[00:59] <hmmmm> hi, I'm trying to downgrade from focal to bionic.  i have changed any instances of 'focal' to 'bionic' in /etc/apt/sources.list, and then created /etc/apt/preferences with Package: *   Pin: release a=bionic   Pin-Priority: 1001, then apt update which worked, but then apt upgrade fails with several unmet dependencies which results in a "broken packages" error.  what did i do wrong or what am i missing?
[01:05] <quadrathoch2> Hmmmm reinstall as downgrading is not supported
[01:06] <hmmmm> well really all i need to do is downgrade the kernel to a certain version, is that doable at least?
[01:09] <quadrathoch2> hmmmm you can try, by downloading the deb file from packages.Ubuntu.com. But not sure how broken your system is right now
[01:09] <hmmmm> I think i should be able to get it back into shape by undoing my changes and apt updating again
[01:09] <hmmmm> I need kernel 4.4.0-x, nothing in focal has a hard dependency on a 5.x kernel right?
[01:12] <quadrathoch2> 4.4? Wow that’s old. Nope, but I’m sure there are some apps not working probably
[01:15] <hmmmm> actually I can get away with 4.15
[01:15] <hmmmm> https://packages.ubuntu.com/bionic/kernel-image/4.15.0-1004-oem-di ?  this is what i'd download?
[01:16] <quadrathoch2> amd64?
[01:17] <hmmmm> yes
[01:18] <hmmmm> there's a big warning that says it's not to be installed on ubuntu
[01:18] <hmmmm> i'm wondering if i'm making a huge mistake
[01:20] <quadrathoch2> hmmmm Linux-image-generic should give your the right one from bionic
[01:49] <GSMarquis> Hi there
[01:49] <quadrathoch2> Hi
[01:52] <GSMarquis> I have a 6th gen x1 carbon. 20.04 and 20.10 both lock up after awake by lid open. The Activity menu still works but all apps do not accept any keyboard or mouse inputs. I can still launch apps from dock but have to close from activity menu after non of them accepting input. I cannot alt-f2 and restart gnome either. after a reboot all is good.
[01:53] <quadrathoch2> Sounds like your laptop has issues with suspend. Did it ever work?
[02:34] <fattywompus> Hey, anyone know if I install ubuntu 20.04 server if it automatically handles finding an existing windows install and conveniently adds it to grub the way a regular ubuntu install would do? Does ubuntu server even install grub or is it some completely different animal?
[02:38] <tomreyn> fattywompus: ubuntu server installs grub as well. if you install os-prober (after ubuntu server is already installed), it will also detect a windows installation which shares the same boot method (uefi or legacy bios, depending on how ubuntu server was installed)
[02:46] <fattywompus> Ok thanks.  And do you know if I get ubuntu server installed, which to my understanding has no desktop environment, is there any way I could install xfce and still have a choice to run it, or boot into the no gui mode?  Or once I put xfce on there will it always be xfce desktop unless removed?
[02:47] <steve_> I feel like it worked with 18.04. I had never used a bluetooth mouse before 20.04 though. I also feel like this has happened on 20.04 before I had a BT mouse. Other than that this is a straight forward ISO install. Nothing else.
[02:48] <Bashing-om> fattywompus: If you do not also install a desktop manager ( say lightdm or such) then you will boot to a terminal - from there you may start that GUI.
[02:48] <tomreyn> fattywompus: correct, ubuntu server comes with no desktop environment. it also defaults to systemd-networkd for network configurations, rather than network-manager as desktops do. as such, in case you'Re planning to, setting it up with a wireless connection can be more involved.
[02:50] <fattywompus> No its a desktop PC that will be strictly ethernet
[02:51] <steve_> I Just upgraded did a fresh repartition, reinstall from ISO, completly wiping everything to 20.10. I havent had that same lockup sense but its been just 4 hours or so and few awake sessions.
[02:53] <tomreyn> steve_: were you meaning to say something before "I feel like it worked with 18.04."? because that's the first we got to see since you joined.
[02:54] <steve_> I am having ID troubles with freenode. I am also GSMarquis.
[02:56] <tomreyn> i see.
[02:57] <tomreyn> steve_: after you initially discussed the suspend issues, there was: <quadrathoch2> Sounds like your laptop has issues with suspend. Did it ever work?
[02:57] <steve_> Basically, I have a stock X1 Carbon 6th gen, integrated graphics. It locks upon waking but not fully. I can still use activity menu. I can launch apps via dock but no other keyboard or mouse inputs to running apps. Just can close via actiity menu.
[02:58] <steve_> A reboot fixes it until it happens again. I cannot alt-f2 R to restart gnome.
[03:00] <tomreyn> have a look at your system logs, see what's logged before and after lines which have " PM:"
[03:00] <steve_> Which logs do I pull?
[03:01] <tomreyn> so something along the lines of this may help: journalctl -b -1 | grep -C 20 ' PM:'
[03:01] <tomreyn> but also look at (not only) acpi initialization during boot
[03:06] <steve__> Intersting to say the least. I should wait till this lockup happens again, reboot to gain control, then pull that log?
[03:10] <tomreyn> steve__: maybe try solving this network connectivity / nick regain issue first, otherwise supporting you gets difficult.
[03:18] <steve_> I know :).....I will work on that now.
[03:29] <fattywompus> So I've come up with the fact that xfce4 provides "startxfce4" to launch it from terminal and as you say having no display manager will make it so it naturally just boots to terminal mode.  If I started xfce then wanted to return it back to terminal mode can that be done as well?
[03:33] <Bashing-om> fattywompus: The GUI runs in TTY7 - the "terminal" is on TTY1.
[03:33] <quadrathoch2> steve_ you cant restart gnome probably because you are on Wayland not x, so you could change that (and maybe your issue goes away) or stay with wayland (and gnome will at some point reimplement the restarting of the shell)
[03:33] <quadrathoch2> hm gone kk
[03:33] <abhijit> Hi All. How to view thumbnail preview of RAW (ARW) files in Ubuntu 20.04 file browser?
[03:35] <abhijit> I found one article which asks to edit /usr/share/thumbnailers/gdk-pixbuf-thumbnailer.thumbnailer . is it safe to edit that file directly?
[03:35] <deltab> fattywompus: switch between virtual consoles using ctrl+alt+f1--f12, or exit xfce
[03:36] <GSMarquis> Alt-F2 does allow a restart with "r" when not locked up.
[03:38] <complib2891> can i just report package issue here?
[03:43] <Bashing-om> !bug | complib2891
[03:49] <lotuspsychje> complib2891: before filing bugs, we also advice to ask it in the support room first, maybe volunteers know existing bugs
[03:50] <complib2891> i've just filled using ubuntu-bug but from 18.04 instead of 20.04, but it's generally python2 code in python3 only environment.
[03:51] <guiverc> complib2891, it should be filed on the impacted system, as package versions are recorded; it won't be able to be re-created if wrong detail is recorded (ie. ubuntu-bug done on another different box)
[03:51] <complib2891> correction, from 20.04, the installationmedia say it's from 18.04
[03:52] <complib2891> i think it's similar, as it's 20.04, and greylistd is not those complex python program imo.
[03:52] <Bashing-om> complib2891: You are aware that python2 is depreciated ?
[03:52] <complib2891> yes i'm aware, but we still can install python2 if we want.
[03:53] <guiverc> complib2891, the bug reports you on 20.04, yes installation media was 18.04 is there too
[03:53] <complib2891> i've looked inside greylistd package, it's python2 code, i have to replace some files from https://github.com/eurovibes/greylistd then it's working
[03:54] <complib2891> i think maintainer tag Groovy correctly, but focal using old code, let me check groovy's
[03:55] <tomreyn> fattywompus: i think the better approach will be to "sudo apt update && sudo apt install xubuntu-desktop^", then use "sudo systemctl set-default multi-user.target" to make it boot to tty by default. then use "systemctl isolate graphical" to switch to graphical mode, or "systemctl isolate default" to switch back to tty mode.
[03:55] <guiverc> complib2891, the more detail put in the bug report, the easier & possibly earlier it'll be fixed.  I'd suggest adding your/these comments to the bug report
[03:59] <complib2891> @guiverc, i've update my notes to the bug report. thanks.
[04:03] <tomreyn> complib2891: looks like you're requesting a stable release upgrade (SRU) for focal (20.04 LTS) on a universe package imported from debian. that's not very likely to happen.
[04:04] <complib2891> tomreyn, oh, so it's debian?
[04:05] <tomreyn> not unless you can provide much more convincing arguments, such as a serious security risk, anyways. preparing it yourself may help, but you'll still need to reason for an SRU.
[04:05] <tomreyn> this package is probably imported from debian.
[04:07] <lotuspsychje> tomreyn complib2891 guiverc bug #1875257
[04:07] <complib2891> it's related to spam, greylisting should help as people from #exim suggested. it just kept coming in, i've added spamassassin clamav, denies on acl...
[04:07] <complib2891> not sure how spam related to security issue
[04:09] <lotuspsychje> complib2891: can you check if the bug above is what you experience?^^
[04:11] <complib2891> lotuspsychje, i think it's similar to #1875257 i've even tried the suggestion on #1875451
[04:12] <complib2891> to find out the whole package is not for focal without python2 (my guess)
[04:13] <tomreyn> if the package is broken (as it was in debian, according to 1875257), then an SRU could be possible. you'd still need to follow the SRU process https://wiki.ubuntu.com/StableReleaseUpdates
[04:15] <complib2891> so i should go #ubuntu-security?
[04:15] <complib2891> or just email security@ubuntu.com?
[04:15] <tomreyn> i don't think anyone suggested either of these
[04:16] <lotuspsychje> complib2891: if you feel its the same as #1875257 please affect yourself to the bug, then dupe your new bug towards it for start
[04:18] <complib2891> will see how to mark dupe, do i mark my dupe or find 1875257 and mark that as dupe?
[04:20] <lotuspsychje> complib2891: i duped it for you, now affect 187527 in the left upper corner please
[04:20] <complib2891> yes, thank you
[04:21] <lotuspsychje> complib2891: now you can checkout the link tomreyn provided you for SRU
[04:22] <complib2891> lotuspsychje, yea i scan tru it, and find the email and the channel name, i tot I should just email them lol
[04:25] <complib2891> hmn, i think i was reading a different link for the SRU, it's much more lengthy it seems.
[04:29] <complib2891> so i add more description to #1875257? with [Impact] [Test Case] [Where it could occur]?
[04:30] <lotuspsychje> complib2891: not sure howto handle SRU myself, maybe other volunteers can guide you for that
[04:31] <complib2891> lotuspsychje, thanks. i think i'll leave it at that... i manage to get it to work with simple file replace, and there's so much red tapes here.
[04:32] <lotuspsychje> complib2891: at least the existing bug has more affects now, so thats a good thing its known, and reported
[04:32] <lotuspsychje> complib2891: thank you to report
[04:32] <complib2891> lotuspsychje, yeah, i've tried.. cheers, thanks for the reply and help. appreciate nonetheless.
[04:43] <Seven_Six_Two> I've got ping packet loss on an ubuntu 20.04 laptop using an internal wifi. a usb nic in the same machine has no packet loss ping to gateway and 8.8.8.8, second ubuntu laptop also no packet loss to gateway and 8.8.8.8. I've looked at all machines' ifconfig -a as well as in the router for any conflicting ip addresses, and there are none.
[04:43] <Seven_Six_Two> I don't see a pattern yet, but packet loss varies between 5% and about 75%
[04:45] <Seven_Six_Two> ufw is disabled, and pinging localhost and nic IP show no loss
[04:47] <Seven_Six_Two> I've used dhcp and also set a static ip to see if there's a computer somewhere that I forgot about which might conflict. no change
[04:47] <Seven_Six_Two> I've disabled ipv6 completely, no change.
[04:53] <Seven_Six_Two> every danged time....Whenever I spend time asking for help, typing out all the things I've done, I seemingly without fail find the solution to my problem. I'm sorry to waste your time! I found a 'band' option in an "advanced network settings" app that isn't in other applets. Anyhow, switching to b/g fixed the issue. I'm in a back room!
[04:56] <Seven_Six_Two> hm..or maybe not. I just got a dropout, and a test showed 97% loss
[05:05] <Randolf> I'm trying to install VirtualBox manually because the one in apt is broken.  Which distribution is 20.04?  Is it eoan, bionic, xenial, buster, or stretch?  The instructions aren't clear about this.  Thanks.
[05:05] <deltab> Seven_Six_Two: that's a very common thing to happen. 'Rubber-ducking' refers to talking to a rubber duck to get the same effect
[05:07] <aeplus> Randolf, the virtualbox download site suggests to use eoan
[05:07] <Randolf> aeplus: Okay, thanks.
[05:07] <deltab> Randolf: 20.04 is focal. See https://en.wikipedia.org/wiki/List_of_Ubuntu_releases
[05:07] <Randolf> deltab: Ah, that will be helpful in the future.
[05:11] <RandolfR> Well, the "deb" command can't be found.  These instructions seem to be in need of some improvements:  https://www.virtualbox.org/wiki/Linux_Downloads
[05:11] <RandolfR> All I want to do is virtualize Windows 10 on my daughter's computer (she's in elementary school) so she can play games that only run in Windows (they don't work with WINE); stuff like Roblox and whatnot.
[05:12] <RandolfR> Whoops, I misread.  Never mind.
[05:13] <bindi> sounds like you should dualboot or get another machine altogether
[05:14] <RandolfR> No, I'm not dual booting.  No f---ing way I'm letting Windows run bare metal on any box as much as I can help it -- it f---s up everything way too much.  I'm done with trusting Microsoft.
[05:14] <RandolfR> Windows has trashed my systems way too many times.  They can go to hell.
[05:18] <Konomi> hi I installed kitty (terminal) and htop via snap
[05:18] <Konomi> but when running htop in kitty I get "Error opening terminal: xterm-kitty"
[05:21] <Randolf> Nope, this manual installation fails too.
[05:22] <Randolf> Never had any problems running VirtualBox on Ubuntu 18, but under Ubuntu 20 it just won't work.  Even when I try to sign keys for EFI and all that jazz.  This sucks.
[05:22] <Randolf> Maybe I'll try VMWare.
[05:37] <Randolf> VMware fails too.
[05:37] <Randolf> Is it even possible to virtualize Windows anymore?  (I'm using Ubuntu Linux 20.04.)
[05:45] <Konomi> it seems the ubuntu software couldn't find kitty as a snap so it installed it via apt instead and then installed htop as a snap
[05:45] <Konomi> removing the snap of htop then installing it with apt fixed the issue
[05:45] <Konomi> thought snaps were supposed to be easier
[05:47] <Randolf> Snaps are a nightmare.
[05:47] <Randolf> I've had software installed with snaps that failed, and then after removing them before re-installing the same application from apt without snap it couldn't read the data (this was with Mozilla Thunderbird).  This was my introduction to snap, and it was a disaster.
[05:48] <Randolf> I find that getting updates breaks snap applications too.  So I always try to use apt so I don't get the snap version, and then everything works properly.
[05:48] <Konomi> I don't understand why snaps are the default if they're so problematic
[05:51] <Konomi> https://github.com/maxiberta/htop-snap/issues/9
[05:51] <Konomi> fun
[05:54] <Seven_Six_Two> hey I totally figured out my packet loss problems!! Oh man.. It was my router! I've got a D-Link with a feature called Smart Connect that serves 2.4 and 5 on the same SSID. As soon as I disabled that, my packet loss stopped.
[05:57] <Konomi> does the ubuntu software package work without snaps?
[06:00] <Randolf> I'm trying to create a virtual machine in KVM now.  But it's failing when I tell it the maximum size of the disk (I increased from 40 GBs to 250 GBs because Windows 10 needs more than 40 GBs once you get all the updates in), and I'm getting the error "A storage path must be specified" but there's no option to specify a storage path.  How can I fix this?  Or do I need to do a manual installation somehow because KVM's GUI is b0rked?
[06:01] <Randolf> Rebooting Linux resolved this problem.
[06:09] <fattywompus> tomreyn: do you think there's some benefit to using xubuntu over xfce?  I thought xfce might be lighter and less chance of it interfering with the underlying ubuntu server in some way. Also thanks to you and Bashing-om and deltab for all the help thus far
[06:19] <Konomi> my fn+pgup doesn't seem to work in ubuntu 20.04 and the key combo produces no output in xev any ideas?
[07:59] <Konomi> fixed it and made a pull commit for upstread https://github.com/systemd/systemd/pull/17750
[07:59] <Konomi> upstream*
[08:00] <Konomi> pull request* I think the summer heat is melting me
[08:34] <otisolsen70> I tried asking in #ubuntu-server but got no response. I guess it might be on topic here as well: How do I make my server request an IP address from DHCP on boot up?
[08:35] <pho}e{nix> otisolsen70 set up the network interface to use dhcp?
[08:36] <otisolsen70> pho}e{nix, I already did during install.  When I installed it, the machine got an IP address from the DHCP server. The I took down the server and moved it to a different network on this network, the server will also receive network config via DHCP. The problem it, however, that when I boot up the server, for some reason it just configures the network with the original IP address that it got form the DHCP server.
[08:36] <otisolsen70> *from the other dhcp server.
[08:36] <otisolsen70> And it does not request IP from current DHCP. If I run dhclient -r <NIC> ; dhclient <NIC> then it obtains a working IP from the DHCP server. But this does not happen automatically.
[08:36] <pho}e{nix> otisolsen70 also, why would you use DHCP on a server? Don't you want the server to be reachable always on the same IP?
[08:37] <pho}e{nix> otisolsen70 can you paste your network config file please?
[08:37] <otisolsen70> pho}e{nix, dhcp server has configured a static lease for each server.
[08:37] <otisolsen70> pho}e{nix, I just configured this for dhcp during install. Not sure what file to paste.
[08:38] <pho}e{nix> why not just set a static IP on the server itself?
[08:38] <pho}e{nix> so that the DHCP server does not even play a role here
[08:38] <pho}e{nix> otisolsen70 I think it should be under /etc/netplan/
[08:38] <otisolsen70> pho}e{nix, sure, I might do that. But shoulnd't it work?
[08:39] <otisolsen70> Also, I would prefer using DHCP because this is what I do for all other machines on this network. With static leases. So I only need to manage network config in a single place.
[08:40] <pho}e{nix> otisolsen70 well it depends.. I NEVER do it the way you did it. I always set a static IP address for all my servers and not configure the DHCP server to mess things up
[08:40] <pho}e{nix> I usually set a static IP outside of the DHCP range as well so there's no IP conflict
[08:41] <pho}e{nix> otisolsen70 You don't need to turn off the DHCP server...just asign a static IP address on your server which is outside the DHCP IP range and you'll be fine.
[08:42] <pho}e{nix> "Managing network config" is not just asigning IP addresses to devices. That's a small part that shouldn't be changed often anyway.
[08:42] <otisolsen70> pho}e{nix, I do understand that. And I can live with static IP addresses on these servers, but I would still prefer having DHCP work.
[08:47] <pho}e{nix> otisolsen70 well...have you double(triple) checked the configuration on the DHCP server that doesn't work correctly?
[08:47] <pho}e{nix> Is there an IP address conflict maybe?
[09:07] <otisolsen70> pho}e{nix, no ip address conflict
[09:09] <locsmif_work> Hi all. My Iiyama 4K monitor won't turn back on automatically after I unlock my laptop. Apparently, according to discussion on StackOverflow, this is related to DP conenctions, however DVI-D doesn't properly support 4K, so I can't switch connectors. Is there something I could tweak? My otehr full HD monitor turns back on normally and so does my laptop screen.
[09:09] <locsmif_work> The 4K monitor I have to turn on manually.
[09:11] <otisolsen70> pho}e{nix, the client (ubuntu server) does not request any network config from DHCP on bootup. It does when I request it manually using dhclient. This is what I am trying to fix.
[09:12] <otisolsen70> pho}e{nix, the server works fine and is part of an infrastructure that have been running for a decaade.
[09:18] <otisolsen70> pho}e{nix, this is the network config under netplan: http://paste.yt/p13091.html
[09:41] <Konomi> is there anyway to have "ubuntun software" prefer apt pacakges over snap packages?
[09:41] <Konomi> aka if it can't find a apt package then it gets the snap version as opposed to what it seems to do at the moment which is prefer snaps over debs
[09:51] <lotuspsychje> Konomi: try synaptic or gnome-software instead of ubuntu software(snap store) perhaps?
[09:52] <Konomi> lotuspsychje: yeah I did a bit of reading if I install gnome-software and then install gnome-software-plugin-snap will that allow me toinstall debs first then snaps later etc?
[09:53] <lotuspsychje> Konomi: i dont think there's an easy workaround to 'prefer' snaps vs debs
[09:54] <lotuspsychje> Konomi: you best use the tools that completely seperate them
[09:54] <Konomi> that's kind of lame
[09:54] <Konomi> okay
[09:54] <Konomi> so possible gnome-software and snap-store
[09:54] <lotuspsychje> Konomi: there are wishlists/bugs on that, feel free to affect them
[09:54] <lotuspsychje> Konomi: yes, you can install any software manager you like
[09:54] <Konomi> I'm new to ubuntu so I'll wait on doing that long term debian user
[09:55] <lotuspsychje> Konomi: you can divide easy from terminal apt vs snap too
[09:55] <Konomi> yeah the terminal is easy it's just the store bothering me
[09:55] <lotuspsychje> apt-cache search foo - snap find foo
[09:57] <lotuspsychje> Konomi: bug #1789943
[10:19] <tacco> hi there. Did anything changed in the CPU handling behavior between 18 and 20? Because if i start a vm with bionic beaver i can see all my CPUS (248) and use them. With 20.04 i can see only a max of 64. Same behavior i have at debian. But not sure what is the problem here. Seems to be kernel related. Missing Flags or maybe a bug? dunno
[10:30] <BlueEagle> tacco: so `nproc` only yields 64 and not 248 as expected?
[10:33] <ace_me1> how to find bind9 user ?
[10:33] <BlueEagle> tacco: What does `lscpu` yield?
[10:33] <ace_me1> I've enabled logging and want to allow that user access to file
[10:34] <BlueEagle> ace_me1: `ps aux|grep bind` should show you which user runs the process I think.
[10:34] <lusrx> I installed Ubuntu 4 times yesterday and Elementary OS 1 time. They all lead to same problem: works fine up until I install all the latest updates, after which system freezes on next boot with only a black glowing screen (display is working) or with my laptop brand logo showing in the center of the screen. What could this be? I have tried installing with and without Nvidia proprietary drivers and packages, there's no difference.
[10:34] <ace_me1> BlueEagle: true && thx\
[10:34] <BlueEagle> lusrx: Did you try to boot with the 'nomodeset' option?
[10:35] <ace_me1> BlueEagle: still show wne bind9 restart  isc_stdio_open '/var/log/named_querylog' failed: permission denied
[10:35] <ace_me1> rwxrwxrwx 1 root bind 0 Nov 27 12:32 /var/log/named_querylog
[10:37] <tacco> BlueEagle: to nproc, yes it shows 64 with --all it shows 248 as expected.
[10:37] <tacco> means generaly they are there. but not online.
[10:37] <tacco> lscpu shows that the CPUs that are missing marked as "offline"
[10:37] <BlueEagle> tacco: Are you booting with nolapic?
[10:38] <tacco> if i try to manual enable i got a cannot allocate memory
[10:38] <pikapika> What is cron's behavior if the system was turned off when a job was supposed to run? Does it simply drop it or immediately runs it on next boot?
[10:39] <tacco> not sure. have to take a look. using the cloud image without any change for now
[10:39] <lusrx> BlueEagle: I did. However, I did not get consistent results. It was like... it worked in 1 out of 5 attempts (including power cycling). I had better luck with removing "quiet splash" and adding "debug systemd.log_level=info" but even that only worked sometimes.
[10:40] <pikapika> I don't remember how I did it but I get the usual stream of kernel and system messages instead of a splash screen now
[10:40] <pikapika> I think it involved unstinstalling some package via apt
[10:41] <BlueEagle> pikapika: cron does not keep track of previously run jobs so it cannot run a "missed" job as far as I know.
[10:41] <pikapika> So the only way a job would run is if cron is running (which presupposes the system being online of course) as well as it being the right time? Otherwise its simply dropped?
[10:42] <pikapika> I was just wondering if things like mlocate updatedb simply not happen for many days if we hit the unlucky shutdown times
[10:42] <pikapika> its a desktop not server obviously, so it isn't up 24x7
[10:43] <BlueEagle> pikapika: from the man page: "Cron examines all stored crontabs and checks each job
[10:43] <BlueEagle>        to see if it needs to be run in the current minute."
[10:44] <BlueEagle> pikapika: The obvious workaround is to have a job run at startup that checks when the job in question was last run. That again would require the cron job to leave a time stamp somewhere to be checked.
[10:45] <lusrx> I am puzzled with this problem. How come both Ubuntu and Elementary have the same problem with my laptop, and only after installing all the latest updates? I have now installed Debian 10.6 and so far so good, no problem. But I did install using the net install image (meaning no overwrites or updates needed as everything is pulled in from get go).
[10:45] <neorpheus> is there some general reason that some of my keyboard kees seem no tto work under ubuntu 20.04? i have a run of the mill logitech usb keyboard that works fine under windows and other distros. i have also tested my wifes keyboard on my system with the same "dead" keys. there may be others but the ones ive noticed are ALT, F10, and F12. i do know F5. Shift, CTRL work.
[10:46] <pikapika> neorpheus, try a live boot environment on the same laptop
[10:46] <pikapika> oh
[10:46] <pikapika> hmm
[10:46] <pikapika> Try to see at various points
[10:46] <pikapika> such as type some shit in bios time before os boots
[10:46] <pikapika> to see if its os fault or something deeper
[10:47] <pikapika> neorpheus, also believe it or not, sometimes wifi and such signals can interfere even with wired keyboards
[10:47] <pikapika> I've had it happen to me
[10:48] <neorpheus> i mean its def OS fault. as stated ive tested this keyboard and know its good on other hardware. or same hardware with other OS. also tried a second known good keyboard and it shows the same nonresponsive keys in ubuntu 20.04. but works fine on same desktop with different os
[10:48] <BlueEagle> neorpheus: If you open xev does it also show no events reaching the X server for those keys?
[10:48] <pikapika> neorpheus, no
[10:48] <pikapika> You can't be sure until you have tested it on same laptop
[10:48] <pikapika> yes xev is a good test
[10:48] <pikapika> After that try bios
[10:48] <pikapika> It could in fact be a looseness of usb connection or similar
[10:49] <pikapika> I have had some really strange keyboard shenanigans so trust me you have to test on same hardware
[10:49] <pikapika> same laptop/pc that is
[10:49] <tacco> BlueEagle: nolapic isn't set in grub.conf
[10:49] <pikapika> to be fully sure
[10:50] <tacco> and here is my lscpu http://paste.openstack.org/show/800491/
[10:50] <pikapika> the series of tests is as he stated xev, then bios, then live boot environment
[10:52] <neorpheus> maybe i didnt explain myself fully. the keyboard as well as all 8 usb ports on the desktop have been ruled out as the cause of the error. i have tested 2 keyboards on 2 differnt computers. both keyboards work fine under other os's on both systems.  neither keyboard works full on the problem system that is running ubuntu20.04. i cannot test the keyboards against ubuntu on the other system. i will look into this
[10:52] <BlueEagle> Norkle: So xev does not show any events for the keypresses?
[10:53] <BlueEagle> Norkle: Sorry, wrong target
[10:53] <tacco> lscpu from the HV, there is also ubuntu http://paste.openstack.org/show/800492/ but 18.04
[10:53] <BlueEagle> neorpheus: so xev does not show any events for the keypresses?
[10:53] <BlueEagle> tacco: echo on > /sys/devices/system/cpu/smt/control
[10:53] <BlueEagle> tacco: That appears to have solved a similar problem.
[10:54] <BlueEagle> tacco: https://www.nixcraft.com/t/i-only-have-8-cores-cpu-but-linux-kernel-only-4-cores-when-i-run-top-htop-and-gui-tools/2822/10
[10:54] <tacco> echo on > /sys/devices/system/cpu/smt/control
[10:54] <tacco> -bash: echo: write error: No such device
[10:54] <Konomi> You can use showkeys instead of xev to pick up keys xev can't see
[10:55] <BlueEagle> tacco: Then I have extended the length of my google-fu for this issue. :/
[10:55] <lusrx> neorpheus: how did you install Ubuntu? What keyboard layout did you select? Maybe you selected the hacker version that says it excludes dead keys? Perhaps by mistake?
[10:55] <Konomi> But it has to be done in a vt not xorg/waylabd
[10:56] <lusrx> I would reinstall Ubuntu...
[10:58] <neorpheus> sorry, this is the output from xev: https://pastebin.com/hfGvh9qt
[10:59] <neorpheus> it seems to see the keypress
[11:00] <neorpheus> trying to map these keys in games acts like the key was never pressed. same results for example when trying to use F10 to exit cli progs. F10 works fine in BIOS
[11:01] <BlueEagle> neorpheus: Ok, so then it is something else trapping these keypresses before they reach the game. Not sure what that would be, though.
[11:02] <lusrx> neorpheus: Sorry but I don't understand any of that. What am I looking at? What keys did you press?
[11:02] <tacco> BlueEagle: im 100% sure this hat nothing to do with SMT because in ubuntu 18.04 i have also smt/control -> notsupported and see all cpus in the VM as expected.
[11:02] <tacco> and in the HV i already can see all cpus.
[11:03] <neorpheus> left alt. it calls out ALT_L
[11:03] <tacco> thats not the point. I can do a 1:1 diff between ubuntu cloud-image bionic and focal. They have the same machine description and so on. Only the volume differs. one with focal the other with beaver. beaver shows 248 CPUs and focal only 64.
[11:04] <tacco> and have no clue why. because nproc --all or --hardware shows that there would be more CPUs than available/online
[11:05] <lusrx> Ah OK, "KeyPress event" and "KeyRelease event". That I do understand. You pressed "Alt_L" (left Alt) and then released it.
[11:07] <lusrx> neorpheus: I would say you definitely pressed that left Alt key then. But I don't know how to help you resolve the problem (other than being a voice of reason).
[11:08] <lusrx> I don't even know what xev is, but if this is some kind of judge on what key was pressed, then it has given its verdict: the key was pressed. What now?...
[11:08] <neorpheus> yeah, i got same results for F-10, F12, "Backslash", and "Scroll Lock". no worries. ive found other keys te rebind my game controls until i figure it out. just a pain not having half my F keys
[11:08] <TJ-> tacco: what does this report in terms of max CPUS "journalctl -k | grep smpboot"
[11:10] <TJ-> neorpheus: have you ensured the keyboard map is set correctly? Might be worth trying "sudo dpkg-reconfigure keyboard-configuration"
[11:11] <BlueEagle> neorpheus: I am not sure of a good way of finding which application traps keys. It would be possible to start a "clean" X-server and start one and one application to see when the key response disappears.
[11:11] <neorpheus> man that would be a pain but your right it would probably work.
[11:13] <ace_me1> I did configure a bind9 like is shown here https://www.digitalocean.com/community/tutorials/how-to-configure-bind-as-an-authoritative-only-dns-server-on-ubuntu-14-04
[11:13] <ace_me1> and I try to nslookup from local to the bind9 ip address: nslookup, then server xxx.xx.xx.xx and then I query www.configureddomain.com but it answer server can';t find www..... SERVFAIL
[11:13] <ace_me1> any hints please ?
[11:18] <ace_me1> anyone a little help please with bind9 ?
[11:18] <BlueEagle> !patience | ace_me1
[11:18] <ace_me1> sure ubottu
[11:19] <lusrx> Every time I install Ubuntu 20.04 (using 20.04 or the 20.04.1 installer, with or without Nvidia drivers and packages) it works perfectly for my needs. But then I go and install all the latest updates and it causes the laptop to freeze during bootup, leaving me with either a black glowing screen (display working) or with my computer brand name glowing in the center of the screen. What's causing this? Installing Debian 10.6 worked perfectly and still
[11:19] <lusrx> does (but I used net installer so there were no updates to install).
[11:23] <lusrx> Sorry to bother you guys. If someone is familiar with this problem please ping me. I would like to go back to using Ubuntu. But for now I will be using Debian. Not being able to update the system without risking breaking the boot process doesn't seem like a sane or reliable choice.
[11:45] <TJ-> lusrx: when it happens, as an alternative, are you able to intercept the boot process and get to the GRUB menu? if so then it tells us the issue is likely something to do with a) updated kernel b) updates in the initialramfs (initrd.img file)
[11:51] <lusrx> TJ-: when it happens it hangs itself, there is not much I can do at that point other than force a power off. The magic sysrq key combo does not work to kill it off gently. I can anticipate when it will happen. It will happen every time I install latest updates to a new Ubuntu 20.04 installation, and on every fifth or so bootup thereafter. Not sure how I would intercept it?...
[11:51] <ace_me1> where I can find a bind zone file to use it as an example ?
[11:53] <BlueEagle> ace_me1: Typically examples are in /usr/share/doc/ or there abouts
[11:54] <BlueEagle> ace_me1: also googling found this site: https://ubuntu.com/server/docs/service-domain-name-service-dns
[11:57] <lusrx> TJ-: I would like to try what I did with Debian. I used a special iso image called firmware-10.6.0-amd64-netinst.iso that includes drivers for my Intel Wifi antenna, and it boots a minimal (I think it's called core) system that then pulls down all the latest packages to my laptop and installs them. So there is no update later on that breaks my installation. Does Ubuntu have something like this? I'm looking at "Netbooting the live server installer"
[11:57] <lusrx> instructions and it seems to be a different kind of net install using PXE boot. I don't think this is the same thing?...
[11:57] <lusrx> https://discourse.ubuntu.com/t/netbooting-the-live-server-installer/14510
[11:58] <ace_me1> BlueEagle: thx... also found a online validator https://bind.jamiewood.io/#
[12:04] <lusrx> This is for installing a server? I want to install a desktop Ubuntu. I found this: https://askubuntu.com/questions/1233746/download-ubuntu-minimal-iso-20-04lts
[12:04] <lusrx> Will this help me install desktop version of Ubuntu? Is the mini ISO no longer supported?
[12:06] <lusrx> Does it include drivers for Intel Wifi antennas?
[12:07] <EriC^^> lusrx: mini iso can let you install desktop, but it doesnt support uefi just so you know
[12:13] <TJ-> lusrx: I don't think that is going to help - better to identify the root cause. As I said above, we need to know if this is at the boot-loader (GRUB) or kernel/OS stage. If it happens "on every fifth or so boot" that suggests it is a kernel/OS issue and therefore we ought to be able to narrow it down quickly
[12:14] <chosig> set weechat.color.chat_delimiters 29
[12:14] <chosig> oops
[12:15] <lusrx> EriC^^: thanks! I enabled CSM and now I can boot it. This looks familiar. It's just like the Debian, but with very simple graphics and no mouse support. But it does support my Intel Wifi antenna so that's good.
[12:16] <locsmif_work> Hi all. My Iiyama 4K monitor won't turn back on automatically after I unlock my laptop. Apparently, according to discussion on StackOverflow, this is related to DP conenctions, however DVI-D doesn't properly support 4K, so I can't switch connectors. Is there something I could tweak? My otehr full HD monitor turns back on normally and so does my laptop screen.
[12:16] <TJ-> lusrx: What I'd do is add "break=top" to the kernel command-line to be able to pause booting at the initialramfs stage
[12:16] <locsmif_work> The 4K monitor I have to turn on manually.
[12:17] <locsmif_work> I disable dpms in .xinitrc, which works, but it's not a solution, it's a pretty grim workaround
[12:17] <EriC^^> lusrx: no problem, alright sounds good
[12:18] <lusrx> TJ-: OK. So that's the linux line in GRUB cmd config. When or how do you suggest I add that in? I mean it freezes first after the update and on first reboot, before it allows me to look at the GRUB menu and edit boot parameters. So I need to add that line from within Ubuntu then? To catch it early on?
[12:20] <TJ-> locsmif_work: maybe DPMS tooling could help: "xset dpms on" (see 'man xset')
[12:20] <BlueEagle> lusrx: If you can boot to a live cd and then chroot into the installed environment you should be able to update grub I think.
[12:20] <TJ-> lusrx: based on what you've just said it sounds like your issue is UEFI boot
[12:22] <TJ-> locsmif_work: typo! should be "xset dpms force on"
[12:25] <lusrx> TJ-: maybe so... I don't know. What could be the issue though? I mean it works the first time around. Up until I decide to install latest updates. And Debian makes it look like a child's play, and even supports UEFI in its mini bootable image. Not that I would want to switch to Debian...
[12:27] <lusrx> I have not tried other derivatives of Debian, other than Ubuntu and Elementary. And Elementary displays the same issue. I can clean install it using its regular image and reboot it many times. Up until I update it... then it starts freezing my laptop on bootup.
[12:29] <tacco> TJ-: http://paste.openstack.org/show/800497/ here we go. sorry for the late response
[12:29] <tacco> i posted poth bionic beaver with all CPUs and focal that only showup with 64 CPUs
[12:30] <tacco> as i told you thats the same machine only changing the underlying image from focal to bionic
[12:35] <coconut> Is there a known linux hardware store reseller which sells external blu-ray burners?
[12:35] <tacco> it came to my mind that this could propably have something to do with CPU-Bug mitigation or something similar?
[12:36] <tacco> coconut: usualy you shouldn't have mouch trouble with them et all
[12:36] <TJ-> tacco: that output suggests only physical CPUs are recognised
[12:37] <TJ-> tacco: how many physical cores does the host have?
[12:37] <tacco> i have 256 cores in physical
[12:37] <TJ-> tacco: cores or threads ?
[12:38] <tacco> 2 sockets with Core(s) per socket  64 and 2 threads each core
[12:38] <tacco> AMD EPYC 7742 64-Core Processor
[12:38] <tacco> without any problems at bionic only focal won't work
[12:39] <coconut> tacco, ok but i rather have this guaranteed one way or another :)
[12:40] <tacco> no problem. Wanna see the lcspu from the host system as well?
[12:40] <tacco> or anything else i can show.
[12:41] <tacco> http://paste.openstack.org/show/800498/ thats the host CPU. and this is running on bionic.
[12:42] <tacco> this is my openstack HV, with qemu/KVM and as VM i tested with bionic, all fine, but focal not.
[12:42] <tacco> i could try to ignore this and only use bionic, but i'm almost sure that this maybe could be a kernel bug.
[12:42] <tacco> because on debian i have exactly the same issue.
[12:43] <tacco> but there i only tested with buster.
[12:43] <tacco> but buster has a never kernel version then bionic.
[12:43] <tacco> maybe the difference can be found between 4.19 and 4.15
[12:44] <TJ-> tacco: can you give us the entire kernel log from the focal guest? "pastebinit <( journalctl -k )"
[12:46] <tacco> sure. :) give me a minute
[12:47] <tacco> oh.. pastebinit is nice.
[12:47] <tacco> https://paste.ubuntu.com/p/pny9253m3g/
[12:47] <tacco> every day something new. :D
[12:54] <BluesKaj> Hi all
[12:55] <FurretUber> Hi, I am trying to configure a live media to have a persistent partition only for snaps. Initially I tried creating a mount point for /var/lib/snapd, but many files are added to /etc, which were lost on shutdown. Is it possible to make snaps persist and work?
[12:57] <TJ-> tacco: I've looked at the kernel source code, and it looks like topology_update_die_map() isn't called for the remaining CPUs. That is called from validate_apic_and_package_id() called from identify_secondary_cpu() from smp_store_cpu_info() from .... rabbit-hole!
[12:58] <TJ-> tacco: good news from kernel-log is, it sees all the APICs
[13:02] <TJ-> tacco: looks like ti could be a KVM/NUMA issue
[13:21] <TJ-> tacco: seems you're not alone https://forum.mikrotik.com/viewtopic.php?p=826858
[13:34] <geosmile> is there a nice ui for maintaining multiple machines through ssh - which shows update/upgrades are pending etc?
[13:40] <tacco> TJ-: thanks a lot for your help. Numa was my fist tought, but what makes me totaly confused is that with newer version of ubuntu i get only 64 but with a older version i'm able to get all cpus. means, it isn't a general issue with the setup or that i missed something.
[13:40] <BlueEagle> geosmile: I use Tinycp for managing my two servers.
[13:42] <BlueEagle> geosmile: Doesn't use SSH, but you can set it up on SSL I guess.
[13:48] <tacco> TJ-: and this behavior also happend to debian and fedora, maybe a CPU problem, i guess not many people have this CPU that cost around 6.5k
[13:48] <geosmile> BlueEagle, will look at tinycp, thank you!
[13:49] <tacco> still think it could be cpu-bug mitigation
[13:54] <TJ-> tacco: smp: Brought up 2 nodes, 64 CPUs
[13:54] <TJ-> tacco: and compared to other systems, it looks like you have 2 relevant issues. 1) node #1 isn't being brought up
[13:57] <TJ-> tacco: 2) the logical cpus (threads) on the real CPUs aren't being reported in that number, but may still be available via other means. I say this after comparing your output with my (admittedly smaller example): "smpboot: Allowing 16 CPUs, 8 hotplug CPUs" ... "smpboot: CPU0: AMD Ryzen 7 3700U with Radeon Vega Mobile Gfx" -> "smp: Brought up 1 node, 8 CPUs"
[13:58] <FurretUber> Hi, I am trying to configure a live media to have a persistent partition only for snaps. Initially I tried creating a mount point for /var/lib/snapd, but many files are added to /etc, which were lost on shutdown. Is it possible to make snaps persist and work? Currently the snaps are saved but they don't start after rebooting
[13:59] <TJ-> tacco: however there are other indications all 248 CPUs are available (I'm comparing your VM guest kernel log to mine on bare hardware) -> "SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=248, Nodes=2"
[13:59] <TJ-> tacco: also you have "rcu:         RCU restricting CPUs from NR_CPUS=8192 to nr_cpu_ids=248"
[13:59] <LiveWire> FurretUber: yuo tried creating it with mkusb?
[14:00] <tacco> 2 nodes because of the numa nodes. that was defined by myself
[14:00] <tacco> in the flavor.
[14:01] <TJ-> tacco: right; 2 sockets
[14:01] <TJ-> tacco: 2 (sockets) x 64 (cores per socket)  x 2 (threads per core)
[14:02] <tacco> i know. I've also tried to define this verry exact in the flavor but didn't helped
[14:03] <TJ-> tacco: it feels like in the guest node 1 isn't being brought online
[14:03] <FurretUber> LiveWire: I created a FAT32 partition with the squashfs with the system and SYSLINUX, and a partition which is mounted with a systemd service
[14:03] <tacco> but only with 20.04
[14:06] <LiveWire> FurretUber : I created quite a few persistent live usb sticks using mkusb. Not sure what your trying to accomplish though
[14:06] <TJ-> tacco: hmmm, look what I've found on mine: /sys/devices/system/cpu/offline:8-15
[14:07] <tacco> uhh strange
[14:07] <TJ-> tacco: this is v5.10-rc5 with Ryzen 7 3700U (8 cores, 16 threads)
[14:09] <N3bulaK> Hey all
[14:09] <N3bulaK> Ubuntu on my machine takes around 1 minute and 40 seconds to boot
[14:10] <N3bulaK> is there a way to speed this up?
[14:10] <FurretUber> LiveWire: I want only snaps persisting and the bas OS (/usr, /lib, ...) not persisting. It's for computers with 4 GB storage only
[14:11] <FurretUber> So I was thinking on keeping the browsers installed as snaps, which would be always updated, while keeping the base system intact
[14:11] <FurretUber> Also, it's storage slower than 1 MB/s write speed...
[14:11] <coconut> N3bulaK, yeah you can replace your hdd with an pci-e nvme ssd.
[14:12] <coconut> that would boots speeds with booting
[14:12] <coconut> *boost
[14:13] <lotuspsychje> N3bulaK: sudo apt install preload haveged stacer and start tweaking your startup items + systemd services
[14:20] <TJ-> tacco: what does this show? "grep . /sys/devices/system/cpu/cpu*/topology/thread_siblings_list"
[14:20] <TJ-> tacco: this Ryzen 7 3700U actually has 4 cores and 8 threads, so not sure how it is getting reported as 8-core 16 thread
[14:25] <LiveWire> TJ-: I just looked my 3700 shows that also
[14:27] <TJ-> LiveWire: weird!
[14:30] <LiveWire> TJ-: guess im a dumb ass "AMD RYZEN 7 3700X 8-Core 3.6 GHz" is what i bought
[14:38] <tacco> TJ-: https://pastebin.ubuntu.com/p/gMbkrFMHQJ/
[14:42] <Konomi> N3bulaK: https://www.freedesktop.org/software/systemd/man/systemd-analyze.html#systemd-analyze%20blame
[14:44] <TJ-> LiveWire: :)
[14:45] <SimAV> Hi, after upgrading ubuntu to 20.10 on a Lenovo D330-10IGM, the Goodix Capacitive TouchScreen stopped working. It is not found by xinput anymore. Dmesg shows https://paste.ubuntu.com/p/H8BpTgTt3Y/
[14:45] <TJ-> tacco: extremely weird! what are the 2 kernel versions (working/not-working)
[14:47] <tacco> 4.15 working (bionic) 4.19 not working (debian buster) 5.4 not working (debian buster backports) 5.4 not working (focal)
[14:48] <SimAV> has anyone else encountered problems with goodix touchscreen on ubuntu 20.10? Or any idea?
[14:54] <lotuspsychje> SimAV: we dont have much touchscreen bugs for gnome yet, so i would advice, if it was working before, i would file a new bug for that to help yourself and the community
[14:55] <Guess59061> Is there a way to make files in a certain directory non-executable while in that dir, but executable when outside of it (i.e. while keeping their (+)x flags intact)?
[14:57] <pavlos> N3bulaK: pastebin <( systemd-analyze blame )
[14:57] <Guess59061> I'd also like to be able to easily add/remove files from that dir.
[14:57] <SimAV> lotuspsychje, it is completely independent of the desktop environment. It doesn't work on KDE either nor does it on the login screen.
[14:58] <lotuspsychje> SimAV: but you said it was working before your upgrade?
[14:58] <SimAV> yes, the touchscreen worked before.
[14:59] <lotuspsychje> SimAV: so i would consider that as an (upgrade) bug
[15:00] <SimAV> Me too, as it was working on eoan
[15:01] <lotuspsychje> SimAV: ubuntu-bug package from a terminal to file a !bug
[15:03] <SimAV> lotuspsychje, thanks for the pointer. Is this dist-upgrade or release-upgrade or rather Other problem?
[15:05] <lotuspsychje> SimAV: in this stage, we need to investigate more, filing the bug will import more info about your system, so volunteers and devs can start search/debug your issue
[15:06] <SimAV> When choosing "Other problem", the question is which package. I'm tempted to suspect the kernel module, as I can provoke the error message a second time in the dmesg using rmmod goodix; modprobe goodix
[15:07] <lotuspsychje> SimAV: not sure wich package they use exactly for touchscreen, maybe linux or xorg
[15:10] <tacco> TJ-: will try what happend if i jump from 4.15 to 4.18 in bionic
[15:14] <SimAV> lotuspsychje, linux-image-5.8.0-25-generic seems to be the culprit. When booting 5.4.0-26-generic the touchscreen works.
[15:16] <tacco> TJ-: so with bionic. i was able to install 4.18 with all 248 CPUs. now i upgraded to 5.4 to see what happend, and it's only 64 CPUs.
[15:21] <TJ-> tacco: so, now to find out what changed!
[15:24] <tacco> TJ-: so 100% sure in bionic 4.18(248 cpus) is the laters 4. version upgraded to the first 5. version 5.0.0-15 (64 cpus) no luck
[15:24] <tacco> and in debian buster with 4.19 i had the same issue. maybe something changed between 4.18 and 4.19 will crawl trough the change logs
[15:27] <tacco> and 4.19 is from late 2018 and in this time i immediately think about spectre and meltdown mitigations.
[15:33] <tacco> TJ-: maybe related https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1734249.html
[15:36] <tacco> https://lkml.org/lkml/2018/8/14/731
[15:50] <coconut> Is there any software in the repo for burning blu-ray discs?
[15:51] <TJ-> tacco: no, the kvm parts (looking at your kernel log) seems OK
[15:54] <tacco> TJ-: ok. puuh.. i'm so done with this week. Wish you all a nice weekend. will keep my hands away from computer until monday. See you then and stay safe!
[15:55] <jwash> i installed this copy of ubuntu ~2 years ago with encrypted disk. i regularly change my os level password. how do i cycle the encryption password for my disk?
[15:56] <DJones> coconut: https://help.ubuntu.com/community/CdDvd/Burning This may help it references a ppas so that Brasero and K3b can burn blueray discs, not sure whether its been superceeded simce it was written though
[15:56] <coconut> *click*
[15:56] <jwash> is blank blu-ray media cost effective?
[15:59] <TJ-> tacco: this info might help us: "grep . /sys/devices/system/cpu/cpu2/topology/* "
[16:01] <TJ-> tacco: even more helpful to use * and list all cores!  "grep . /sys/devices/system/cpu/cpu*/topology/* "
[16:04] <tacco> http://paste.ubuntu.com/p/qXbBvdHxYB/
[16:04] <tacco> i've tested on bionic, that with 4.18 i have all cpus with 4.19 not anymore.
[16:04] <tacco> So the problem exists since 4.19
[16:05] <coconut> jwash, i have not looked at that
[16:11] <TJ-> tacco: thanks! I'm working my way through the commits between those two versions that affect SMP
[16:11] <TJ-> tacco: how about creating a quick ubuntu bug report so I can keep you updated there and we can get the kernel team involved?
[16:12] <tacco> TJ-: hope you have a buy me a coffee link for me.
[16:13] <TJ-> tacco: "ubuntu-bug linux" and give me the bug #
[16:14] <TJ-> tacco: there appear to be only 6 related commits between those versions
[16:15] <TJ-> tacco: My gut feel is its commit bc2d8d262cba5736332cbc866acb11b1c5748aa9
[16:18] <TJ-> tacco: there's something wrong with your topology for the guest, for v4.19+ -- your pastebin shows that it thinks you've got all those CPUs on the same core!
[16:20] <TJ-> tacco: -"/sys/devices/system/cpu/cpu63/topology/core_id:0" -- for physical cores that would be core_id:63 (no SMT) or core_id:32 (2 thread SMT)
[16:21] <TJ-> tacco: if it is possible, it would be 'interesting' to boot that host with the same OS/release/kernel and capture the kernel log and the sysfs topology as above, then compare with the guest view
[16:21] <quadrathoch2> jwash I guess it really depends on what your goal is with blu rays. imho it's a cheap way for getting a ro backup strat for example
[16:22] <jwash> quadrathoch2: that was my idea
[16:22] <jwash> i've got about 35gb of compressed data which will grow to about 50gb in the next year
[16:22] <jwash> by 2022
[16:23] <quadrathoch2> jwash look imho into m-disc, it's the data storage variant of bd, I get 100gb discs for 13c/gb
[16:26] <tacco> TJ-: would it help to have the virsh dumpxml of the VM?
[16:28] <TJ-> tacco: indeed
[16:32] <tacco> TJ-: here we go. https://pastebin.ubuntu.com/p/pFp7zGjJPf/
[16:35] <TJ-> tacco: this doesn't look correct to me "<topology sockets='248' cores='1' threads='1'/>"
[16:37] <tacco> hm.. will try to fix this. for 248 CPUs this would be 62 cores and 2 threads right?
[16:38] <TJ-> tacco: I'd think so, yes, with 2 sockets
[16:40] <TJ-> tacco: from what I can see, "openstack image set $IMAGE_ID --property hw_cpu_sockets=2 --property hw_cpu_cores=62 --property hw_cpu_threads=2"
[16:42] <ace_me> when I query my new DNS server I try to configure I get
[16:42] <ace_me> server can't find www.domain.com: SERVFAIL
[16:42] <ace_me> where that domain.com is a type master zone
[16:43] <tacco> TJ-: recreated the flavor and used hw:cpu.. but yes. it ends at the same. Thanks for the help
[16:46] <TJ-> tacco: OK, so that seems to indicate the KVM topology isn't being passed through. what kernel version is on the host (where the kvm module lives)
[16:50] <tacco> TJ-: https://pastebin.ubuntu.com/p/sPWHVMn5FX/
[16:54] <TJ-> tacco: looks OK - and you say the guest kernel log for v4.18 on that looks the same as the earlier kernel log?
[16:54] <tacco> TJ-: http://paste.ubuntu.com/p/5txHGPHpBT/
[16:55] <tacco> with 4.18 i had no issues because all 248 CPUs came up.
[16:55] <TJ-> tacco: yeah. that topology output indicates the kernel thinks there are 32 cores with 2 threads each!
[16:56] <TJ-> tacco: which is an improvement!
[16:56] <tacco> :)
[16:56] <TJ-> tacco: "64" seems to be a limit of some kind here. That was the 1st thing I looked for earlier in kernel CONFIG_* in case we had an arbitrary limit, but didn't find anything
[16:57] <tacco> yea the flavor has now --property hw:cpu_sockets='2' --property hw:cpu_threads='2' --property hw:cpu_cores=62
[16:57] <tacco> my scope was CONFIG_NR_CPUS*
[16:57] <tacco> in debian this is set to 64
[16:58] <tacco> but at ubuntu MAXSMP is enabled and raise this values up to 8192
[16:58] <tacco> but tis also don't help
[16:58] <TJ-> tacco: yes, and then it's dynamically altered at runtime to match. The weird thing here is the per-cpu regions are done correctly
[16:59] <tacco> indeed
[16:59] <tomreyn> ace_me: SERVFAIL is quivalent to a HTTP servers' 5xx, an unexpected situation occurred on the server, review its logs.
[17:16] <chosig> Budgie related but the channel is dead, is there a way to diable Alt+Q to pop open the Quake terminal?
[17:26] <CorvusCorax> hi. I reported a kernel bug on ubuntu18 yesterday. now I booted from a ubuntu 20 live CD and managed to reproduce it there, too. I'd like to report it, but "ubuntu-bug" just hangs indefinitely. is there a way to skip "ubuntu-bug" and report a bug manually?
[17:28] <tacco> TJ-: You may only assign yourself because you are not affiliated with this project and do not have any team memberships.
[17:28] <tacco> TJ-: https://bugs.launchpad.net/launchpad/+bug/1905993 maybe have a look by yourself.
[17:29] <TJ-> tacco: thanks
[17:29] <tacco> hope this is the right place. If not. let me know. I'm off for now but you can leave a private message any time, will read it later.
[17:29] <tacco> TJ-: no Thank you!
[17:30] <TJ-> tacco: I've moved it to the correct project and package: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1905993
[17:31] <TJ-> tacco: just so I'm clear; the host kernel remains at 4.18 but this occurs when you upgrade the guest kernel?
[17:32] <tacco> yes
[17:33] <tacco> or when i use another distro like focal
[17:33] <tacco> i don't change anything on the hypervisor. only the VM with always the same flavor settings
[17:33] <tacco> and with plain bionic it works.. with kernel above 4.18 not.
[17:34] <tacco> and a debian guy told me yesterday that he could find the same issue, but this seems not to affect intel cpus.
[17:34] <tacco> maybe something special with theese AMD cpus have no clue anymore but maybe the best idea to escalate this to a kernel dev.
[17:42] <nedbat> i can see the motd by cat'ing /run/motd.dynamic.  What creates that file?
[17:48] <CorvusCorax> nvm. ubuntu-bug works if I run it before the kernel-oops was triggered. but after the OOPS happened, ubuntu-bug will freeze collecting information. so its actually quite simple, just have to report the bug before it occurs and upload the dmesg/trace later
[18:12] <GSMarquis> Having lockup issues upon waking. X1 Carbon 6th gen, integrated graphics. Happens on 20.04 stock ISO install. The activity menu still works and I can launch apps but cant control them. No keyboard/mouse inputs work below the activity menu.
[18:13] <GSMarquis> I cannot alt-f2 r gnome after this lockup happens. If I reboot, all comes back as normal.
[18:13] <GSMarquis> This lock up can happen mins after a wake from sleep.
[18:15] <GSMarquis> I format, reinstall 20.10 ISO. Has not happened yet. About 24 hrs, multiple wake from sleeps. I stick to lts on this machine.
[18:34] <Blade-Runner> After installing Ubuntu after win 10 Grub does not appear boot loader does. Fast start in win 10 is disabled what else can i do?
[18:35] <matsaman> Blade-Runner: boot up your install image and come back to the channel
[18:36] <Blade-Runner> ok
[18:36] <matsaman> one possibility is that you installed GRUB somewhere other than the first boot device
[18:36] <matsaman> so the hardware has no reason to boot it
[18:36] <Blade-Runner> nope i didnt choose where to install grub
[18:36] <matsaman> one possibility is that GRUB was installed somewhere other than the first boot device
[19:30] <nedbat> i can see the motd by cat'ing /run/motd.dynamic.  What creates that file?
[19:34] <BlueEagle> nedbat: http://manpages.ubuntu.com/manpages/xenial/man5/update-motd.5.html
[19:34] <matsaman> https://wiki.debian.org/motd
[19:34] <nedbat> BlueEagle: thanks. that says "at login", does that include when I run a command via ssh?
[19:36] <BlueEagle> nedbat: I am not sure that I understand your question.
[19:37] <nedbat> BlueEagle: when I use "ssh my-machine cat /run/motd.dynamic" i see that it changes each time.  So I guess my ssh causes the file to be recreated?
[19:37] <BlueEagle> nedbat: Well, I think I understand the question, but I do not undestand what you are trying to achieve.
[19:38] <BlueEagle> nedbat: The opening of a shell causes the scripts in /etc/update-motd.d to be executed and they update the motd file.
[19:38] <nedbat> BlueEagle: it's partly just curiosity.  I used the ssh command to get a quick look at the state of the machine, and it seemed "live", so i was wondering how it's done
[19:38] <BlueEagle> nedbat: So yes, runnign the command over a ssh link causes the motd to be updated, but it is not ssh that does it.
[19:38] <nedbat> right, i knew it wasn't ssh directly.
[20:06] <Blade-Runner> how do i access irc when using live image?
[20:07] <matsaman> Blade-Runner: just run a client, like hexchat or irssi
[20:07] <matsaman> Blade-Runner: you can even install one on the live OS if one isn't present, but one probably is
[20:07] <Blade-Runner> how do i find it?
[20:08] <BlueEagle> open terminal and type irssi
[20:08] <Blade-Runner> ok
[20:08] <BlueEagle> Blade-Runner: If it is not installed: sudo apt install irssi
[20:08] <CorvusCorax> he's gone
[20:08] <matsaman> Blade-Runner: you can just run the 'dash', I forget the shortcut
[20:08] <matsaman> tab or super or windows key or something
[20:09] <CorvusCorax> maybe irrsi worked ;)
[20:09] <matsaman> =D
[20:10] <CorvusCorax> is there any way to update the kenrel of a live CD / live USB stick?
[20:10] <CorvusCorax> kernel
[20:10] <matsaman> probably, but not likely a way worth pursuing
[20:11] <matsaman> why would you want to?
[20:12] <CorvusCorax> background: I'm normally running ubuntu18 LTS - i run into a kernel bug which i reported, i wanted to see if it also affects ubuntu 20, so i downloaded the live CD / USB stick, booted it, and - indeed - it is also affected... but the kernel of the live CD is not the latest version you'd have on a native install thats fully updated
[20:12] <CorvusCorax> i don't wanna install it, i just want to see if the latest kernel also crashes
[20:13] <matsaman> CorvusCorax: aha, well
[20:13] <matsaman> CorvusCorax: it's very likely you can use most any version kernel with your regular install, for testing
[20:13] <BlueEagle> CorvusCorax: Is that the green "baddie" from that british cartoon that we saw to learn English back in school?
[20:13] <BlueEagle> Sorry, wrong channel.
[20:14] <matsaman> CorvusCorax: https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel
[20:14] <quadrathoch2> CorvusCorax I would suggest to just install ubuntu onto the usb stick, and then go from there
[20:14] <BlueEagle> CorvusCorax: And it was "corvax" anyway, so there's that.
[20:15] <CorvusCorax> hmm, yeah that's an idea. just use a free partition on the stick as install device... it's gonna take forever ;) but who cares
[20:15] <quadrathoch2> CorvusCorax why would it take forever? usb2?
[20:15] <CorvusCorax> corvuscorax is just the latin term for "Raven" - although there's also a band by that name
[20:16] <CorvusCorax> yeah, USB2
[20:16] <quadrathoch2> oh okay :/
[20:16] <matsaman> won't take that long
[20:53] <lusrx> can i get rid of the "printer x has been added" messages in ubuntu 20.04 on every boot? printer does not need to be added every time i start my computer, does it?
[20:53] <lusrx> it's a network printer so ubuntu sees it when it's on the same network.
[20:58] <quadrathoch2> lusrx look into avahi
[21:32] <chieta> is it possible to save ssh detail connection on command line
[21:32] <chieta> password and the key
[21:32] <chieta> on every server i've login
[21:32] <chieta> ?
[21:33] <chmod> Hello I am trying to install go lang and it's giving problems. I did apt install golang-go, but then I run go run something and it gives errors
[21:33] <chmod> src/golang.org/x/net/html/render.go:8:2: cannot find package "bufio" in any of:
[21:34] <chmod> Oh and it's the old version
[21:35] <geosmileus> https://landscape.canonical.com/try-landscape - any alternatives?
[21:37] <chmod> Ok nevermind. I will not do it ubuntu repository way
[21:37] <matsaman> geosmileus: webUI for server management?
[21:37] <chmod> Ok, I got 1.15 from upstream and installed it and now it works fine.
[21:38] <matsaman> geosmileus: for the same reason you don't normally want GUI apps on a server, you don't really want webUIs with root access, IMO
[21:40] <halvors> I am trying to switch root device (/) to use a new disk that has an identical filesystem, i've changed /etc/fstab to do so, reinstalled to grub bootloader and update initramfs, system still boots into old system, am i missing something?
[21:42] <Jonii> Hello. My grub crashes after like 10s in menu, and doesn't react to primary keyboard
[21:42] <geosmileus> matsaman, what is a good software so that I can remotely update/upgrade monitor things?
[21:43] <Jonii> Ubuntu 20.04, it's been acting up in all sorts of ways after upgrading(first from 16.04 to 18.04 and now this)
[22:07] <matsaman> geosmileus: ssh
[22:11] <BlueEagle> geosmileus: I use tinycp for that.
[22:12] <BlueEagle> matsaman: a web ui on ssl can have the same security as a secure shell session tbh.
[22:20] <matsaman> BlueEagle: nope
[22:21] <matsaman> not unless it's only viewable locally
[22:31] <BlueEagle> matsaman: So how is SSH more secure than SSL?
[22:32] <matsaman> BlueEagle: you're talking about logging into a website (webUI) that's protected by "SSL", right?
[22:33] <BlueEagle> matsaman: I am.
[22:35] <BlueEagle> matsaman: SSL (or rather TLS as that is what all the cool kids use) both do public key encryption before any sensitive data is transmitted, and their key strengths are similar.
[22:35] <matsaman> BlueEagle: okay so we're comparing openssh, a 20-year-old industry standard, to whatever random authentication library landscape is using _and_ whatever other _masses_ of code it uses that might have any flaw in it
[22:36] <BlueEagle> matsaman: No, I am saying that a web ui can have the same security as a secure shell session.
[22:37] <BlueEagle> matsaman: I have not made any look into the source code of Landscape to verify the integrity of that system.
[22:37] <matsaman> theoretically it can, sure
[22:37] <matsaman> you can even have a web system that uses only openssh in the backend to authenticate
[22:37] <matsaman> but I've never seen any webUI in the wild do that
[22:37] <matsaman> they always use some other auth lib instead
[22:38] <matsaman> I'm not saying HTTPS and ssh aren't both encrypted, if that's what you mean
[22:43] <BlueEagle> matsaman: You are replying to my comment that "a web ui on ssl can have the same security as a secure shell session tbh" which implies that ssl/tls is inherintly less secure than ssh. While there are webUIs that have attack vectors that are not available on secure shell, that is besides the point I was trying to make.
[22:44] <BlueEagle> matsaman: My experience is that there are way more brute force attempts made at my secure shell server than man webui though.
[22:44] <matsaman> I appreciate that you have clarified that is your point, but my own point was never that HTTPS and ssh are not both encrypted
[22:44] <matsaman> merely that openssh is a much better choice than anything else
[22:44] <matsaman> including whatever library landscape uses for authentication
[22:45] <BlueEagle> matsaman: Then you should have read my comment more clearly to be honest.
[22:45] <matsaman> should've worded it better =)
[22:46] <matsaman> a web ui on ssl can have the same security as a secure shell session only in theory
[22:46] <matsaman> and not because it's ssl, but because theoretically the authentication lib you're using over any protocol _can_ theoretically be as good as openssh's
[22:50] <BlueEagle> matsaman: Wasn't it 2018 when SSH2_MSG_USERAUTH_SUCCESS was still accepted without a request, though?
[22:51] <BlueEagle> matsaman: And when was the last time you rotated your keys anyway?
[22:52] <gordonjcp> evening
[22:52] <BlueEagle> Good evening, gordon.
[22:52] <gordonjcp> I'm trying to boot 20.04 off a USB stick on an oldish PC that doesn't deal with UEFI
[22:52] <gordonjcp> it doesn't even give me a GRUB prompt, just a flashing unhappy cursor
[22:52] <gordonjcp> Alpine boots just fine
[22:53] <BlueEagle> gordonjcp: Which image are you attempting to boot?
[22:54] <gordonjcp> BlueEagle: x86_64
[22:54] <gordonjcp> or, amd64 as it's labelled
[22:54] <gordonjcp> tried "stock" 20.10 and Ubuntu Mate 20.04, which is what I have lying about
[22:54] <BlueEagle> net install? live disc?
[22:55] <gordonjcp> live USB
[22:56] <matsaman> BlueEagle: I don't use keys, so I'll never have that problem
[22:58] <gordonjcp> matsaman: then your ssh is insecure
[23:01] <BlueEagle> gordonjcp: Not neccessarily. A 12 character password with sufficient complexity equals a 1024-bit key, which is a common key length.
[23:02] <matsaman> gordonjcp: nope
[23:02] <matsaman> keys serve a couple of purposes
[23:03] <gordonjcp> BlueEagle: huh, disabling Fast Boot fixed it
[23:03] <gordonjcp> BlueEagle: <shrug>
[23:03] <matsaman> most people use them solely as a convenience, thereby making their entire process for logging in actually less secure
[23:03] <gordonjcp> BlueEagle: a 12-character password with sufficient complexity to have 1024 bits of entropy equivalent is 100% insecure
[23:03] <matsaman> but they can also help with not giving your typed-in password to someone pretending to be your regular remote address
[23:04] <gordonjcp> BlueEagle: because you're never going to remember that jumble of mixed-case and symbols, so you'll either need to use a password manager (Whoops!  There goes the security) or write it down (oopsiedaisy, cardinal sin right there)
[23:04] <gordonjcp> you can somewhat mitigate that with CorrectHorseBatteryStaple passwords
[23:05] <gordonjcp> basically, though, single factor authentication is like a Yale lock
[23:05] <gordonjcp> it'll keep out someone who doesn't actually want in
[23:05] <gordonjcp> like if someone is deliberately trying not to break into your PC, that'll be adequate
[23:06] <gordonjcp> if you actually want to keep people out, MFA or GTFO
[23:06] <matsaman> you're saying it's less secure than keys because you won't remember it?
[23:07] <gordonjcp> that being said, I work in a public safety environment where barely air-gapped machines run Windows XP and have three letter passwords, so <shrug>
[23:07] <gordonjcp> matsaman: you have to trust something else to remember it for you
[23:07] <matsaman> I mean I appreciate the xkcd argument, that normal people might put a sticky note with their password in plain text
[23:07] <matsaman> but I don't do that =P
[23:07] <matsaman> gordonjcp: nope
[23:07] <matsaman> you could just remember it
[23:07] <matsaman> whether you use something that makes it easier (xkcd), or not
[23:07] <gordonjcp> matsaman: a separate password, for every host, from every host?
[23:08] <gordonjcp> matsaman: I think you're missing the point of keys
[23:08] <matsaman> keys don't "keep out someone who doesn't actually want in" either
[23:08] <matsaman> no determined attacker will ever be stopped
[23:08] <gordonjcp> matsaman: yes, they do
[23:08] <matsaman> that's why the government puts their computers in highly defended compounds
[23:08] <gordonjcp> matsaman: here's a host - minecraft.gjcp.net
[23:08] <gordonjcp> matsaman: here's a username - admgordon
[23:09] <matsaman> do you imagine the location of this server is a secret?
[23:09] <gordonjcp> matsaman: here's a password - quetup]Orv5
[23:09] <gordonjcp> matsaman: go to it and good luck
[23:09] <BlueEagle> gordonjcp: A good enough password combined with brute force mitigation should prove insurmountable. As for having a password writen down, if the attack vector you're working against is not internal, then writing down a password is completely legitimate. If someone wants my data and are willing to break in to my house to get it, then they can probably just take the server with them.
[23:09] <gordonjcp> matsaman: you'll find a VM running a desktop, which I'm currently driving through Guacamole
[23:10] <gordonjcp> matsaman: have your CV on my ~/Desktop and you've got a job as a pentester
[23:10] <matsaman> gordonjcp: cool good to know
[23:10] <gordonjcp> matsaman: that is the genuine username and password for it
[23:10] <gordonjcp> fire in
[23:10] <gordonjcp> seriously
[23:10] <gordonjcp> literally anyone who wants a crack it it
[23:10] <gordonjcp> *at it
[23:12] <matsaman> I just got done saying openssh is great, so why would I bother
[23:13] <BlueEagle> gordonjcp: Your minecraft server is so secure I can't even play Minecraft on it.
[23:13] <gordonjcp> BlueEagle: really?
[23:14] <BlueEagle> gordonjcp: Connection was forcibly closed by remote host.
[23:14] <gordonjcp> BlueEagle: hm, maybe the service has stopped
[23:14] <matsaman> BlueEagle: although I would say taking a server is likely to be less effective than some kind of evil-maid-style attack
[23:14] <matsaman> since disk encryption can be pretty tedious to deal with
[23:14] <gordonjcp> seing a lot of valiant efforts from someone on hinet.net
[23:14] <gordonjcp> *seeing
[23:14] <gordonjcp> matsaman: anyway the point is, that username and password is useless without ssh keys
[23:15] <BlueEagle> matsaman: It is at this point you should try to log in to the server and bank that gordonjcp is logging in at the same time and wrongfully authorizes MFA on your login.
[23:15] <gordonjcp> matsaman: since your machine can't prove it has the right to be doing anything
[23:15] <matsaman> gordonjcp: and I could give you a key that's useless without my password
[23:15] <matsaman> because everything I use is useless without my password
[23:15] <matsaman> since nobody knows my password and can't prove they're me
[23:15] <gordonjcp> matsaman: right but a key and password is lazy 2FA
[23:15] <matsaman> 'cause it's a password =)
[23:16] <matsaman> lazy 2fa is more annoying than it is helpful
[23:16] <matsaman> and honestly, 2fa has the same problem we've already talked about
[23:16] <matsaman> if you can break into a person's house, you can damn sure steal their phone or watch them authenticate
[23:16] <BlueEagle> matsaman: A key is used in lieu of a password, so a 1024 bit key is the same as a password like: My milksh akc br1n gs all th3 bois to t#e jard
[23:16] <gordonjcp> BlueEagle: should be on minecraft.gjcp.net:25565
[23:17] <matsaman> or borrow their phone and duplicate it entirely, etc., etc.
[23:17] <gordonjcp> BlueEagle: should be running 1.16.4
[23:17] <BlueEagle> gordonjcp: Who plays 1.16.4 o.+
[23:17] <gordonjcp> BlueEagle: weird and slighly worrying if you can't get in but maybe $teenager has set it whitelisted again
[23:17] <gordonjcp> BlueEagle: what do you recommend?
[23:17] <matsaman> if you allow enough unthrottled attempts to crack even a short password, you're already doing it wrong
[23:18] <gordonjcp> BlueEagle: 35565 has usually got the current snapshot
[23:18] <k_sze> Shit. I fucked up. I have 18.04 on my laptop and I started do-release-upgrade last night before going to sleep, but I forgot to plug my laptop in and there was no big fat warning telling me that I wasn't plugged in.
[23:18] <BlueEagle> gordonjcp: Minecraft peaked at 1.12.2
[23:18] <gordonjcp> BlueEagle: valid
[23:18] <BlueEagle> gordonjcp: any good mods on at least?
[23:18] <gordonjcp> BlueEagle: bone stock, always interested to hear about decent mods
[23:19] <k_sze> I'm not sure whether do-release-upgrade had started installing new packages yet.
[23:19] <gordonjcp> BlueEagle: I set it up as a kind of experiment and now use it instead of Skype to chat to some friends
[23:19] <BlueEagle> gordonjcp: Not sure there are any. I run 1.12.2 with ic2. Got to get buildcraft in too though.
[23:19] <gordonjcp> BlueEagle: nod
[23:19] <BlueEagle> gordonjcp: We should prolly move this to offtopic, though.
[23:20] <gordonjcp> BlueEagle: sure
[23:21] <k_sze> And if I run do-release-upgrade, it tells me I have to finish installing updates first.
[23:21] <k_sze> If I run `lsb_release -a`, it tells me that I'm still on bionic.
[23:25] <tomreyn> k_sze: chances are your apt sources point to the target release already, but packages haven't been installed, yet. there should be a log at /var/log/ though
[23:26] <k_sze> Do I just run `apt full-upgrade`?
[23:27] <tomreyn> *if* my theory is correct (which you should confirm by inspecting the logs), you could just undo the change to the apt sources, pointing them to bionic again, as they used to.
[23:27] <matsaman> was impressed with minetest
And people wonder why Linux does not gain desktop market share. Even one of the most user-friendly distros does not implement something as fundamental as warning the user about plugging in before starting a long upgrade.</rant>