[03:51] <lotuspsychje> good morning
[07:04] <ducasse> good morning
[18:21] <lotuspsychje> tomreyn: looks like a cheap flaky machine to me, https://incoherency.co.uk/blog/stories/ubuntu-18.04-ideacentre-310s.html
[18:23] <tomreyn> it's definitely a low price platform
[18:24] <lotuspsychje> havent seen his dmesg neither yet
[18:24] <tomreyn> lotuspsychje: the system does work fine as sold, though, just not with this seperately acquired nvidia card, on linux (does work on windows).
[18:25] <lotuspsychje> tomreyn: i also find it weird, 440 doesnt showup his ubuntu drivers list
[18:25] <tomreyn> he's posted several journalctl -b    by now.
[18:25] <lotuspsychje> and -server drivers showup
[18:26] <tomreyn> here's what he gets when he boots with the 450 drivers installed:  http://termbin.com/601ye
[18:26] <tomreyn> same thing but with recovery kernel chosen on grub menu (which also sets "nomodeset"): http://termbin.com/e115
[18:28] <lotuspsychje> tnx
[18:34] <daftykins> pentium silver :( just wow what people try and polish
[18:35] <lotuspsychje> daftykins: whatever the user trys, his nvidia drivers seem to fail
[18:36] <tomreyn> this is the exact model: https://psref.lenovo.com/Detail/IdeaCentre/ideacentre_310S08IGM?M=90HX000PUK
[18:36] <daftykins> yep got that page open
[18:36] <daftykins> there'll be insufficient power i bet, to run a discrete card
[18:37] <tomreyn> 65W is really not much, not sure
[18:37] <tomreyn> i dodn't notice this
[18:37] <tomreyn> *didn't
[18:37] <daftykins> see how it says 65W adapter? so it's unlikely to push the full power spec through the PCIe slot
[18:37] <daftykins> what model is it he's putting in, something like an nvidia 1050 Ti?
[18:37] <lotuspsychje> https://certification.ubuntu.com/hardware/201912-27573 a bit similar lenovo show 18.04 with oem kernel here
[18:38] <lotuspsychje> also with the intel 605
[18:38] <tomreyn> 10de:1d01 is the nvidia
[18:38] <lotuspsychje> PCI-E Nvidia GT 1030
[18:38] <tomreyn> yes
[18:38] <tomreyn> but he said it works on windows
[18:39] <daftykins> i see a 'nvidia-uvm' claiming to load, what's that about o0
[18:39] <daftykins> hrmm 30W card - nvidia's specs say you should have a 300W PSU
[18:40] <daftykins> for me personally, i'm not too persuaded of the claim Windows is fine - it's a bad combo, would refuse
[18:41] <daftykins> 10W CPU, so say it were loaded, that's already 40W of the PSU taken, with the remaining 25W to run disks, motherboard silicon, anything else in there... lol :D
[18:42] <lotuspsychje> 278 euro lenovo :p
[18:46] <tomreyn> looks like he calculates power consumption differently
[18:46] <daftykins> with a hope and a prayer?
[18:47] <tomreyn> :)
[18:49] <daftykins> it might be borderline, seems like the CPU is an SoC there, so the 10W could be full package power, with no further on the board
[18:50] <daftykins> makes so much more sense buying the right tool for the job though
[22:38] <tomreyn> out of interest, i just looked for alternatives to dreamscapes' https://psref.lenovo.com/Product/IdeaCentre/ideacentre_310S08IGM?ch=90HX000PUK (which sells for €265 on ebay) the computer  https://geizhals.eu/?cat=sysdiv&xf=11069_integrierte+Grafik+(IGP)~11121_2~13393_4~445_8192~447_allgsosd~6764_AMD~9718_Dali~9718_Renoir~9724_15&v=k&hloc=at&hloc=de&hloc=pl&hloc=uk&hloc=eu&sort=p&bl1_id=30&togglecountry=set
[22:42] <daftykins> seems like you're keen on AMD ;)
[22:43] <tomreyn> he wanted low power, and better graphics than intel igp
[22:43] <tomreyn> to me, this means amd apu
[22:44] <daftykins> no reason for it to be integrated
[22:44] <tomreyn> also, yes, i prefer anything over nvidia
[22:44] <tomreyn> power savings are a reasn for it to be integrated
[22:44] <daftykins> i find it funny how you think nvidia = trouble :)
[22:46] <tomreyn> i find it sad how i came to think this.
[22:51] <tomreyn> anyways, the first system on the list has twice the ram, a proper ssd (his had 220 MB/s write speed max.), an extra hdd, certainly faster cpu threads (but two cores less), 802.11 ac,(4) USB 3.2 Gen 1 Type-A, HDMI v1.4b/HDCP 1.4, RS-232
[22:52] <tomreyn> and is cheaper :)
[22:53] <daftykins> 'pentium' was all you need know about the current one
[22:55] <tomreyn> :) well, it's like the amd "athlon" nowadays, isn't it
[22:56] <tomreyn> a marketing name, usually referring to the slower / power efficient and, most of all, cheap ones
[22:59] <daftykins> think it's a greater difference than that
[23:06] <tomreyn> yes, probably, they're also not the most current ones. leftover stock which needs clearing. ;-)
[23:07] <daftykins> didn't look too closely, it's foreign! plenty of options going a few generations back on ebay and the like
[23:08] <daftykins> what was the result anyway, were they given up on?
[23:14] <tomreyn> he gave up, saying he'll use his old one instead
[23:17] <daftykins> PC or graphics card?
[23:20] <tomreyn> pc
[23:20] <tomreyn> which was an older low energy pentium
[23:20] <daftykins> TJ-: \o greetings, all well?
[23:21] <TJ-> Hiya :) just arrived in the office for an overnighter!
[23:21] <daftykins> ooh another one
[23:22] <tomreyn> i'm about to arrive in my bedroom for an overnighter
[23:22] <tomreyn> see you!
[23:22] <TJ-> yeah - closed down for everything except tech deployment for 5 days from yesterday, we got a bit behind schedule today so trying to catch up
[23:22] <TJ-> tomreyn: :D
[23:22] <daftykins> TJ-: i had an amusing situation with the encrypted VM, thought you might enjoy hearing... i had the UUID wrong in crypttab, so when the machine booted it would only take to the -25 kernel that was preinstalled post-debootstrap, the real /boot never got mounted to /boot, so instead a /boot as a subdirectory of / was where newer kernels were getting put xD easily fixed up
[23:22] <TJ-> delay mostly caused by Windows Server defrag STILL running 12 hours later!
[23:23] <TJ-> daftykins: ouch - been caught out by that myself once or twice
[23:23] <daftykins> hah, what are you doing messing with that nasty thing, a completely degraded small business server 2011 machine i fixed up was much quicker even with 50% fragmentation
[23:24] <daftykins> TJ-: more seriously though i might have to give up, as i want the module xenfs loaded so the guest utilities can talk to dom0, but on reboot modules keep disappearing from the disk - and i don't understand why at all O_O
[23:25] <daftykins> it's not a custom addition, just one that comes in from linux-modules-<version>
[23:26] <TJ-> we're wiping Windows out completely; currently we have a HyperV Linux on the HP server with PCI pass-through of 2 NVMEs + 4x Ethernet. Now we're going to make it dual boot so need to shrink the OS partition, and the aim is repeat the Windows shares using NFS off the NTFS, and later totally wipe out Windows
[23:26] <TJ-> modules disappearing?
[23:27] <daftykins> yep, it's a magic trick - i can get it loaded once; sudo depmod, modprobe xenfs, but on reboot the actual files are gone from the disk and i have to apt install --reinstall the linux-modules-version package to get them back
[23:28] <TJ-> I can think of one cause; if the module is in a package linux-modules-$VERSION but you do not have the generic non-versioned package that depends on that installed, when a kernel update is added the related linux-modules won't be installed
[23:29] <TJ-> otherwise, you've got a storage failure of some kind. what is the underlying file-system ? LVM + ext4?
[23:29] <daftykins> mmm, nah the modules-generic is present and correct... yeah matched your guide identically but then it's a VM atop XCP-ng
[23:30] <TJ-> give me a specific version you've had problems with so I can check
[23:30] <daftykins> mmm 5.4.0-28 is current on that VM i think
[23:31] <daftykins> the /xen/ files don't disappear on a kernel upgrade though, solely a reboot
[23:32] <TJ-> so linux-modules-5.8.0-25-generic has an rpdends to linux-image-5.8.0-25-generic
[23:32] <TJ-> well, that suggests you've got a file-system problem of some kind. Sure you've not got a stray tmpfs ?
[23:34] <daftykins> hrmm during package installs i think i saw some weird messages about 'leaks' due to LVS invocation, something along those lines
[23:34] <daftykins> some comments online claimed they are harmless
[23:34] <TJ-> that would be grub-install
[23:35] <TJ-> the package installs to /lib/modules/$VERSION/ so if that entire director sub-tree disappears, or files within it do, then there's a storage system fault
[23:35] <TJ-> is there hardware RAID behind it?
[23:36] <daftykins> i'd moved it off of one and onto a single 2TB M.2 PCIe drive during these antics
[23:36] <daftykins> i could move it back and see if anything changes
[23:36] <TJ-> is it possible you've got some problem in fstab entries causing the wrong file-system being mounted?
[23:37] <mgedmin> does debsums -c linux-modules-$(uname -r) find other missing modules, or just the xen one?
[23:37] <daftykins> it's a pretty basic two-liner, LUKS_BOOT and the main / ext4 - can't really see i'd get as far as i do if it were too wrong
[23:38] <TJ-> you can use "dpkg --verify linux-modules-$(uname -r)" instead of debsums :)
[23:38] <daftykins> mgedmin: mm saw you post that the other day, but i wasn't in a position to work on the system then
[23:38] <TJ-> daftykins: OK --- trying to rule out the obvious
[23:38] <daftykins> i will go and fire it up, it'll be a good 15 mins before it's up
[23:38] <TJ-> daftykins: as it is a VM makes me wonder if the hypervisor is doing something bad like not flushing to the real storage
[23:41] <daftykins> not had a problem on my other VMs during setup and configuration
[23:42] <TJ-> I'd check /var/log/apt/history.log in case something in the background is unning a package remove operation!
[23:42] <TJ-> it depends if the entire package content is going missing, or just part of it, and which part
[23:43] <daftykins> it's a select number of files under there, but i can't give the path until it's fired up, lost from memory :D
[23:43] <TJ-> if apt/dpkg still shows the package as installed but files from it are missing then something weird is going on. I'm wondering if the file-system journal isn't being flushed to the file-system (e.g. the ext4 journal being lost so next time the file-system is mounted (at boot) the file-system has effectively been rolled back
[23:47] <daftykins> the issue with that theory, is that when i added xe-guest-utilities - that change persisted
[23:49] <TJ-> yep - losing journal would show up in the kernel log at boot
[23:50] <TJ-> what module are you losing again?
[23:50] <daftykins> xenfs
[23:52] <TJ-> the only thing I can think is you've got some DKMS module that is replacing/removing the original, but again that's doubtful in your scenario
[23:52] <TJ-> but DKMS does move package modules
[23:53] <daftykins> weird, i am missing linux-generic linux-headers-generic linux-image-generic today
[23:54] <TJ-> how 'missing' - reported as not installed?
[23:55] <daftykins> ah i had remembered incorrectly, i'm on 5.4.0-52-generic
[23:55] <daftykins> yep, i'm going to hold off of adding them back just for a moment though
[23:55] <TJ-> you sure it's not using -lowlatency rather than -generic ?
[23:55] <TJ-> "apt list --installed 'linux*' "
[23:55] <mgedmin> hey TJ-, do you know how to interpret dpkg --verify's output?  https://paste.ubuntu.com/p/XWkpJYrFpd/
[23:55] <mgedmin> "??5??????" is not exactly self-explanatory
[23:56] <daftykins> positive because i installed via linux-generic post-debootstrap
[23:56] <TJ-> mgedmin: 5 mean md5sum
[23:57] <daftykins> ok so in my VM, /lib/modules/5.4.0-52-generic/kernel/drivers/xen contains only "xen-front-pgdir-shbuf.ko" whilst there should be plenty more there
[23:57] <TJ-> mgedmin: it's rpm format output, with 9 checks. ? means it wasn't performed
[23:57] <TJ-> mgedmin: so '5' means md5sum didn't match
[23:58] <daftykins> in that case i just got 5's across everything
[23:59] <daftykins> /lib/modules/5.4.0-52-generic/kernel/drivers/xen/xenfs/xenfs.ko - is the specific one i'm after
[23:59] <mgedmin> TJ-: thanks!