to resolve name "pppX" (X = number of the dynamic interface created) everytime a user connects. anyone has an idea about this?
[18:19] <pulsar12> i am struggling with weird issue on pppd daemon: it is trying to resolve name "pppX" (X = number of the dynamic interface created) everytime a user connects. anyone has an idea why this happens?
[18:40] <coreycb> jamespage, the new sahara with api-paste.ini is in mitaka staging now
[20:48] <dorrek> Hi Guys. Im stuck... I run apache2 server, with user dir http_public. I am unable to add alias to /home/user/public_html. Its just showing 404. In whith file i should look? Also, everything works just fine by domain/~user/adress
[21:23] <sarnold> dorrek: what did you try? do you get any errors, warnings, or access entries in the logs when you try to load it?
[21:25] <dorrek> Nvm... Well.. Im tired... I forget to close <VirtualHost>
[21:30] <sarnold> dorrek: aha :)
[21:58] <guampa> hello
[21:59] <guampa> do you know if it's possible to shutdown or reboot LCXs via libvirt?
[21:59] <guampa> I'm failing to get virt-manager GUI or virsh to do that
[22:00] <guampa> *LXCs
[22:01] <sdeziel> guampa: are you using the libvirt-lxc driver? Or were those LXC containers started by something else (LXD|LXC)?
[22:02] <guampa> these were created via virt-manager, so I guess it's libvirt-lxc
[22:03] <guampa> I got these messages from within virsh http://hastebin.com/qukacuzafo.lua
[22:04] <guampa> yes, I see the xml under /etc/libvirt
[22:07] <sdeziel> ugh, a paste service that needs javascript to display ...
[22:07] <guampa> do you prefer another?
[22:08] <guampa> http://hastebin.com/raw/qukacuzafo
[22:08] <sdeziel> guampa: I'm used to paste.ubuntu.com is all
[22:09] <guampa> the raw version shows plain text
[22:09] <sdeziel> guampa: as for your shutdown problem I am afraid I cannot help you much. I'm not familiar with this libvirt driver
[22:10] <guampa> neverming, thanks for taking interest at all
[22:10] <guampa> d*
[22:10] <sdeziel> guampa: https://libvirt.org/drvlxc.html#usageStop ?
[22:10] <guampa> I haven't found much docs on the issue, so I'll keep digging
[22:11] <guampa> yeah, that's what I tried in the paste
[22:11] <guampa> haven't tried with destroy, I'm looking for a graceful shutdown
[22:11] <sdeziel> guampa: you didn't include the virsh invocation so I wasn't sure if you had used the "-c lxc:///" part
[22:12] <guampa> ah, that's because I was in the virsh shell already
[22:13] <sdeziel> guampa: maybe you could strace the container's init and see if it receives the SIGTERM?
[22:13] <guampa> if everything fails I can resort to script it over ssh
[22:13] <guampa> yes that could be useful
[22:14] <sdeziel> guampa: there is also LXD that might interest you. Works really well with containers
[22:14] <guampa> I could try it, but libvirt interests me especially since I can manage both KVMs and LXCs from the same interface
[22:15] <sdeziel> in Ubuntu, that's where the focus is with containers in general. The libvirt-lxc driver receives less attention
[22:15] <guampa> I'm experimenting with two-level virtualization, actually virtualization + conteinarization
[22:16] <sdeziel> then you could really well have libvirt driving your KVM that could run LXD inside
[22:16] <sdeziel> many use it that way
[22:16] <guampa> it's working beautifully, just added DHCP+DDNS and you fire up machines at any level and everything is reachable
[22:16] <guampa> sdeziel: but I would loose managing everything from the same UIs
[22:16] <sdeziel> true
[22:18] <sdeziel> guampa: "shutdown --mode acpi" and the "agent" mode won't work for containers. Those are for virtualized guests only
[22:19] <sdeziel> the other 2 modes should have though. The fact that they don't seems to point at a miss-configured container
[22:19] <guampa> the error "Container does not provide an initctl pipe" suggests a possible solution
[22:20] <sdeziel> guampa: here on a Xenial lxc container: https://paste.ubuntu.com/15839958/
[22:21] <guampa> I see
[22:23] <guampa> all I see in /dev is console  fd  full  log  null  ptmx  pts  random  shm  stderr  stdin  stdout  tty  tty1  urandom  xconsole  zero
[22:24] <sdeziel> what's PID 1 in that container?
[22:25] <guampa> init [2]
[22:26] <guampa> there's a /run/initctl
[22:26] <sdeziel> in the container, if you send  a SIGTERM to init, what happens?
[22:26] <guampa> nothing
[22:27] <sdeziel> can you try symlinking /run/initctl with /dev/initctl
[22:28] <guampa> just what I was trying now
[22:29] <guampa> yesssssssssss
[22:29] <guampa> does work
[22:29] <guampa> :)
[22:31] <guampa> I'll investigate why it isn't being linked from starters, I see the KVM that hosts it hasn't got the link either
[22:31] <guampa> the KVM has acpi though, so it's able to reboot and shutdown
[22:57] <pulsar12> i still have the problem with the pppd trying to resolve name "pppX"(X being the number of dynamic interface created), every time a user tries to connect. I dont find any explanation for this behavior and havent found anything on google
[22:58] <sarnold> pulsar12: hey, bummer you're still fighting that.. I got to wondering last night, why not just add a billion entries to /etc/hosts for that? set ppp0 to 127.1.0.1, ppp1 to 127.1.0.2, and so on..
[22:59] <pulsar12> hi sarnold :). it would solve it but that would be a hack, and i prefer going to the elegant solution, and also im curious why this happens
[23:00] <sarnold> pulsar12: yeah, I can understand that, but sometimes you just need things to work,hehe :)
[23:01] <pulsar12> maybe next step will be using gdb to know exactly what is happening
[23:01] <pulsar12> it will be the first time i use it, i have to learn