| jamie12 | anyone know how to enable numa when not on by default? I'm running a thread ripper 2950x on desktop because it's well... a desktop but it's technically epyc | 06:11 |
|---|---|---|
| lordievader | Does the thread ripper itself have multiple numa nodes, or do you have multiple installed in your machine? | 06:14 |
| lotus|NUC | jamie12: ^ | 06:17 |
| jamie12 | lordievader it's multidie | 06:17 |
| lordievader | jamie12: https://serverfault.com/questions/877018/ryzen-threadripper-cpu-does-not-report-multiple-numa-nodes | 06:19 |
| jamie12 | I might have found the solution... Asus removed it from BIOS and only changable in ryzen master | 06:21 |
| jamie12 | ugh booting Windows and enabling it to see what happens | 06:22 |
| jamie1 | lordievader: so.... you need windows to enable numa XD | 06:27 |
| lordievader | Interesting. | 06:28 |
| jamie1 | lordievader: basically amd asked asus to remove the bios setting so people would use ryzen master utility to enable or dissable it | 06:29 |
| lordievader | And there is no similar utility on Linux, I suppose? | 06:30 |
| jamie1 | no, sadly there isnt | 06:30 |
| lotus|NUC | or run that ulitity from wine? | 06:30 |
| jamie1 | doesnt work that way, it needs access to core os due to os communication with uefi | 06:30 |
| lordievader | Someone tries that here, via wine: https://www.reddit.com/r/Amd/comments/5zaypz/any_word_on_a_linux_version_of_ryzen_master/?st=jnlcp0te&sh=ab9426a0 | 06:31 |
| jamie1 | also the main reason is most of the sensors dont even work on linux yet for x399 | 06:32 |
| jamie1 | so that would be needed even start with | 06:32 |
| lotus|NUC | https://linuxconfig.org/how-to-overclock-your-amd-ryzen-cpu-on-linux | 06:32 |
| jamie1 | after that we may see ryzen master on linux but that is still questionable due to oems never like to help linux | 06:33 |
| lotus|NUC | also interesting jamie1 https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=051f3ca02e46432c0965e8948f00c07d8a2f09c0 | 06:37 |
| lotus|NUC | !info numactl | jamie1 | 06:39 |
| ubottu | jamie1: numactl (source: numactl): NUMA scheduling and memory placement tool. In component main, is optional. Version 2.0.11-2.1 (bionic), package size 32 kB, installed size 116 kB (Only available for linux-any) | 06:39 |
| jamie1 | lotus|NUC: thats what i use to make sure numa is running right | 06:45 |
| BlackDex | hello there | 10:40 |
| BlackDex | i'm having some issues with lvm | 10:41 |
| BlackDex | getting the message that "Device /dev/sdb1 not found (or ignored by filtering)." | 10:41 |
| BlackDex | there is no filter | 10:41 |
| BlackDex | the disk is an RAID-1 of 2 SSD's | 10:42 |
| BlackDex | i can partition the disk without any issues | 10:42 |
| BlackDex | lsblk shows that the partition is there | 10:42 |
| BlackDex | but the strange thing is that it also shows something called mpath | 10:42 |
| BlackDex | i don't know if that is something wich is causes issues here. | 10:42 |
| BlackDex | Someone have an idea? | 10:43 |
| BlackDex | disks are wiped with dd, but beginning and ending of the disks. | 10:43 |
| BlackDex | s/but/both/ | 10:43 |
| lordievader | BlackDex: What is the output of `sudo pvscan`? | 11:54 |
| ahasenack | good morning | 12:02 |
| rbasak | ahasenack: what would you like me to do about this build-source regression? | 12:18 |
| rbasak | I can revert the MP that broke it. | 12:18 |
| rbasak | Or if you want to try to tackle the implementation of -S within the build command and remove build-source, I'd be happy with that too. | 12:18 |
| rbasak | Or I can leave it for now for someone to fix later? | 12:18 |
| ahasenack | well, I use build-source extensively, and that broke my workflow | 12:19 |
| ahasenack | maybe wait to see if smoser has an insight later today? | 12:19 |
| ahasenack | if not, I'd revert it | 12:19 |
| ahasenack | unless that would reintroduce a bug that is more serious | 12:19 |
| rbasak | OK. I'll revert after smoser has a chance to respond - unless he has an alternate plan. | 12:23 |
| elfranne | anyone know a modern apt repo (create your own repo) ? the ones i can find are unmaintained ... | 14:32 |
| avu | elfranne: reprepro still seems to be active | 14:39 |
| elfranne | this is what i was looking for: https://www.aptly.info/ found it months ago but forgot the name | 14:54 |
| elfranne | but thanks avu | 14:55 |
| joakim_ | Hi got a issue with my ubuntu/netplan, the problem is that after i restarted the (remotely ovh hosted) server all my ip configs disappered when i try to do any netplan command i get that "netplan: fatal error: cannot bind to port 2983, is another daemon running?, exiting." same issue is descriped here https://ubuntuforums.org/showthread.php?t=2400673 but no answers, how do i debug this? | 19:40 |
| sarnold | joakim_: you've probably confused the netplan calendaring tool with netplan the network configuration tool | 19:42 |
| sarnold | joakim_: install netplan.io not netplan | 19:42 |
| joakim_ | sarnold thx, but could you please explain that a bit more?...... ah thx :) | 19:42 |
| sarnold | joakim_: when netplan.io was named we forget to check if the name was already used... | 19:44 |
| sarnold | (not the first time :) | 19:44 |
| sarnold | joakim_: apt-get install netplan.io netplan- should probably get you most of the way to a fix | 19:44 |
| joakim_ | sarnold :/ that what happens i'll try, but what i wundering about is why have it been working for month and then just stopped, i never installed netplan.io manually or remember removing it | 19:45 |
| sarnold | joakim_: hrm :( | 19:46 |
| sarnold | maybe my guess is wrong .. | 19:46 |
| sarnold | try the apt-get command and see what falls out | 19:46 |
| joakim_ | and doint a apt search netplan.io dons't show it as installed on ubuntu 18.04 | 19:46 |
| joakim_ | (have 3 servers + 1 workstation | 19:46 |
| joakim_ | sarnold however your suggestion worked | 19:48 |
| cyphermox | joakim_: so, OVH system that got deployed by them? | 19:48 |
| joakim_ | cyphermox ? | 19:49 |
| cyphermox | you are doing this on a system in OVH, right? | 20:39 |
| cyphermox | was it deployed / installed by them or did you do it yourself? | 20:40 |
| tomreyn | joakim_: ^ | 20:46 |
| Epx998 | cosmic was released? :D | 21:15 |
| tomreyn | good morning, nvidia. | 21:16 |
| Epx998 | buh | 21:16 |
| tomreyn | < not affiliated with anything, i'm allowed to say this. :-P | 21:17 |
| Epx998 | lol | 21:20 |
| Epx998 | neh'veed'de'uh | 21:20 |
| Epx998 | vmware not digging the display after network preseed on a netboot | 21:20 |
| tomreyn | on 18.10 server live installer? | 21:22 |
| Epx998 | 18.10 server iso | 21:22 |
| tomreyn | which one? | 21:23 |
| Epx998 | http://cdimage.ubuntu.com/ubuntu/releases/18.10/release/ubuntu-18.10-server-amd64.iso that one | 21:24 |
| Epx998 | well no wait sorry im jumoing around | 21:25 |
| tomreyn | i think commonly mini.iso is used with pxe. | 21:25 |
| Epx998 | i grabbed http://archive.ubuntu.com/ubuntu/ubuntu/dists/cosmic/main/installer-amd64/current/images/netboot/netboot.tar.gz - let me check md5sum to make sure i got the right one | 21:26 |
| Epx998 | ill grab the ones off the mini iso too and see | 21:27 |
| Epx998 | nope | 21:29 |
| Epx998 | hmm | 21:29 |
| tomreyn | so netboot.tar.gz contains the kernel and initrd for booting. the pxe server would tell the booting 'client' where to load them from, using tftp. once those are loaded, the 'client' would load the installer files from the tftp server, too. what exactly that is depends on what you put on the tftp server. | 21:35 |
| tomreyn | usually somewhere below /var/lib/tftpboot | 21:35 |
| tomreyn | commonly, the /var/lib/tftpboot/pxelinux.cfg/default file on the tftp server would instruct the 'client' what to download (and then load the userspace) from where. | 21:37 |
| Epx998 | i need the right kernel video= option - so far nothing is working past the network preseed | 21:38 |
| tomreyn | i'm not into vmware, so not sure. | 21:39 |
| Epx998 | its trying different overrides until i get one that weorks | 21:39 |
| tomreyn | does 18.04 work with this setup? it should be pretty much the same. | 21:44 |
| tomreyn | vmware recommends to use vmxnet3 instead of e1000e virtual nics with 18.04 (they dont support 18.10 yet) | 21:46 |
| tomreyn | maybe the issue is that you have nvidia hardware passed through to this vm? their graphics card drivers are sadly proprietary and thus, for many years now, often don't work flawlessy and out of the box on linux. | 21:48 |
| Epx998 | there are no gpu's in the chassis other that the onboard server chip | 21:49 |
| Epx998 | Getting a kernel panic tho | 21:50 |
| tomreyn | https://wiki.archlinux.org/index.php/kernel_mode_setting#Forcing_modes discusses what can be used for video= (if you really need to) | 21:52 |
| tomreyn | vga=791 helped me boot 18.04 on virtualbox when they didn't support this version, yet. i didnt get a panic though, just a black screen. | 21:55 |
| Epx998 | prob not even a video issue | 22:09 |
| Epx998 | headache is its in a different DC, so I need to deploy cosmic to another DC to test baremetal | 22:12 |
| Epx998 | ok its my kickstart'= | 22:25 |
| Epx998 | or at least i thought it was | 22:28 |
| teward | RoyK: what're the chances my setting up netconsole will detect the panic and give me any kind of useful information? | 22:44 |
| teward | assuming, of cours,e you can edit the boot cmd of the live images to do that. | 22:44 |
| RoyK | teward: netconsole will probably tell the receiver end what the panic is all about, if there is one | 23:00 |
| RoyK | teward: did you try ctrl+alt+f1 ? | 23:01 |
| teward | RoyK: again, dies and goes nonresponsive at boot, will try again, if i can't get details i'll failover to netconsole, and if that fails too then I'll ahve to see if I can get the normal 18.04 system to boot then extract logs at the filesystme level from a 16.04 instance. | 23:03 |
| teward | 16.04 live desktop instance* | 23:04 |
| teward | RoyK: call trace looks like it dies at the kernel loading level long before it even starts the full boot process :| | 23:05 |
| sarnold | the last time I tried netconsole I didn't actually get anytthing useful out of it. if you're similarly unlucky, maybe the kernel crashdumps would be more productive. (I've never used those, but afaik we still have them :) | 23:07 |
| teward | sarnold: this assumes I can get them | 23:07 |
| teward | does the kernel always crashdump on panics? | 23:07 |
| sarnold | teward: excellent question :) | 23:08 |
| teward | RoyK: ctrl+alt+f1 doesn't seem to help :| | 23:10 |
| teward | because it's not even getting that far | 23:10 |
| teward | hmm, I wonder... *downloads an older 18.04 installer disk from his local mirror to test* | 23:10 |
| RoyK | teward: what about the "magic sysrq" stuff I posted? | 23:10 |
| RoyK | teward: just do-release-upgrade - it'll take you to 18.04 | 23:11 |
| teward | didnt see it probably got eaten by my internet | 23:11 |
| teward | RoyK: um... | 23:11 |
| teward | it was 18.04 | 23:11 |
| teward | to begin with | 23:11 |
| RoyK | I thought you wrot 16.04 | 23:11 |
| teward | can't upgrade to what it already is :p | 23:11 |
| teward | RoyK: no, that's my fallback plan | 23:11 |
| RoyK | ok | 23:11 |
| teward | because that's my emergency recovery USB stick for my laptop that boots the desktop ISO with persistence and my recovery toolkuts | 23:12 |
| * RoyK uses debian when striving for stable systems | 23:12 | |
| teward | ahah! | 23:12 |
| teward | the 4.15 kernels explodify here | 23:12 |
| teward | but not the 4.13 that was 'older' | 23:12 |
| teward | iiiinteresting | 23:12 |
| teward | NO IDEA how to debug that one. | 23:13 |
| teward | better question is why it's using 4.15 :| | 23:13 |
| teward | (this was an upgrade to a newer release so the old kernels must've been around) | 23:14 |
| RoyK | no idea - try to use another kernel, hten | 23:14 |
| RoyK | *then* | 23:14 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!