[05:32] <cpaelzer> good morning
[06:06] <lordievader> Good morning
[07:37] <UDworker> Hi, anyone can install docker-ce on ubuntu 18.04 ?
[07:37] <UDworker> from deb [arch=amd64] https://download.docker.com/linux/ubuntu bionic stable
[07:39] <UDworker> I had the answer on the other channel, the package is not stable :(
[10:59] <goodi___> I'm looking into user Subiquity via PXE and pressed - can anoyone point me to the variable names it accepts? e.g. for 16.04 it wsa url=http://somehost/preseed.cfg
[10:59] <goodi___> preseed ;-) not pressed
[11:22] <cpaelzer> goodi___: I thought it is not meant to be used hat way (old preseed still works)
[11:22] <cpaelzer> goodi___: but you can wrap the answers
[11:22] <cpaelzer> see https://ubuntuforums.org/showthread.php?t=2390710
[11:24] <goodi___> cpaelzer: how do I wrap the answers? I found the https://github.com/CanonicalLtd/subiquity/blob/master/examples/answers.yaml and it looks so nice and clean vs the current preseed.cfg I use. How 'should' it be used? or is there now a better way to fully remote install machines>?
[11:37] <cpaelzer> goodi___: #1 you can still use preseeds, #2 the link I had describes how to repack the squashfs (if you want to change the subiquity answers) #3 depending what you want to remote install maas/juju can help you
[11:38] <cpaelzer> goodi___: thanks that found answers.yaml nice and clean (were design points of subiquity), but IIRC remote/preconf install was not
[11:38] <cpaelzer> goodi___: OTOH you have to understand that this just generates a curtin yaml
[11:38] <cpaelzer> which is what subiquity uses, but also what maas uses
[11:38] <cpaelzer> so we circle back to either hacking something, or understanding the triplet of curtin/maas/cloud-init for this use case
[11:39] <cpaelzer> goodi___: some pointers https://readthedocs.org/projects/curtin/ http://cloudinit.readthedocs.io/en/latest/ https://maas.io/
[11:40] <cpaelzer> smoser: and cyphermox might have better pointers if you really want to go this way
[11:40] <cpaelzer> cyphermox: was subiquity-preseed every a thing and I missed it?
[11:40] <cpaelzer> s/every/ever/
[11:46] <goodi___> cpaelzer: thank you very much, missed the first link
[11:49] <goodi___> cpaelzer: so I read it like: build your own image that includes the answer.yml I need, and then do most of the installation via curtin/cloud-init if needed?
[11:49] <goodi___> does that make roughly sense?
[11:49] <cpaelzer> goodi___: if you insist on subiquity answers yes, but I think that is the wrong approach
[11:50] <cpaelzer> I can't decide which is the right approach for you, but subiquity answers only gets you subiquity loaded to generate a curtin yaml for you
[11:50] <cpaelzer> so I wonder IFF you really want to go that way, why not provide a curtin.yaml right away
[11:51] <cpaelzer> in all of this cloud-init is what you'd use initially to get the target doing what you want
[11:51] <goodi___> cpaelzer: I'm open for any solution that is simpler/fully unattended
[11:51] <cpaelzer> if you do that via pre-baked custom images or via external datasource is up to you
[11:51] <cpaelzer> hmm - simpler ...
[11:52] <goodi___> cpaelzer: prefered external datasource, but can I use subiquity and only provide a curtin yaml and get if unattended installed?
[11:52] <cpaelzer> honestly, unless cyphermox / smoser beat me with better answers I think for "simpler" you could just stick with classic preseeds for now
[11:52] <Gargoyle> Hi All. I'm trying to get the MOTD section which shows how many packages can be updated (and which are security updates) from a 14.04 system. However it looks like /etc/update-motd.d/90-updates-available is no longer there. Just 91-release-upgrade. Is there any way to manually fetch this info (Need to produce a report for updates required on a bunch of servers). ?
[11:52] <cpaelzer> Gargoyle: does "apt list --upgradable" work on 14.04 already ?
[11:53] <Gargoyle> yeah.
[11:53] <goodi___> cpaelzer: oki, will give it a try; the new one looks just so much nicer; esp. using yml over the current config files it a way nicer IMHO; I just don't really understand how to provide my own datasource/how it should look and if it's even possible todo unattended - but will dig
[11:54] <cpaelzer> goodi___: mass uses cloud-init/curtin to do its installs, so feel free to read through the latter and then see maas for one example how to integrate it
[11:54] <goodi___> I think I'm just missing the picture how Subiquity talks/work with curtin and then cloud-init ;-)
[11:54] <Gargoyle> cpaelzer: So I can see all the packages that need an upgrade, but the count (especially the one that separated security updates) was just a nice short summary
[11:54] <goodi___> cpaelzer: I tried maas a bit, but I always run into issues regarding storage detection if it's not typical server hardware/without ipmi etc ;-)
[11:55] <goodi___> but good point - will look into it
[11:55] <cpaelzer> Gargoyle: I still have /etc/update-motd.d/90-updates-available from update-notifier-common btw
[11:56] <Gargoyle> on 14.04 ?
[11:56] <cpaelzer> and (the last generated data) is in /var/lib/update-notifier/updates-available
[11:56] <cpaelzer> Gargoyle: yes
[11:56] <cpaelzer> well a contianer of 14.04 not updated for a few weeks
[11:56] <cpaelzer> let me update if it is dropped
[11:57] <Gargoyle> hmmm. guess these servers are not vanilla then. (They existed before "my time" here)
[11:57] <cpaelzer> updated, not broken by an update
[12:01] <Gargoyle> Thanks cpaelzer. My /var/lib/update-notifier/ dir is empty. Don't really want to install anything on these servers to get a report. They are in production.
[12:01] <Gargoyle> Most likely we'll provision 18.04 replacements and migrate the app across.
[12:24] <Gargoyle> Lets say I "apt upgrade" my staging server on a Monday, and deploy the app. On Wednesday QA come back with the thumbs up for deployment. What's the best strategy to make sure that when I run "apt upgrade" on production on Wednesday I get exactly the same packages I got on staging on the previous monday?
[12:54] <cyphermox> cpaelzer: what is this about?
[12:55] <cyphermox> cpaelzer: I think subiquity has some form of a preseed, but I'm not sure of the details -- I think it was more just some kind of logic somewhere so it could be automatically tested, not to do specific installs
[12:55] <cyphermox> cpaelzer: what I was told some time ago was that the preseeding story for subiquity was to use MaaS
[12:56] <cyphermox> (but I think we still want to properly support preseeding somehow in subiquity, I just don't know if that's been done yet
[12:59] <Gargoyle> I've found dpkg --get-selections based answers, but that doesn't seem to include package versions - just the names, so I assume any resulting target server could be a few minor revisions ahead on some packages depending on the time between getting the list from one server and installing it on another.
[13:04] <cpaelzer> cyphermox: thanks, you confirmed my assumptions - and for official subiquity preseeding at least I haven't heard of it
[13:20] <samba35> i have configure tun/tap interface with ip command and tun/tap interface configure with openvsitch also  (ifconfig also show tun/tap interface) but i could not see tuntap interface with virt- manager on ubuntu 18.04 ,can some one please give some idea
[13:21] <samba35> i want to use that interface for internet (can i ?)
[13:21] <samba35> sorry exploreing  new thing
[13:24] <rbasak> ahasenack: when you get in, would you like a HO to sync on nvdimm? I think it might be easier/quicker to discuss the outstanding points in person.
[13:24] <rbasak> s/person/realtime/
[13:27] <cpaelzer> samba35: to not seen it is a known issue
[13:28] <cpaelzer> samba35: libvirt can only manage devices that are handled through old ifupdown, but Ubuntu for a while migrated to networkd/NetrowkManager
[13:28] <cpaelzer> samba35: you'd not extra add a tun/tap to get the guest internet access
[13:28] <cpaelzer> I mean usually
[13:28] <cpaelzer> I haven't done so for years
[13:29] <cpaelzer> even in virt-manager just add an interface map it to the default net and that is it
[13:29] <cpaelzer> all else will be done by libvirt for you
[13:29] <samba35> cpaelzer, thank you .understood
[13:29] <cpaelzer> sorry I mostly use libvirt through xmls/virsh so I can't lead you through virt-manager UI for that
[13:30] <samba35> reaily i need to learn more cmd base configuration part
[13:30] <samba35> but i stuck with passing parameter to guest or adding it to guest
[13:31] <cpaelzer> worst case that libvirt doesn't support just one extra arg you can add it there as raw qemu arg still
[13:31] <cpaelzer> https://libvirt.org/formatdomain.html
[13:31] <cpaelzer> and for all it supports you become agnostic to changing qemu versions and such
[13:31] <cpaelzer> as it will abstract it for you
[13:32] <samba35> ok ,thanks again
[13:35] <samba35> cpaelzer, can you please tell me after creating vm with virt-manager can i make changes /add NIC or VGA device to esixting guest with virsh ?
[13:35] <ahasenack> rbasak: hi
[13:35] <ahasenack> rbasak: I was thinking about making an MP this time
[13:36] <ahasenack> rbasak: base branch would be what you reviewed, and new branch would be my changes as a response
[13:36] <cpaelzer> samba35: virsh edit <guestname>
[13:36] <cpaelzer> samba35: look at the definition of one interface and you'll understand
[13:36] <cpaelzer> samba35: use the link I pasted above to understand what is not obvious
[13:37] <samba35> ok ,i tryed with edit part but it failed to boot guest (may be wrong fomation/editing )
[13:43] <rbasak> ahasenack: I was under the impression I still owed you some responses?
[13:46] <samba35> still i am not able to understand why ubuntu 18.04 is using consuming  6 gb ram and i could not see it with top (top 1o process using 2 gb max )
[13:47] <samba35> can some one help me to dig into issuse
[13:47] <samba35> i am using on 18.04
[13:48] <samba35> i have only 1 windows guest with 2 gb many time it fail to load due to low pc.ram message
[13:49] <ahasenack> rbasak: that is true
[13:50] <ahasenack> there is a mix
[13:51] <ahasenack> rbasak: https://hangouts.google.com/hangouts/_/canonical.com/standup-server
[13:55] <samba35_> sorry
[14:06] <lordievader> samba35_: What is the output of 'free -m'?
[14:06] <samba35_> Mem:           7934        6268         294         244        1370        1799
[14:07] <samba35_> Swap:          4095           0        4095
[14:10] <samba35_> may be stupid quastio but i reaily fail to copy text from terminal in 18.04 with less effort
[14:10] <tomreyn> !paste
[14:11] <tomreyn> !pastebinit
[14:12] <samba35_> sorry ,no i am trying to copy multiple lines with text but fail to copy them
[14:13] <Gargoyle> samba35_: got htop?
[14:13] <samba35_> yes
[14:14] <Gargoyle> sort the list by mem%
[14:14] <tomreyn> "ps auxw | sort -nk6 | tail" should likst the most memory hungry processes
[14:18] <samba35_> tomreyn, this this syntex is correct or typo ?
[14:19] <tomreyn> samba35_: works here
[14:20] <samba35_> ok ,sorry
[14:20] <Gargoyle> On 32GB RAM I have, chrome 0.7%, /proc/self/exe 0.8% (??), atom 0.8%, atom 0.8^, chrome 0.9% + 1.0% + 1.1%, gnome-shell 1.1% and slack 1.2%
[14:24] <samba35_> seems gnome-shell is problem with many many entry with 240 mb each
[14:25] <samba35_> gnome  shell is use by user and gdm
[14:25] <Gargoyle> I don't think it works like that. IIRC the "many entries" are child processes who all report the parents mem usage?
[14:26] <samba35_> is it becasuse of @reboot cron ?
[14:27] <samba35_> Gargoyle, sorry i did not get your point about paremnts mem usage
[14:28] <samba35_> there is only one terminal open /but even without opening terminal ( gnome-shell ?) usage is high
[14:29] <samba35_> english is not native :) (:
[14:30] <Gargoyle> samba35_: If you use htop, and put it in tree mode
[14:30] <samba35_> ok
[14:31] <Gargoyle> samba35_: Some of the entries are white, and some are green. The green ones always just report the exact same numbers as their parent white one (which not always the next white one above them in the list).
[14:32] <Gargoyle> Since I literally have 100's of chrome processes all reporting 1GB usage, they can't all actually be using 1GB!
[14:36] <samba35_> deja-dup monitor is 96 g with virt is that ok ?
[14:38] <samba35_> using server as a desktop :) is that a problem ?
[14:39] <Gargoyle> What do you have running?
[14:39] <Gargoyle> As in, main Apps?
[14:40] <samba35_> xchat ,terminal and firefox
[14:41] <samba35_> libvirt
[14:41] <Gargoyle> No vm's actually running at the mo?
[14:42] <samba35_> dpdk with 2048 m huge page
[14:42] <samba35_> yes not vm running
[14:43] <Gargoyle> OK. So assuming that the memory is being managed properly, what's the actual issue you are having?
[14:45] <samba35_> 8 gb ,HugePages_Free:     2048
[14:45] <samba35_> is that you are asking  ?
[14:45] <samba35_> sorry
[14:45] <samba35_> 8 gb total memory on system
[14:46] <samba35_> teamviewer  ok to you ?
[14:46] <Gargoyle> samba35_: If you are going to go through your entire process list and ask if it "looks ok to you", no-one is going to answer.
[14:48] <Gargoyle> Did you say that the error is that a guest vm fails to start?
[14:48] <samba35_> yes becasue of pc ram
[14:49] <Gargoyle> what version of windows are you trying to run
[14:50] <samba35_> windows 7 and 2 gb ram
[14:50] <samba35_> should i pastebin log
[14:51] <Gargoyle> Have you tried giving it more?
[14:51] <Gargoyle> set the vm to 4 gig?
[14:52] <samba35_> ok let me try with 4gb
[14:53] <samba35_> https://pastebin.com/57xQBT1v
[14:57] <Gargoyle> samba35_: Ahh ok.
[14:58] <Gargoyle> An error from qemu not inside the guest.
[14:59] <ahasenack> rbasak: would this be a Vcs-Browser url for a package in lp? https://code.launchpad.net/ubuntu/+source/sssd
[14:59] <Gargoyle> You are gonna need a bigger computer or a lighter desktop.
[14:59] <cpaelzer> ahasenack: https://code.launchpad.net/ubuntu/+source/sssd/+git maybe?
[14:59] <ahasenack> hm
[15:00] <ahasenack> yeah, better, that one has cloning instructions
[15:00] <Gargoyle> oh.
[15:00] <rbasak> Yeah the +git one
[15:01] <rbasak> Personally I prefer https://git.launchpad.net/ubuntu/+source/sssd but I suppose that's not canonical.
[15:01] <Gargoyle> samba35's xchat probably just got OOM killed! :/
[15:01] <Gargoyle> :D
[15:01] <ahasenack> tand vcs-git git://git.launchpad.net/ubuntu/+source/sssd ?
[15:01] <compdoc> samba35_, I installed 18.04 and had to learn to set interfaces in /etc/netplan/*.yaml, but virt-manger cannot see bridges created there. virt-manager wants the bridges set in /etc/network/interfaces. but thats fine, works great
[15:01] <rbasak> Yeah
[15:01] <ahasenack> seemed to work without the +ssh bit
[15:01] <ahasenack> ok
[15:02] <Gargoyle> compdoc: is virt-manager (qemu, etc) significantly better than virtualbox? (for the additional headache that seems to be involved in getting it running)?
[15:04] <samba35> sorry ,have to pull wire from wall ,
[15:04] <samba35> system hang after vm start
[15:05] <Gargoyle> samba35: You're gonna need a bigger machine or a much lighter desktop env.
[15:06] <samba35> agree
[15:07] <Gargoyle> I've got pretty much the same as you running: Desktop, Browser, few chat apps. Nothing *major*. and I am using 5GB RAM.
[15:07] <samba35> but on same system i was running 3 guest on ubuntu 16.04.03 (4)
[15:08] <Gargoyle> I play with 3 or 4 virtualbox VM's most days, so got a Ryzen 1800X + 32GB RAM.
[15:09] <samba35> when i install system for couple of day it was 2-3 gb but after 1 day system update /upgrade memory spike right after that
[15:09] <samba35> r u using 18.04 ? or 16?
[15:09] <Gargoyle> 18.04
[15:10] <samba35> which  kernel version
[15:11] <samba35> 4.15.0-20-generic ?
[15:11] <Gargoyle> yup
[15:12] <Gargoyle> git:(master) ✗ uname -a
[15:12] <Gargoyle> Linux ryzen1804 4.15.0-20-generic #21-Ubuntu SMP Tue Apr 24 06:16:15 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[15:12] <samba35> Linux shrikant 4.15.0-20-generic #21-Ubuntu SMP Tue Apr 24 06:16:15 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[15:14] <Gargoyle> samba35: Not sure it will help, but here's meminfo... https://paste.ubuntu.com/p/bVh627k929/
[15:22] <Gargoyle> samba35: Do you have a large value for unevictable?
[15:25] <samba35> Unevictable:       54888 kB
[15:29] <samba35> how do you copy text from terminal ? i could not select all  cat /proc/meminfo
[15:34] <samba35> VmallocTotal:   34359738367 kB
[15:34] <samba35> ??
[15:44] <samba35> Gargoyle, what is size of cpu and motherboard /which make and model
[15:44] <samba35> sorry what is price of ......
[16:01] <samba35> brb
[17:12] <Gargoyle> OK. Slightly different approach. Anyone know of any existing systems or guides for setting up an apt mirror so we can effectively freeze updates during our testing window. Eg, on a Monday we would update and sync with main ubuntu repo's, and then all our servers use our own internal mirror?
[17:17] <tomreyn> https://wiki.ubuntu.com/Mirrors
[17:20] <tomreyn> https://wiki.ubuntu.com/Mirrors/Scripts
[17:22] <Gargoyle> thanks tomreyn
[18:13] <ahasenack> rbasak: should I add systemd to build-depends because of my usage of systemd-detect-virt? systemd itself has "Priority: important"
[18:16] <rbasak> ahasenack: is systemd-detect-virt usually in the build environment without it?
[18:16] <rbasak> ahasenack: if so, you could condition it on a test -x
[18:16] <rbasak> That'd save the expense of such a major build dep possibly.
[18:17] <ahasenack> systemd is there in all vms and containers I tried
[18:17] <ahasenack> but maybe not in a chroot
[18:17] <ahasenack> I can treat it not being there as "ok, run the tests"
[18:17] <ahasenack> seems like a safe default
[18:17] <rbasak> +1
[22:39] <hehehe> :0