Criggie | does synergy run under beowulf? I'm doing a dist-upgrade from Ascii and it wants to remove synergy, which is important to me. | 00:40 |
---|---|---|
fsmithred | where did you get it? | 00:41 |
fsmithred | it's not in the repo | 00:41 |
Criggie | yeah its the pro version from symless directly, so its packaged for debian generically. | 00:42 |
Criggie | likewise the steam isntaller is going to be removed if I dist-upgrade | 00:42 |
fsmithred | so you're using the stretch version? There should be a buster version. | 00:42 |
fsmithred | I don't know the proper way to handle those things, but you might have to remove them, upgrade, then reinstall newer versions of them. | 00:43 |
Criggie | hang on - checkling | 00:43 |
Criggie | okay thank you - I get confused by all the various names of versions and their equivalents sometimes. | 00:44 |
fsmithred | beowulf=buster, chimaera=bullseye (still in testing) | 00:44 |
Criggie | yeah but its something I have to go look up, its not top of my head info. | 00:46 |
Criggie | 1569 upgraded, 475 newly installed, 85 to remove and 22 not upgraded. | 00:47 |
Criggie | bah I suspected this update would go badly | 01:09 |
Criggie | Nice to be right now and again :-\ | 01:09 |
fsmithred | did you look at the upgrade guide? | 01:10 |
fsmithred | at devuan.org | 01:10 |
Criggie | yep. | 01:10 |
Criggie | Not a mention of nvidia shennanigans | 01:11 |
Criggie | "your video card is no longer supported by nvidia-driver - would you like to install nvidia-legacy-390xx-driver instead?" | 01:11 |
Criggie | ...okay.... | 01:12 |
Criggie | then the update for "10-nvidia-legacy-check" errored, apt quit, and I'm stuck with the Unmet Dependencies loop | 01:13 |
furrywolf | what're the odds of successfully upgrading a debian 5 system to devuan? | 01:13 |
Criggie | probably shouldn't have let it go so long between updates :) | 01:13 |
fsmithred | 5??? | 01:13 |
Criggie | 5 ? | 01:14 |
fsmithred | isn't/wasn't that lenny? | 01:14 |
Criggie | this was Devuan Jessie | 01:14 |
Criggie | so not really that far back. | 01:14 |
furrywolf | yes, lenny. | 01:14 |
Criggie | I think its nvidia as the initial stumbling block here | 01:15 |
fsmithred | furrywolf, I don't know. dev1fanboy did debian jessie to devuan beowulf (skipping ascii) | 01:15 |
fsmithred | lenny didn't even have udev, did it? | 01:16 |
Criggie | yay update is rolling again | 01:16 |
plasma41 | furrywolf: I recommend upgrading to Wheezy first, but it "should work"(TM). | 01:16 |
Criggie | lenny didn't have systemd either :-PPP | 01:16 |
furrywolf | upgrading to wheezy sounds like more work. :P | 01:16 |
Criggie | its possible the sources aren't available any more either | 01:17 |
fsmithred | jessie is at archive.devuan.org | 01:17 |
plasma41 | Debian keeps good archives. The sources are still available. | 01:17 |
mason | Criggie: Has anyone pointed you at Barrier yet? It's what exists in Synergy's space nowadays. | 01:17 |
fsmithred | deb http://archive.devuan.org/merged jessie main | 01:18 |
systemdlete | I have completely uninstalled mate because I noticed there were 2 different versions going on. Some of it was 1.20 packages, others were at 1.22, part of my experiment to see if an upgrade would eliminate the panel memory leak. | 01:23 |
systemdlete | But now when I try to install mate-desktop-environment, it fails due to some incompatibility with mate-desktop-environment-core, which in turn won't install due to some other incompatibility, etc. | 01:24 |
systemdlete | Is there some "magic" command that will just automatically take care of business for me and simply install mate desktop from the regular repos? | 01:26 |
systemdlete | (I've removed the repos for the 1.22 versions and I've run apt update umpteen times now) | 01:26 |
fsmithred | maybe pin hezeh repo to -1 | 01:28 |
fsmithred | not sure | 01:28 |
plasma41 | systemdlete: I recommend aptitude's interactive dependency resolver. Press 'e' from within aptitude to enter the resolver. Accept and Reject line items in it's proposed resolution by pressing 'a' and 'r', respectively. Press '.' to make it propose a new resolution based on your Accepts and Rejects. To apply a proposed resolution, press '!' then 'g' then 'g'. | 01:32 |
systemdlete | fsmithred: Already removed hezeh repo from sources, and ran update. I'd think it would be all gone by now. | 01:36 |
systemdlete | Was it bad to remove it? | 01:36 |
fsmithred | no, if there are newer versions, they won't be removed automatically | 01:36 |
fsmithred | other way to deal with it is to install older versions from /var/cache/apt/archives with dpkg --force-downgrade -i <package> | 01:37 |
plasma41 | systemdlete: If you use the aptitude interactive resolver, I recommend you leave all repos enabled. The resolver can often provide better resolutions if it's playing with all the cards, so to speak. | 01:39 |
Criggie | Ah ha! The trick is to anwser "yes" to "Install unsupprted nvidia graphics card driver package ? " | 02:09 |
Criggie | this upgrade is so confused - it wants to uninstall xinit and apt-transport-https too. | 02:25 |
systemdlete | http://paste.debian.net/1178565/ | 02:53 |
systemdlete | I ran apt install --fix-broken | 02:54 |
systemdlete | no help | 02:54 |
systemdlete | plasma41: I tried using aptitude, but for some reason, the terminal was being finicky about my keystrokes (uxterm, xterm). It actually thought I wanted to remove aptitude itself! I could not even quit aptitude; I had to kill the uxterm. | 02:56 |
systemdlete | It wouldn't take any of my keystrokes, as if it were ignoring me. | 02:56 |
rwp | systemdlete, I assume you did a fresh 'apt update' to make sure the index "Packages" files are up to date? That's a common problem. | 03:01 |
systemdlete | rwp: I have run "apt update" so many times that I swear my fingers will be falling off soon. | 03:01 |
rwp | What does this show? "apt-cache policy mate-desktop-environment-core" It should show: "Candidate: 1.20.0+5" Does it? | 03:02 |
systemdlete | How come apt won't automatically figure out how to install the core package? | 03:02 |
systemdlete | Take a look at the link I just gave, above | 03:02 |
rwp | I did, and it said: Depends: mate-desktop-environment-core (= 1.20.0+5) but it is not going to be installed. | 03:03 |
systemdlete | It says that it depends on that package (core) but it is not going to be installed. | 03:03 |
rwp | But it is available for me on my Beowulf system okay. | 03:03 |
systemdlete | yes, that^ | 03:03 |
systemdlete | What is going on? Why can't it just reach into the repos and pull it down, and install it? | 03:03 |
systemdlete | Like usual? | 03:03 |
rwp | No problem here: https://paste.debian.net/1178568/ | 03:04 |
systemdlete | I don't doubt that it is fine by your system. It is mine that has a problem. | 03:04 |
rwp | Your questions are what I was trying to determine. | 03:04 |
systemdlete | So what do you think at this point? | 03:04 |
rwp | Correct. Something on your system is the problem. The way to find it is to work through it and debug why. | 03:04 |
systemdlete | That's what I've been trying to do. | 03:04 |
rwp | I think the problem is something between your sources.list and the downloaded Packages files. | 03:05 |
systemdlete | One thing I note is that mate depends on caja | 03:05 |
rwp | The current problem you reported in the paste is that mate-desktop-environment-core (= 1.20.0+5) is not available. Why not? It's damned available to me. | 03:05 |
systemdlete | If I try to install caja, it wants to install the hezeh repo version (1.22) not the regular one (1.20) | 03:05 |
systemdlete | But the hezeh repo is disabled atm | 03:05 |
systemdlete | (and, yes, I did apt update) | 03:06 |
systemdlete | Why does it want to update from hezeh repo if it is disabled? | 03:06 |
rwp | What is the hezeh repo of which you speak? | 03:06 |
systemdlete | Is apt cacheing info someplace that could be blown away maybe? | 03:06 |
systemdlete | hezeh has mate 1.22 | 03:06 |
rwp | I recommend that if you are having problems that you use only the standard and recommended sources.list files. | 03:07 |
systemdlete | I was trying that to see if the panel memory leak might be fixed (it wasn't, so I want to roll back to 1.20) | 03:07 |
systemdlete | Yes, of course. | 03:07 |
rwp | Mixing sources.list sources from various random places has often caused exactly the type of breakage that you are reporting now. | 03:07 |
fsmithred | maybe you still have some 1.22 that's trying to pull something in | 03:07 |
systemdlete | It is just that the memory leak was becoming such a problem, causing the system to crash. | 03:07 |
fsmithred | hezeh is relatively safe - it's run by the person who maintains mate for devuan | 03:07 |
systemdlete | What is the magic for getting a list of all packages installed from hezeh? | 03:09 |
rwp | I also see: "E: Unable to correct problems, you have held broken packages." Do you have "held" packages? | 03:09 |
fsmithred | not sure. | 03:09 |
rwp | apt-show-versions should do it. | 03:09 |
fsmithred | is hezeh in the version? | 03:09 |
rwp | apt-show-versions is not core and so you probably will need to install it. | 03:09 |
rwp | Probably a useful use of the command: apt-show-versions | grep -v -e /beowulf | 03:10 |
fsmithred | there's a way with aptitude search, too | 03:10 |
rwp | Also I often do: apt-show-versions | grep -v -e uptodate | 03:11 |
fsmithred | systemdlete, did you re-enable the repo before you tried aptitude? | 03:14 |
systemdlete | thanks, rwp. That was the magic I needed. Apparently, in purging the hezeh stuff, I overlooked a few packages. | 03:15 |
rwp | \o/ | 03:15 |
systemdlete | fsmithred: No, but I think I am on my way now. | 03:15 |
systemdlete | We will see in a moment when I try to install mate again... | 03:15 |
systemdlete | This is looking much more hopeful. Thanks to you both! | 03:16 |
systemdlete | I wish that the apt tools would give more complete reasons why it will not proceed. The problem here it seems is that there was dependency chaining going down deep into libraries and other infrastructure. But the error gives the highest level of failure, not the lowest, or any other levels. | 03:18 |
fsmithred | aptitude usually gives more information | 03:18 |
fsmithred | and alternatives | 03:18 |
rwp | The crazy thing to me was in your paste that mate-desktop-environment=1.20.0+5 was available but mate-desktop-environment-core=1.20.0+5 was not available. | 03:19 |
fsmithred | aptitude -s whatever | 03:19 |
systemdlete | I also wish that apt's own packaging would include these more useful tools like apt-show-versions and some others like apt-file by default. Having to go installing these just when the packaging tools are not working sufficiently is inconvenient. | 03:19 |
fsmithred | if you know to look for them | 03:20 |
rwp | Well... The problem is one of feature creep. There is always yet another feature that is interesting, useful, needed. But some of them are conflicting. | 03:20 |
rwp | Because even though everyone loves swiss army knives I don't know anyone that actually can practically everyday carry one of these: https://www.gearpersonal.com/largest-swiss-army-knife/ | 03:24 |
systemdlete | I wish pacapt were packaged for beowulf; it was for ascii. | 03:24 |
Criggie | less is more. | 03:24 |
systemdlete | From now on, all such experimentation as this will be done in a scratch VM, not a system I actively use. | 03:27 |
systemdlete | VMs are better for this kind of thing. | 03:27 |
systemdlete | I did not have a VM readily available a week or so ago. Today, I have several generic VMs for cloning. Very convenient. | 03:28 |
rwp | VMs are awesome for testing! | 03:30 |
rwp | I looked up pacapt and it looks like this was why it was removed from Testing: https://bugs.debian.org/886874 | 03:30 |
rwp | FTBFS due to missing build dependencies. A patch was submitted to fix the problem but the maintainer has not responded. | 03:31 |
rwp | I would guess that it would build from source for anyone who wanted it okay. | 03:32 |
systemdlete | "build from source" Heheh. It's a shell script. Single file. Can download it from the source repo for the pacapt project. | 03:32 |
systemdlete | That's how I've been doing it. | 03:32 |
rwp | I didn't look into it that far. I have not every used Arch and so am not for me interested in pacman. But it's okay if others are interested in it. | 03:35 |
systemdlete | I tried arch, but didn't care for it. However, I did like their package took, pacman. | 03:37 |
systemdlete | but we are getting o.t again, sorry | 03:37 |
rwp | I am pretty much happy with apt-get, dpkg, apt-cache, apt-show-versions, apt-file, apt-mark, dctrl-tools, awk, grep, sed, perl, and so forth. | 03:37 |
rwp | I haven't felt the need to learn the newest addition to the familiy the command line "apt" command. | 03:37 |
rwp | dlocate, deborphan, probably a few others. It's a toolbox with many tools available. | 03:38 |
unixbsd | virtualbox fails to install on a new debootstraped devuan ascii from pkgmaster. However, it runs and installs fully on a debootstrap devuan ascii -- but installed from the DVD iso https://mirror.leaseweb.com/devuan/devuan_ascii/installer-iso/devuan_ascii_2.1_amd64_dvd-1.iso ! | 08:02 |
gnarface | i think it's a known issue | 08:36 |
gnarface | not sure, i think you just need a different build of virtualbox | 08:37 |
unixbsd | that's strange that pkgmaster cannot allow to build virtualbox. however on same ascii from dvd. | 10:09 |
unixbsd | the vb deb installs well, but it looks like that some files arent working for vbox. | 10:10 |
n4dir | "however on same ascii from dvd". Huh? | 10:15 |
gnarface | n4dir: he means the copy on the dvd, unupdated (you can use the install isos as apt sources, too) | 10:24 |
n4dir | main thing was that the question and problem are anything but clear. | 10:30 |
unixbsd | On devuan, does this module mount BSD Freebsd or another BSD? 4.9.0-11-amd64/kernel/fs/ufs/ufs.ko | 12:14 |
unixbsd | it seems that ufs does not mount ffs of netbsd. | 12:17 |
unixbsd | installation on virtualbox of cdrom devuan fails when it is given KDE to be installed. | 13:31 |
unixbsd | standard system installation only works. | 13:32 |
unixbsd | kde fails to install on devuan... here a termbin. : links termbin.com/ef82 | 13:54 |
fsmithred | unixbsd, try aptitude - it may give you some other options. Also, maybe try specifying the version to install. | 14:14 |
fsmithred | about virtualbox: it should not be on any devuan CD or DVD. There's no virtualbox in ascii or beowulf. It's in ascii-backports and also in ceres. | 14:15 |
unixbsd | I used the official deb of debian for ascii devuan. | 14:45 |
unixbsd | from www. virtualbox org. | 14:45 |
Helle | unixbsd: I mean that one is not official in the sense that Devuan provides it | 16:08 |
Helle | there is a reason it wasn't included, which was the horrendous code quality and lack of care from upstream | 16:09 |
gnarface | unixbsd: does it have to be virtualbox? qemu is a pain in the ass to set up the first time but it's much better supported | 16:27 |
unixbsd | well virtualbox works very fine. it is ultra polished and well running. | 18:01 |
unixbsd | I know oracle, but still qemu is as "crappier" as vbox, but well, they aree today our bests. | 18:01 |
unixbsd | dosbox is highly well designed in comparison. | 18:01 |
Helle | unixbsd: qemu/kvm is secure, stable and safe, virtualbox has had security issues that upstream refused to address for years | 18:05 |
Helle | (see also Debian bug tracker, it got heated) | 18:05 |
unixbsd | well, I use it on lan intranet... no risk | 18:07 |
unixbsd | qemu is very ultra slow. | 18:07 |
unixbsd | and it does not run well any single OS. | 18:07 |
fsmithred | it's not bad with kvm | 18:07 |
unixbsd | try on a raspberry ;) | 18:07 |
fsmithred | I've switched t... | 18:07 |
unixbsd | I did windows on pi ;) | 18:07 |
fsmithred | nm | 18:07 |
fsmithred | lol | 18:07 |
fsmithred | did you figure out how to install kde? | 18:08 |
unixbsd | https://bellard.org/jslinux/ still it runs even on a browser with little javascript | 18:08 |
unixbsd | well, I have an old debootstrap.tar.gz image, that I use all the time when pkgmaster is dead. | 18:09 |
unixbsd | I did installed netbsd to put Slackware ramdsik with grub2, then I scp my tar.gz and then apt-get isntall kde-standard. but pkgmaster didnt work out. | 18:10 |
unixbsd | I tried the debian debootstrap, and it works all the single time (every time). | 18:10 |
nemo | I had to roll back my SO's virtualbox to 5.2 | 18:22 |
nemo | the new 6.1 requires a feature I was unable to get working in her old XP image. | 18:22 |
nemo | well. either in the host or the XP image - not sure who is at fault yet | 18:23 |
nemo | (Devuan Beowulf) | 18:23 |
nemo | frankly was kind of in a rush to get it working again, she had an urgent deadline and it was just to get a copy of MS Word running | 18:23 |
nemo | on TODO is to look into running that in wine | 18:23 |
nemo | https://forums.virtualbox.org/viewtopic.php?f=6&t=97027 this one to be exact | 18:25 |
nemo | rechecked bios everything seemed fine there, so... dunno. | 18:25 |
Hurgotron | Hi, fsmithred | 18:26 |
nemo | unixbsd: were you using qemue+kvm ? | 18:27 |
nemo | *qemu | 18:27 |
unixbsd | sometimes but ceertainly not on PI raspberry | 18:46 |
unixbsd | nemo: having to use MS is hell. Recently a new update of Microsoft office 365 allows to not start and give an error message. no clue, no help, no solution. welcome MS. SOlution: buy another pc and pray it works. | 18:48 |
fsmithred | hi Hurgotron | 18:49 |
nemo | unixbsd: hm. surprised virtualbox run at all on RPI | 19:12 |
nemo | it was just the comparison of vbox to qemu that perked me up | 19:12 |
nemo | seemed proper comparison would be qemu/kvm | 19:12 |
nemo | at least in terms of performance | 19:12 |
nemo | unixbsd: and yeah, I'd dearly love it if everyone would agree on libreoffice as a common document interchange format, but no love | 19:13 |
nemo | if you want editable docs with history, MS Word is the only game in town. | 19:13 |
nemo | I do wonder if we'd had any better luck if docx hadn't gotten accepted as a standard | 19:13 |
hagbard_ | So, I updated xfce to 4.16 (from devuan unstable), but xfce4-session is the only xfce package that is only available as 4.14 there. I assume this is due to a devuan fork of that package that still needs to be made. Would it be safe to just use the version stright from a debian repo? | 19:44 |
hagbard_ | Just did it anyways. Was installable. | 19:57 |
hagbard_ | Only difference between the debian package and the devuan fork seems to be a systemd recommends of the package, there have been no changes in the actual content itself. | 19:59 |
hagbard_ | So another fork should be trivial, with a single line changed in the control file. | 20:04 |
DashiePie | hi, can anyone help me with an amdgpu issue? | 20:49 |
hagbard_ | maybe, depends on the issue | 20:50 |
DashiePie | well, it's crashing and locking up my computer | 20:50 |
DashiePie | before and after adding amdgpu.gpu_recovery=1 | 20:51 |
DashiePie | if you need the errors, I can pastebin them or something, if you'd like | 20:52 |
DashiePie | https://pastebin.com/UPTSN24F | 20:54 |
hagbard_ | Hmm, no idea. But the amdgpu driver gets lots of updates and fixes all the time, might me worth to try to use a newer kernel (from unstable, or even directly from kernel.org) | 21:04 |
DashiePie | I had a feeling you might say something like that | 21:05 |
DashiePie | what are the risks of installing a newer kernel? | 21:07 |
n4dir | back in the days the risks were rather low, and the kernel was of the few packages you could get away mixing into stable. You still have been warned | 21:08 |
n4dir | Not that sure how much newer unstable will be compared with backports. Usually not that much, iirc | 21:09 |
fsmithred | DashiePie, did you install firmware-amd-graphics? | 21:13 |
DashiePie | double checked, yes, it's installed | 21:14 |
fsmithred | that's all I've got. Sorry. | 21:14 |
fsmithred | if it's new hardware, you might need newer kernel. I think beowulf-backports has 5.9 now | 21:15 |
DashiePie | it's all old | 21:15 |
DashiePie | the newest piece of hardware in here is the gpu, a sapphire hd 7770 | 21:15 |
fsmithred | if it's 2 years old or more it should be supported with the stock kernel | 21:16 |
DashiePie | it's factory overclocked, and maybe that's why | 21:18 |
DashiePie | I remember having to fix this card on windows, too, but I wasn't the one who did it | 21:18 |
DashiePie | I think the biggest thing is that the guy who fixed it underclocked the card | 21:19 |
DashiePie | anyway, I'll see what I can do with my friend who's been helping me and ask here if we're stumped | 21:20 |
DashiePie | thanks for the help | 21:20 |
fsmithred | oh, that might make a difference | 21:21 |
rwp | DashiePie, Have you previously created a /etc/X11/xorg.conf file that might now be stale and in need of an update? | 21:33 |
rwp | For some time now I have not needed to have any xorg.conf file at all and let it dynamically configure itself. | 21:34 |
rwp | But have run across several older installations that had one from before and then over time of not getting any updates they then cause problems. | 21:34 |
DashiePie | I fresh installed this like, last month | 21:35 |
DashiePie | cause I was tired of windows 7 being outdated | 21:35 |
rwp | Exactly what amdgpu are you running? The lspci id for it would be most canonical. "lspci | grep VGA" | 21:35 |
DashiePie | 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Cape Verde XT [Radeon HD 7770/8760 / R7 250X] | 21:36 |
DashiePie | I mentioned it was the hd 7770 from sapphire | 21:36 |
DashiePie | oh, vapor-x edition | 21:37 |
rwp | Being an older card I would hope the AMD open source Radeon driver would handle it okay. I am personally using the AMD open source driver for the HD 5450 right now. | 21:37 |
DashiePie | the Radeon driver gave me vaporwave vomit | 21:38 |
DashiePie | when I opened dosbox | 21:38 |
DashiePie | no games being played, I just opened it and vaporwave vomit | 21:38 |
DashiePie | all over my screen, colors were screwed | 21:38 |
rwp | So you are using the AMD proprietary drivers at this time? | 21:38 |
DashiePie | no, Mesa | 21:38 |
DashiePie | and RADV | 21:38 |
rwp | Also, could you look over the /var/log/Xorg.0.log file and see if there is anything interesting in there? | 21:38 |
rwp | So far I have had nothing but a perfectly functioning experience with the AMD HD 5450 (older card) with a dual monitor setup and the in-kernel main radeon free software driver. | 21:39 |
rwp | I had until recently been using an nVidia card in a similar setup but on that system had to swap over to an AMD card because just about two or three months ago the in kernel nouveou driver in Sid/Ceres became unstable, after literally years of stable running there. | 21:41 |
rwp | And by unstable in my case the nvidia card with the sid driver started to have timeouts where it would freeze for 2-4 minutes whenever anything changed rapidly. Extremely annoying! | 21:42 |
DashiePie | I don't think there's anything in the Xorg logs, old or new | 21:43 |
rwp | Therefore while I was using the newest Linux kernel and drivers available for me that actually caused my working card to stop working. | 21:43 |
rwp | But if you want then I think using the newest kernel is pretty safe thing to do. | 21:44 |
rwp | The subtle danger lies in if there is a remote security vulnerability for the kernel itself. Because when people apply newer kernels like this they rarely keep track and never update again. | 21:44 |
rwp | But generally you should find that everything works okay if you install a newer kernel and give it a try. | 21:45 |
rwp | Or for that matter an older kernel. Since as I experienced with the nvidia driver that new is sometimes broken too and older is often more working functional. | 21:46 |
DashiePie | well, the Xorg logs are unhelpful, nothing's wrong with Xorg, apparently | 21:47 |
DashiePie | I can say it doesn't have anything to do with wine either, as this crashing happened with 0AD as well | 21:47 |
DashiePie | as well as a couple other games that are native I got from itch.io | 21:48 |
rwp | That's reached the end of my knowledge, sorry. If you happen to have a different display card you might tray swapping cards and pushing a change from the current state that way. | 21:50 |
DashiePie | that's what makes this honestly difficulty | 21:50 |
DashiePie | difficult* | 21:50 |
DashiePie | I've only got the one computer | 21:50 |
rwp | Life's hard with only one computer. On the other hand life is hard when one has dozens too. Too much stuff to support! | 21:51 |
DashiePie | I've had this one since 2011 or so, and it was a prebuilt, I added the dedicated and a more powerful psu later for gaming | 21:52 |
rwp | nemo, Just FYI but I am running two systems here both running libvirt both hosting a Windows XP guest image. One is on Buster and the other is a Frankenstein similar to Beowulf. | 21:53 |
rwp | nemo, I am not a fan of VirtualBox on Linux, it's really mostly for MS-Windows, but KVM on Linux works very solid. | 21:54 |
nemo | rwp: good to know... this virtualbox image is very old, and I was kind of disinclined to burn time on it without a good reason | 21:57 |
nemo | but better performance might be one | 21:57 |
nemo | rwp: is it difficult to convert the images? also windows being windows would want to make sure the system presented itself identically | 21:58 |
nemo | rwp: oh, and WRT amd, my experience on beowulf has been very positive with my graphics card | 21:59 |
nemo | I did have some issues on ascii | 21:59 |
nemo | er wait. I have an off-by-one error π | 21:59 |
nemo | I switched to chimaera to get things working | 22:00 |
nemo | beowulf was broken | 22:00 |
nemo | rwp: this being that whole https://m8y.org/tmp/amdgpu.html I struggled with - turns out switching to "unstable" made life oh so much better βΊ | 22:02 |
rwp | DashiePie, See nemo's comment about AMD and Beowulf. So maybe the Chimera/Ceres kernel would be a good test? | 22:02 |
nemo | I recommend it myself! | 22:02 |
nemo | and it's worked great for me overall | 22:02 |
n4dir | switching to unstable in general, nemo ? | 22:02 |
DashiePie | I was thinking about that awhile back, switching to Chimaera, but it seemed from what I could find that Devuan didn't really do a "Testing" version | 22:02 |
nemo | n4dir: yeah. | 22:03 |
rwp | nemo, I don't know how difficult it would be to convert images. I know images can be converted though. | 22:03 |
rwp | And the wonderful thing about VMs is that you can make the new one work all of the while using the previous one. Don't break the old one until you are assured the new one is good. | 22:03 |
nemo | DashiePie: this machine is the living room gaming family TV youtube daughter's video schooling machine. | 22:03 |
n4dir | nemo: as far i can tell: the problem with unstable is not that it is unstable in general. It *can* work like a charme, it *can* break anytime | 22:03 |
nemo | DashiePie: it has been working wonderfully on chimaera - everything that wasn't working well before now is. Better performance, more games functioning, no more struggling with video | 22:03 |
nemo | n4dir: yes. that is completely true βΊ thus my fear | 22:04 |
n4dir | so if someone wants to avoid a breakage in general, then not the best idea. If one can live with it, anytime, then sure | 22:04 |
nemo | I'm just relieved it seems to be working out well | 22:04 |
nemo | n4dir: but fsmithred here said that it was getting pretty reliable so I took the plunge | 22:04 |
nemo | worst case, restoring /home from backups | 22:04 |
n4dir | nemo: yeah, sure. I'd say that myself. But still, some have to avoid breakage | 22:04 |
rwp | unstable broke my nvidia setup just 2-3 months ago. So... | 22:04 |
nemo | oh. not quite *everything* fixed by chimaera. | 22:04 |
rwp | It's hit *and* miss. | 22:05 |
nemo | $ ./a.out 2>&1 | paste -sd '' | 22:05 |
nemo | WARNING: CPU random generator seem to be failing, disabling hardware random number generationWARNING: RDRND generated: 0xffffffff 0xffffffff 0xffffffff 0xffffffff{d2e3494f-b5c5-498f-aed5-dffe0f03d5af} | 22:05 |
nemo | β β β still broken | 22:05 |
DashiePie | I think I'd have to do some work before I switch, but I always kind wanted more up-to-date software | 22:05 |
nemo | however, Qt handles it properly now | 22:05 |
DashiePie | kinda* | 22:05 |
nemo | and I don't have systemd | 22:05 |
nemo | sooo AFAIK those are the only 2 dumb enough to rely on rdrand anyway | 22:05 |
nemo | hm... that paste would've been more readable with ' ' or '|' βΊ | 22:06 |
nemo | (test program being cout << QUuid::createUuid().toString().toUtf8().constData() << endl; ) | 22:06 |
rwp | nemo, I think I would fix that by adding a USB hardware random number generator to the system. They are available easily enough now. | 22:07 |
nemo | rwp: ehhhh it's a home game machine with a ton of good sources of randomness hooked up already βΊ | 22:07 |
rwp | However most of us have been living without a HW RNG on the system for literally decades. | 22:07 |
nemo | yeah exactly | 22:07 |
nemo | rwp: hell. the video camera she uses for school is probably a good source of randomness if I investigated | 22:07 |
rwp | Or one could cobble together a Raspberry Pi and use the onboard RNG there and then inject that entropy to the other system. | 22:08 |
rwp | One of the many USB HW RNG available for purchase now: https://shop.fsf.org/storage-devices/neug-usb-true-random-number-generator | 22:08 |
nemo | rwp: I have a headless "server" sitting in my personal room that has no hardware RNG - it gets its randomness from whatever the usual linux sources are plus an old infrared mouse dangling by its cord π | 22:09 |
nemo | rwp: it gets bonked from time to time and the sensor bounces off whatever it can pick up | 22:09 |
nemo | /proc/sys/kernel//random/entropy_avail says 3519 so I guess it's fine | 22:10 |
rwp | For actual servers that consume a lot of entropy due to the encrypt everything movement I have been installing 'haveged' to stir more bits into the pool. | 22:11 |
nemo | rwp: does HTTPS really burn that much? | 22:12 |
rwp | Yes. If the server is active then every session start will burn a little entropy. Which will drain the pool. | 22:13 |
rwp | And not just https but if it is active for ssh then that too. | 22:13 |
user282069 | ty for sharing | 22:14 |
rwp | If I mention this next it will sound odd but... I admin for git.savannah.gnu.org and it is very active and had empty entropy pool problems until I broke down and installed haveged there. | 22:14 |
rwp | Since member access is all ssh authenticated that means not just https but also ssh and the connection negotiation drains a little bit of entropy for every new connection. | 22:15 |
rwp | This has not happened there yet but... In a perfect world I would always have a HW RNG on the host and make it available to the guest systems too. | 22:16 |
rwp | https://fedoraproject.org/wiki/Features/Virtio_RNG | 22:16 |
rwp | "The generation of random numbers is too important to be left to chance." | 22:16 |
nemo | heh | 22:19 |
nemo | rwp: I thought that HTTPS used urandom | 22:20 |
nemo | rwp: so wouldn't burn the random pool every single session | 22:20 |
rwp | As per your question I am using nginx and OpenSSH and I believe they won't block waiting for entropy but looking at /proc/sys/kernel//random/entropy_avail would show the system having none available to provide. | 22:21 |
rwp | Other random tools would block waiting for entropy and they would be very annoying. | 22:22 |
nemo | m | 22:22 |
nemo | not familiar with nginx | 22:22 |
mason | There's a LWN article talking about the recent direction of Linux randomness: https://lwn.net/Articles/808575/ | 22:23 |
rwp | I highly recommend Nginx. It's a very fast and efficient web server. | 22:23 |
nemo | rwp: I just checked a devuan VM over here at work that serves about half a million distinct connections per day, and its entropy pool is at 3778 | 22:24 |
nemo | let me get a more accurate number on those connections. one moment | 22:24 |
rwp | nemo, Curious if it has an entropy daemon such as haveged or other installed there to stir more into it? | 22:24 |
nemo | rwp: this is a bog standard devuan I installed myself | 22:25 |
rwp | nemo, Curious if it has a /dev/hwrng device there? | 22:25 |
nemo | it could be ofc the virtual host offers hardware randomness instructions | 22:25 |
nemo | would have to check | 22:25 |
nemo | rwp: one moment. unpacking a week's worth of logs for more accurate check | 22:26 |
nemo | rwp: I can count log lines as distinct connections since they are all fetching same url | 22:26 |
nemo | (json datasource) | 22:26 |
nemo | also because main feature of data is very very dumb | 22:26 |
nemo | *fetcher | 22:26 |
ranix | what did you just call me | 22:27 |
rwp | If it is an HTTP protocol then it might be reusing connections for performance reasons. | 22:27 |
nemo | 2020-12-22 was high point at 560k requests | 22:27 |
nemo | rwp: the remote clients are too stupid for that I think βΊ | 22:28 |
nemo | rwp: it's some crappy java thing someone cobbled together and it doesn't even honour Last-Modified or ETag or gzip | 22:28 |
rwp | I have no idea. I am simply trying to keep the entropy level higher than 20 on the systems. And that is all I really know. | 22:28 |
nemo | rwp: hm. JFDDG'd and seems nginx uses urandom first.. | 22:29 |
nemo | rwp: how many connections do you get per day? | 22:29 |
rwp | And I see that in the last upgrade the munin config lost the entropy trend graph. Fixing that now... | 22:30 |
rwp | We rather intentionally don't log most things. So actually I will need to do some digging to figure that out. | 22:30 |
nemo | interesting | 22:31 |
nemo | rwp: you could log without recording IPs and such I guess, no idea what constraints are over there | 22:32 |
nemo | random stackoverflow says urandom can generate 2ΒΉβΆβ° bits before it has to suck up your random again | 22:32 |
nemo | that's a looooot of SSL connections! | 22:32 |
rwp | I don't know. You have given me something to think about. Questions I do not at this moment have an answer. But something is consuming entropy there. | 22:34 |
nemo | looks like SSLRandomSeed is set to use 512 from /dev/urandom in this bog-standard devuan apache install | 22:35 |
nemo | 2^160 / 512 is... way too many connections for me to care βΊ | 22:36 |
nemo | rwp: could some paranoid person have set either ssh or https to use /dev/random instead? | 22:36 |
nemo | in which case, time to trot out https://www.2uo.de/myths-about-urandom/ again | 22:36 |
rwp | It's possible... | 22:36 |
rwp | Things are running happy there for the most part but as time permits I will dig and see what I find... If it is interesting I will try to remember to let you know what I found. | 22:38 |
nemo | yeah. I'm rather curious | 22:40 |
nemo | rwp: in part due to those machines at work βΊ | 22:40 |
nemo | DashiePie: yeah, link to your logs I guess. in here. there's probably people more knowledgeable. I can just compare to my devuan amd cpu + gpu setup | 22:58 |
nemo | DashiePie: how recent is your card though? I had to use AMD's "pro" driver w/ beowulf just 'cause the xorg/linux distro packages weren't up to date enough | 22:59 |
DashiePie | well, I can tell you it's a Cape Verde card, and it's specifically the Sapphire HD 7770 Vapor-X model, and it should've been bought new in 2013, 2 years after I bought this prebuilt | 23:01 |
DashiePie | I disabled the integrated chip on my CPU in the BIOS, so it shouldn't even be a factor | 23:01 |
nemo | oh wow | 23:02 |
nemo | ancient cards should be fine on linux | 23:02 |
nemo | I mean... apart from being generally ancient and crappy βΊ | 23:03 |
nemo | yeah toss some logs this way | 23:03 |
DashiePie | yeah, well, this is a generally ancient and crappy factory overclocked card | 23:04 |
DashiePie | I think I need to read an archived log, maybe | 23:04 |
DashiePie | https://pastebin.com/CS5wyb6d here's dmesg.0 | 23:10 |
nemo | [ 29.821079] r8169 0000:02:00.0: firmware: failed to load rtl_nic/rtl8168e-2.fw (-2) | 23:10 |
nemo | [ 29.821225] firmware_class: See https://wiki.debian.org/Firmware for information about missing firmware | 23:10 |
nemo | oh nic. meh | 23:10 |
nemo | got excited for a moment | 23:10 |
nemo | there's the amdgpu. neeeevermind | 23:10 |
DashiePie | https://pastebin.com/StNdkBDS | 23:11 |
DashiePie | here's Xorg.0.log.old | 23:11 |
DashiePie | yeah, this mobo has a built-in wifi card or something | 23:11 |
DashiePie | I never used it | 23:11 |
nemo | "amdgpu 0000:01:00.0: kfd not supported on this ASIC" β this might be worth google/DDG maybe | 23:12 |
nemo | that feels like some old kernel thingy | 23:12 |
nemo | [drm] BIOS signature incorrect 2f 7 β too | 23:12 |
nemo | otherwise, dunno, looks like it loaded firmware and is fine | 23:13 |
nemo | what about your old xorg log? | 23:13 |
DashiePie | first result for the first one was ROCm | 23:13 |
DashiePie | I posted it | 23:13 |
nemo | oh | 23:13 |
DashiePie | I'm not trying to use ROCm at all, though | 23:13 |
nemo | huh. glamoregl? | 23:14 |
DashiePie | I have no idea what that even is | 23:14 |
nemo | yeah, I guess it's normal. compared against mine | 23:14 |
nemo | just wasn't familiar | 23:14 |
nemo | you can tell my level of familiarity w/ guts of X π | 23:15 |
DashiePie | that BIOS signature line disturbs me | 23:15 |
DashiePie | I found a link about that, though the guy was having a bit different of an issue | 23:15 |
nemo | xorg log looks normal | 23:15 |
nemo | AFK to make dinner | 23:15 |
DashiePie | https://old.reddit.com/r/linuxmint/comments/6avr4r/amdgpu_driver_ring_0_test_failed/ | 23:16 |
DashiePie | this link | 23:16 |
nemo | DashiePie: old.reddit.com++ | 23:21 |
DashiePie | I don't like nu reddit | 23:21 |
DashiePie | old reddit for life | 23:21 |
nemo | anyway. yeah. that doesn't tell me anything apart from it maybe being your mobo/bios - and I've definitely had issues w/ stoopid mobo manufacturers before | 23:22 |
nemo | most notable one backing up BIOS automagically to HD w/ no disable and no support for >2TB HDs | 23:22 |
nemo | was like. why is my HD instantly corrupted after first post-install reboot... | 23:22 |
nemo | eventually figured out I had to put a tiny sacrificial drive on primary IDE and manually switch boot partition on startup | 23:23 |
nemo | or I'd get corrupted HD and have to restore from backup again | 23:23 |
DashiePie | who made that mobo? China? | 23:23 |
nemo | ummm I might have it lying around. one moment | 23:23 |
nemo | Gigabyte board, Award BIOS | 23:24 |
nemo | https://www.vogons.org/viewtopic.php?t=59627 | 23:24 |
nemo | DashiePie: oh, and my first gen chromebook I accidentally set on fire due to stupid driver fail | 23:25 |
nemo | so yeah. hardware folks suck. esp when they get involved in software π | 23:25 |
DashiePie | I'm not sure what to say | 23:25 |
nemo | or rather I should say *when* otherwise they are probably awesome | 23:25 |
DashiePie | I'm not shocked, but I am disappointed | 23:26 |
DashiePie | "my only solutions are bad and worse, which one do you want?" | 23:27 |
nemo | you know what they say. "The Q in Internet of Things stands for Quality" | 23:28 |
rwp | They also say, "The S in IoT stands for Security." | 23:30 |
DashiePie | the I in IoT stands for Imagination | 23:31 |
DashiePie | you can either imagine it working or you can't | 23:31 |
DashiePie | I failed, I'm sorry, I'm bad | 23:31 |
* DashiePie goes to my corner | 23:31 | |
nemo | rwp: I like that one better due to the IoS | 23:37 |
nemo | so it resonates subconsciously | 23:37 |
rwp | iOS? | 23:42 |
nemo | InternetOfShit | 23:49 |
nemo | its a blog where he reviews awful devices | 23:50 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!