51[00:45:01] *** Quits: thiras (~thiras@replaced-ip) (Remote host closed the connection)
52[00:45:34] <eric23> I found a command in root user history that I swear I never used before. Is it probable the computer has been compromised or some service ran it as root? the command is checkrestart and I do not have the debian-goodies package installed.
53[00:45:54] *** tijara_ is now known as tijara
54[00:46:34] <petn-randall> eric23: And you're 100% sure nobody recommended to run it, and it just failed because the package is missing?
55[00:46:58] <petn-randall> Because "checkrestart" doesn't sound like a program someone would run when compromising a root account.
65[00:52:55] <eric23> yes I am curious about that directory.
66[00:53:26] <petn-randall> Either the root user or some program ran as root user would have created it. This isn't created on installation of packages.
79[01:03:41] <pasiz> is there something that is not debian.org
80[01:04:13] <sney> from that and the fact that it's called 'jail' this sounds like you or some script might have tried to debootstrap a container
81[01:04:29] <freem> I have recompiled some debian packages. How can I make it so that debian does not considers the resulting .deb as "older" than the original one?
84[01:05:03] <freem> I could lock them, sure, but it's not very practical (I'm not warned about potential updates)
85[01:05:16] <sney> freem: change the package version, iirc if you add a + between the 1.2.3-1 and your custom string, apt will treat it as greater than 1.2.3-1
86[01:05:39] <freem> in the control file?
87[01:05:43] <sney> yes
88[01:06:05] <freem> is there's a canonical way to avoid conflicting with a potential update from debian?
89[01:06:28] <petn-randall> freem: Actually, just add a new entry in debian/changelog with a higher version number. Or just pin the packages if you definitely don't want any updates from Debian repos.
90[01:06:42] <freem> petn-randall: <freem> I could lock them, sure, but it's not very practical (I'm not warned about potential updates)
104[01:11:37] <freem> ok, bad luck, no way to version things so that "this is foo debian version, locally patched". Guess I'll have to risk conflicts then
109[01:12:49] <petn-randall> freem: 1.2.3-1+freem will never show up in Debian.
110[01:13:00] <freem> true enough
111[01:13:07] <petn-randall> Or 1.2.3-1.1~freem
112[01:13:17] <sney> freem: there isn't a one-shot trick to always accomplish that, but you can still do it, you just need to be specific about what version you use for your local package.
113[01:14:01] <sney> ~freem is a good tip, since the ~ is what stable-backports uses so that those packages will still be upgraded by the next stable version.
141[01:32:32] <Gerowen> Does anybody know which script for apcupsd is responsible for the occasional terminal broadcast messages that occur if you're SSH'd into a device during a power outage?
155[01:48:44] <frdg> I was wondering how it would be possible to download this package as a deb package (I know nothing about deb packages but the format I want looks like `<name>_<version>_<arch>.deb`). I am not on debian but I need this package in as a .deb. replaced-url
156[01:49:19] <sney> frdg: sure, click on the architecture from the links at the bottom
157[01:49:51] <themill> (that's approximately the .deb filename except for some escaping rules for 'version')
158[01:49:55] *** Quits: black_ant (~antilope@replaced-ip) (Quit: simplicity does not kill)
159[01:50:21] <frdg> sney: yes I have tried that but then I get brought to this link where I see nothing that seems like what I want: replaced-url
160[01:50:34] <frdg> oh haha I get it now
161[01:50:38] <sney> ... yes
162[01:50:49] <frdg> I spent far too much time thinking this was something else
174[02:00:45] <alex11> if you transfer a new hard drive to a computer and back up your data to that drive you need to edit fstab to get the system to recognize it right?
344[04:58:13] <growly> alright this is a long shot but i wanna see if anyone else has ever run into this: the igb driver, when connected to my (netgear gs316) switch, drops its connection every ~15 mins
345[04:58:54] <growly> i get a 'NIC Link is Down/Up [...]' message periodically. when i connected another switch, this didn't happen
607[07:28:16] <UncleKiwi> Hi there, Im being lazy and want to be able to logon with ssh to a debain 10 system using ssh as root with a password - what changes do i need to make to make that possible
760[10:40:38] <gholinbrown> well one last time. if i installed a package does the OS verify its signature against signatures that can already be trusted
761[10:40:50] <gholinbrown> considering if the package was retrieved across http
766[10:42:21] <gholinbrown> is it automatically verified before installation? keeping in mind i may not have the keys, it's just whatever came with the .iso
767[10:42:36] <jelly> gholinbrown, the OS does not verify the signature of the package _itself_, it verifies the package against the signed list of packages on the repo
768[10:43:02] <jelly> gholinbrown, that verification is done at download time, before installation
788[10:48:24] <jelly> the most manual and problematic step is verification of install media. It's somewhat complex to get .iso, get checksum files and their signatures, get the right gpg key with some amount of trust and verify the signature and the checksums
791[10:49:18] <gholinbrown> for .iso i usually just do the hash check. and assuming the https connection to the list of hashes was solid, then the hashes had to be the real ones
792[10:50:21] <jelly> well yeah, if you're getting the hash from the same site you'd get the gpg key, if one is compromised the might will be as well
793[10:50:33] <jelly> the other* might be compromised* as well
795[10:51:45] <jelly> mind you, debian-archive-keyring is also a package that gets upgrades. Its upgrades become important over the years because keys for debian 11 repos might differ from keys used to sign debian 9 repos
815[11:04:42] <jelly> gholinbrown, well you don't need debian-keyring, that's all the developers and maintainers. debian-archive-keyring is enough to get updates and is preinstalled.
823[11:15:14] <jelly> I have a memory leaky systemd service. Is there a way to restart the service gracefully when it reaches some amount of RSS used, and a higher limit to SIGKILL and restart it?
871[12:13:45] <ratrace> jelly: see, that's why PHP sucks. uWSGI for example has exactly an option for that. reload on RSS. since uwsgi can ALSO handle php, perhaps you can take a look at using that instead of PHP-fpm?
881[12:15:59] <serafeim> ratrace: eh yes. i was using it for like 5 years and was glad when got rid of it
882[12:16:11] <ratrace> it's very simple to set up and use. and by that I don't mean the stupid shell init script that's convoluted to hell and back and to hell again
883[12:16:13] <jelly> ratrace, this is precisely the thing a useful service menager would have functionality for
884[12:16:23] <ratrace> serafeim: and we're using it to power our python flagship web apps for 10 years now
885[12:16:37] <serafeim> ratrace: you'd get better milage from gunicorn
886[12:16:41] <ratrace> jelly: yes, so maybe systemd-oomd
887[12:16:43] <serafeim> uwsgi is filled with crap
888[12:17:01] <ratrace> gunicorn has nowhere near the features of uwsgi
889[12:17:12] <serafeim> gunicorn does 1 thing well. i don't care about all that uwsgi features
890[12:17:13] <jelly> ratrace, is it in stretch?
891[12:17:31] <ratrace> serafeim: that's perfectly fine. but also, you can't make blanket statements about uwsgi like that, then.
892[12:18:11] <serafeim> ratrace: as i said i was using it for many years. it kept throwing C dumps for no particula reason. debugging uwsgi errors was hell
893[12:18:43] <serafeim> there is *no* reason to use some C middleman to serve python code
909[12:21:40] <ratrace> uwsgi is an application server
910[12:21:50] <serafeim> ratrace: it also can be used to do like 100 other different, not needed things
911[12:22:06] <ratrace> like tomcat is for java, uwsgi was initially for python but its application servicing is not strictly limited to WSGI, it can to FASTCGI and there' splugins for php
913[12:22:30] <ratrace> PERSONALLY, I would never use uwsgi for anything but python, where it excells and was designed for. but I mentioned thta as a possibility as it CAN be a php application server and has the features that jelly needs.
914[12:23:23] <serafeim> in any case. i had used it and thankfully i abandoned it
915[12:23:30] <serafeim> if it serves you then keep using it
916[12:23:35] <ratrace> well that's fine :) you don't like it, that's okay
917[12:23:46] <serafeim> when you get the C stack overflow dump you'll remember me
918[12:24:10] <ratrace> I really hope you understand how dumb that statement is
919[12:24:27] <ratrace> you can get "C stack overflow" in nginx too. methinks it already had some.
920[12:24:59] <serafeim> ratrace: ok. i did not have problems with nginx only with uwsgi. YMMV
921[12:25:00] <ratrace> you can also get it in cpython, you know, the actual code that interprets your gunicorn? it's written in C
922[12:25:24] <avu> we should ust rewrite it in Rust, I hear that solves all problems
923[12:25:27] <serafeim> lol
924[12:25:34] <serafeim> i'd stick with pure python for this
925[12:25:45] <avu> so, C
926[12:25:52] <avu> your machine doesn't execute "pure python"
927[12:25:53] <serafeim> faster speeds *are not actually needed* for a wsgi server
928[12:26:37] *** Quits: mandeep (uid394387@replaced-ip) (Quit: Connection closed for inactivity)
930[12:30:21] <jelly> clearly you should run that gunicorn on Jython and avoid all that C :-D
931[12:30:27] <ratrace> indeed, "pure python" is powered by code written in C, thus susceptible to "C stack overflows". and python HAS had CVEs over the years, and will likely have more.
942[12:37:59] <ratrace> I mean bottom line, it's fine if you don't like how uwsgi works. but if you base that on "C stack overflow" as a reason, you should reconsider what you think you know about python software :)
944[12:40:11] <mrjpaxton[m]> To be fair, CPython is written in both C and Python. It's not like C is a bad or slow programming language to use. Lol.
945[12:41:53] <ratrace> jelly: shame, seems like systemd-oomd will only send SIGKILLs to monitored cgroup ... I was really hoping it'd have the signal configurable at least, or have an Exec-like option to run an arbitrary command
957[12:46:40] <ratrace> simplicius: if you're using a Desktop Environment, surely it has some Settings for that. Otherwise, xrandr can be used for Xorg. No idea about wayland. Ultimately, it should be automatic.
958[12:47:20] <simplicius> I cannot see the bottom bar
959[12:47:27] <ratrace> and also there's xorg.conf(5) but I'd check that as a very last resort, it's very likely you won't need that
960[12:47:29] <simplicius> xrandr gives me this output replaced-url
961[12:48:15] <ratrace> check the knobs of your monitor, maybe it's misconfigured
962[12:48:25] <ratrace> is it a 1080p screen?
963[12:48:42] <simplicius> I tried out but I cannot control the vertical position
968[12:51:36] <ratrace> see if xrandr's --pos option can help the value given is in AxB format where A and B are offset integers for X and Y coords respectively. I'd assume you want negative Y
969[12:52:10] <ratrace> but ultimately that's a hack. your monitor is connected over digital, not analogue, connection. if anything, it's the monitor's fault
972[12:53:10] <ratrace> well "new" doesn't in any way mean "100% fault proof"
973[12:53:59] <simplicius> When I swtiched on the pc with the monitor for the first time it looked good
974[12:54:08] <ratrace> hardware broken on purchase is not really impossible. for example, they even have a dead pixel tolerance which states that it's likely your new monitor will have dead pixels. indeed, last two monitors I bought (AOC now, LG before that), have had at least one out of the box
975[12:54:17] <simplicius> then it did some autoajustement that messed up everything
976[12:55:09] <ratrace> soooo..... we went from XY problem through XYQPEBKAC and we're now at at PEBKAC? in translation of your "how can I adjust screen resolution" problem, you actually mean "how can I fix the mess I caused?"
977[12:55:22] <ratrace> why, clearly, just do the reverse? :)
992[13:03:29] <ratrace> jelly: cron or a timer is ultimately what I'd use, if the leak is gradual over time so you can either put a constant time on reload, or, time RSS checks and issue a reload after a treshold
993[13:03:51] <ratrace> (since you say it's not PHP but some other ... daemon thingy)
994[13:04:57] <simplicius> for example which value could I use under xrandr --pos?
999[13:07:06] <ratrace> simplicius: how about 0x-200 and see what happens?
1000[13:07:46] <simplicius> xrandr --pos 0x-200 ?
1001[13:07:55] <ratrace> simplicius: I just remembered... which GPU do you use? nvidia for example has its own controls for screen geometry, mostly useful in multimonitor setups but you can adjust single ones too
1007[13:14:24] <marek_> Hi, I just managed to boot my Jessie on a new motherboard: Asrock B550M Pro4. Mostly works but internet went down.
1008[13:14:29] *** marek_ is now known as omarek
1009[13:14:41] <ratrace> jelly: I'm looking through cgroups docs and examples .... fascinating that linux has no ability to control the signal (systemd is just a consumer of that same API), but it CAN register notifiers, so programs can be written to listen to these notifications and act before OOM. I'd be disappointed if such programs don't already exist, regardless of systemd-oomd
1010[13:15:04] <omarek> The mobo says it uses RTL8111H. It appears the driver is in stretch and buster, but not jessie.
1011[13:15:29] <omarek> I'm searching on this page: replaced-url
1012[13:15:37] <ratrace> jelly: (incidentally, that ability, to specify service reload on a limit, has been in freebsd for ages.... so it's even more fascinating it's not already part of linux regular init/service management)
1013[13:16:02] <omarek> Do I have this correct? Is the simplest way to get my internet working again to upgrade/install Buster?
1016[13:16:54] <ratrace> omarek: since jessie is .... oldoldstable, I'd reckon yes, the simplest way is to just get to Stable, that's soon to become OldStable, so your current setup will become OldOldOldStable...
1017[13:17:21] <jelly> ratrace, devs don't care about specific processes / services on linux any more, granulation work is done at container level and horizontally scalable apps are cared about
1039[13:22:25] <omarek> Is Realtek the most linuxhated company after Nvidia?
1040[13:22:54] <jelly> ratrace, you see how important they effectively are, from the lack of tooling. It's not just tools, I have yet to find a non-djb service that responds well to hitting limits
1041[13:23:11] <ratrace> omarek: seems those two packages don't have deps (except dkms which you probably already have installed), so you can sneakernet the .debs in from elsewhere,
1042[13:23:20] <ratrace> omarek: someone here calls it RealCrap
1045[13:23:47] *** Joins: magyar (~magyar@replaced-ip)
1046[13:23:49] <jelly> powerdns hitting limit? Doesn't die. Just hangs. systemd does not restart it. So... I have a cron job.
1047[13:24:10] <omarek> Well at least I got those dual rank ECC memory modules running.
1048[13:24:18] <ratrace> jelly: oh shush. I'm already slowly growing into sns hater and am seriously considering to drop all that in favor of very, extremely simplistic systems even if that means no-features-BSD
1049[13:24:36] <ratrace> the more advanced these systems become, the more dumb and moronic they are
1058[13:27:07] <simplicius> the xrandr command diddn't affect anything
1059[13:27:20] <jelly> so in the meantime people learn how those bits interact and put it in the official docs, Arch Linux wiki
1060[13:28:02] *** debhelper sets mode: +l 1189
1061[13:28:03] <omarek> ratrace: So we're both pedantic :-).
1062[13:28:05] <ratrace> heh "official doccs, the Arch Linux wiki" ... so true )
1063[13:30:01] <ratrace> jelly: I like systemd because it unifies a lot of linux (kernel) APIs into simple to use unit files. but .... as time passes, I see features tossed into it because "someone at RH needs it" with no clear big picture development mode, like the systemd-oomd for example, will likely grow configurable signal sometime in the future when someone does a PR for it
1064[13:31:30] <ratrace> that it doesn't _already_ have it, is just clear sign of the haphazard development model.
1066[13:32:10] <omarek> ratrace: It's too late to return this mobo because I had the bright idea to order from 2 separate shops.
1067[13:32:13] <ratrace> I wonder at which point will my liking of systemd's unified feature set become overpowered by my disliking of the dev model and overall bad implementation
1069[13:32:57] <ratrace> at least in EU you have the right to return the product within 14 days, no questions asked.
1070[13:32:59] <omarek> ratrace: In this country, 14 days.
1071[13:33:09] <ratrace> yea
1072[13:35:46] <jelly> ratrace, it still better that anything else, but I hope someone at some point will get annoyed enough to fork or reimplement with lots of tiny tools instead
1076[13:37:14] <ratrace> s6 is a serious contender. if only someone would find a few million bucks to toss behind s6 development ; instead of knee-jerk herp derp reactionist devuans and friends
1077[13:37:49] <jelly> a bit hard to package for debian at this point, but I last took a look 2-3 years ago
1078[13:38:44] <jelly> it's nice and seems to be modeled well
1079[13:39:05] <ratrace> yea
1080[13:39:07] <jelly> but I don't have even 100k to toss at them
1105[13:57:53] <simplicius> I tried this without success xrandr --output DVI-0 --pos 0x-200
1106[13:58:30] <ratrace> simplicius: did you try some other value? maybe positive? as I suggested, that -200 was just an assumption and an example to try and see
1107[13:59:17] <simplicius> when I give the command nothing chages
1120[14:06:03] <ratrace> omarek: and did you also install the firmware? dmesg | grep -i firmware and if it mentions inability to load up firmware, you're missing it
1121[14:06:19] <ratrace> firmware-realtek package, I assume
1122[14:07:34] <ratrace> simplicius: did you check your Desktop Environment's Settings?
1141[14:24:03] <omarek> ratrace: I just installed 'firmware-realtek'. No difference. Grepping firmware yields 3 lines, but none that reminds me of realtek or network or RTRL8111H.
1145[14:25:52] <omarek> ratrace: r8169 was enabled when I booted with my new hardware. Package description for r8168-dkms says it it covers RTL8111H, and r8168 should only be used if r8169 fails to work.
1146[14:26:33] <ratrace> at this point you'd really have to somehow pastebin or upload to imgur, the relevant outputs from dmesg, to see which nic is it really, and what firmware it may or may not want
1147[14:26:44] *** Quits: fflori (~fflori@replaced-ip) (Remote host closed the connection)
1185[14:59:21] <azeem> fourstepper: do you mean, why Debian switched from ext3 to ext4 at some point, or why Debian chose extX over xfs?
1186[14:59:30] <fourstepper> i suppose :)
1187[14:59:33] <azeem> for the latter, xfs on Linux was not a thing when Debian started
1188[15:00:09] <azeem> for the former, there just hasn't been a serious debate each time we bumped the ext-version I think and/or extX was considered the better/safer choice
1195[15:01:51] <jelly> ext3 over ext2 was an easy choice. ext4 over ext3, well it probably helps that tytso is running debian and maintaining the tools
1219[15:17:38] <omarek> If that fails, I'm going to try installing debian testing 'bullseye' for the 5. kernel.
1220[15:17:57] <omarek> Some internet fairies report success with that.
1221[15:18:10] <jelly> !bdo kernel
1222[15:18:10] <dpkg> Newer kernels for Debian stable releases are available from the <buster-backports> repository. After modifying your sources.list, run «apt update». To install the current backported kernel: «apt -t buster-backports install linux-image-`uname -r|sed 's,[^-]*-[^-]*-,,'`». To list available backported kernel image packages: «aptitude search '?narrow(~nlinux-image,?origin(Debian Backports))'».
1228[15:20:12] <jelly> my (google chrome) browser is having issues with dpaste.org, page unresponsive, searching finds 2x the number of results for "eth" or none for "r8169", weird
1245[15:40:21] <omarek> jelly: Also going for a walk. You have some other above ^^ if you'd like to try. I'm leaning towards installing Bullseye on a new nvme drive. It may be the less complicated way in the long run.
1298[16:18:21] <ratrace> samba35: no. tried with and without the /sys/bus/ path?
1299[16:18:31] <ratrace> I think it's not needed, but in some cases it might be needed.... the device path
1300[16:18:52] <samba35> with bus
1301[16:19:15] <ratrace> btw ... I assumed you did this on a live system, does your motherboard support PCI hotplug?
1302[16:19:56] <samba35> yes system is hotplug enable
1303[16:20:08] <samba35> check with cat config kernel
1304[16:20:50] <samba35> i am trying pci passthrorought
1305[16:20:55] <ratrace> samba35: what do you mean? the kernel does have the options, but are you 100% sure the motherboard supports it? not all do. and some that state that do, don't really
1306[16:21:15] <samba35> ok
1307[16:21:24] <ratrace> samba35: ho-kay. that's XY now. so let's start at the beginning. what are you doing and how exactly. I'm doing pci-passthrough for nvidia for my gaming VM, so I might help
1308[16:21:28] <samba35> how do i check motherboard support hotplug
1309[16:21:42] <ratrace> samba35: in the motherboard's specs. not sure if dmidecode can show that
1316[16:23:34] <ratrace> anyway. if you want pcipassthrough, hot pluggable is not what you need, you just need to specify pciids of devices you wanna pass, for the vfio-pci driver
1317[16:24:14] <ratrace> note one important thing. the _entire_ PCI lane will be affected. you can't just assign _some_ of pciids, on the lane, you need all of them
1340[16:30:41] <ratrace> there's nothing to it, you just specify the netdev to be a virtio-net device, for qemu. if you're using libvirt, then you need to specify that via the xml config for the VM, or use an UI.
1349[16:32:44] <mo1991> Hello, I am having issues with the mouse/keyboard not resuming after hibernation. It seems to happen only after the machine hibernates for a long time. Mouse / keyboard do come back if I hibernate manually, and then resume shortly after. Can anyone point me in the right direction? This is a laptop running debian testing, with a hybrid graphics intel/nvidia.
1350[16:32:54] <samba35> ok
1351[16:32:55] <mo1991> Resume after suspend works fine
1352[16:35:29] *** Quits: Mister00X (~quassel@replaced-ip) (Quit: "I'll be back" — Arnold Schwarzenegger)
1353[16:36:25] *** Quits: lucad111 (~lucad111@replaced-ip) (Remote host closed the connection)
1414[17:25:28] <Manis> It doesn't need to be persistent, btw. So something like pam_group would be good-enough. I couldn't get pam_group to work, though.
1415[17:26:27] <petn-randall> Manis: I added your user to the group. It didn't persist. Done!
1416[17:26:52] <Manis> Persist across sessions that is (of course).
1423[17:33:48] <jelly> Manis, if your NSS reads and shows ldap users, you can add them to local groups without any issues.
1424[17:34:52] <Manis> jelly: What I forgot to mention is that the users are dynamic. Say a new user is added to LDAP, I don't want to go and modify all my /etc/groups.
1506[18:30:08] <sney> backports are not handled by the security team, but once a package is backported, it's common for it to get updated when the corresponding testing package is updated.
1508[18:30:27] <sney> so if there's a security patch in the testing/sid version, backports will usually get it too.
1509[18:30:29] *** Quits: filohuhum (~filohuhum@replaced-ip) (Remote host closed the connection)
1510[18:30:39] <ratrace> it's not in full freeze yet, mind you, just for build essentials and transitions
1511[18:31:01] <sney> right, I was generalizing
1512[18:31:14] *** Quits: sparky4 (~sparky4@replaced-ip) (Remote host closed the connection)
1513[18:31:37] <omarek> Wait, scratch that. I don't need a live CD. I have a separate SSD to install on, I can just fall back to the SATA without internet.
1514[18:32:28] <omarek> I'm finding that with many technical things doing things the hard way ends up being less work.
1515[18:32:29] *** Quits: koniu (~koniu@replaced-ip) (Remote host closed the connection)
1554[18:50:43] <omarek> Hmm, wait, what's the point for me to use a live firmware CD if I already have buster? Because scripts might be better at handling drivers than me?
1555[18:50:43] <ratrace> goot bot
1556[18:51:59] <jelly> omarek, you hinted you wanted to try bullseye
1598[19:10:07] <jelly> add a -v for: (quoting the manual) If the verbosity level is 1 or more, then all the explanations aptitude can find will be displayed, in inverse order of relevance.
1599[19:10:27] <jelly> you will have to man aptitude to see what happens if you -vv
1611[19:23:42] <jelly> Bombo, if you don't have aptitude, you can --dry-run a removal: apt-get -s remove packagename, and see what the removal would have taken away
1620[19:26:28] <sney> the full ,kernels output is 2 lines and full of a lot of other gunk, I was trying to avoid that. uname -r from the bullseye system I'm typing on: 5.10.0-3-amd64
1622[19:27:12] <omarek> When I tried to use a non-free driver, I did 'dpkg -i r8168-dkms', then I manually unloaded the module r8169.
1623[19:28:02] *** debhelper sets mode: +l 1197
1624[19:28:14] <jelly> ,help kernels
1625[19:28:16] <judd> (kernels [--release <squeeze>]) -- Outputs the kernel versions in the archive, optionally restricted to one release. Note that semi-major releases like etchnhalf are treated as separate releases.
1629[19:28:40] <sney> yep, I'm assuming it's on themill's todo list for bot maintenance
1630[19:29:13] *** Quits: mezzo (~mezzo@replaced-ip) (Quit: leaving)
1631[19:29:43] <sney> omarek: it's easy to skip rebuilding dkms modules for a newer kernel, just don't install the headers package. you will need to remove the blacklist for r8169 though.
1632[19:29:59] *** Quits: ayekat (~ayekat@replaced-ip) (Quit: watching this channel in stealth mode now)
1660[19:47:50] <omarek> sney: Internet stopped working when I swapped mobo, CPU, and RAM. I only kept SATA drives, the case, PSU, keyboard, mouse, monitor.
1661[19:48:15] <omarek> sney: what do you suggest I could still do to give r8169 a chance?
1662[19:48:41] <sney> omarek: oh, I believe you. this one sounds like a dud. maybe there's a very minor factory defect on the motherboard or something.
1663[19:49:15] <omarek> sney: Yesterday I spent several hours scratching my head because I didn't push my RAM down hard enough.
1664[19:49:24] <omarek> It's more likely a user error with me I'm afraid.
1674[19:52:34] <sney> omarek: I think your plan of buying a new nic, like an e1000, is a good next step. even if the r8169 can work, the intel is still an upgrade, and they're like $30
1675[19:52:37] <ax562> but as sney said, it might have gave you a clue it wasn't in proper
1676[19:52:38] <omarek> ax562: But there were other possible cases, someone got those 2 leds glowing because he had drives on the same cable.
1687[19:54:47] *** Quits: DarthKnight (~DarthKnig@replaced-ip) (Remote host closed the connection)
1688[19:54:53] <omarek> I'm a bit of a perfectionist (not really a good thing) and it leaves a bad taste that I have an unused component in my mobo.
1689[19:55:09] <ax562> omarek I'm not sure if this was stated, but when dealing with exposed electronics it necessary to make sure you do not have charged static electricity.
1690[19:55:17] <omarek> But I don't use the Vega chip either. I got the Asrock mobo because all of them support ECC.
1691[19:55:46] <omarek> ax562: Do you mean I should finger my PC?
1692[19:56:15] <ax562> they sell ground wrist bands which can keep you grounded or occacsionaly touching something do discharge like a a screw on the back of your tower etc.
1695[19:56:52] <Delf> Is it possible to install Debian on a drive without any partition?
1696[19:57:01] <ax562> it's very easy to destroy electronics seeing that static electricity normally comes with high current
1697[19:57:18] <ax562> Delf yes it will ask to partition while install
1698[19:58:05] <ax562> I recommend partitioning yourself before the install and just instructing debian installer where to install. That's what I normally do (less chance of errors during install).
1699[19:58:07] <Delf> But I don't want to partition
1700[19:58:13] <ax562> then don't
1701[19:58:23] <ax562> the install can and will do it for you.
1702[19:58:50] <sney> delf, are you saying you want to install to the whole disk with no partitions at all? why?
1706[20:00:08] <sney> the installer won't let you do that, I suppose you could debootstrap it, but if simplicity is what you want then just make 1 big partition that uses the whole disk
1707[20:00:33] <ax562> ^
1708[20:01:33] <sney> also if your system is efi you need multiple partitions for that to be able to boot. this is a technical requirement. the installer can set this up for you automatically as well.
1709[20:01:51] <omarek> It's like Intel is having the last laugh at my expense. I bought a Ryzen 5 5600.
1716[20:03:49] <Delf> Why wont the installer let me?
1717[20:03:58] <omarek> Delf: Do you want LVM?
1718[20:04:06] <Delf> No
1719[20:04:36] <sney> because the installer targets normal configurations. there is a fairly wide range of "normal", but using the whole disk without partitions is not within that range
1720[20:05:01] <Delf> What about if I use expert mode?
1721[20:05:29] <sney> partman is partman. expert mode just lets you skip installer steps and override stuff.
1722[20:06:07] <sney> maybe if you booted in expert mode, switched to a terminal, set up the disk and created the filesystem, the rest of the installer would detect it and be able to continue the install?
1723[20:06:11] <Delf> Does that mean I can skip the partman step and carry on with the installation?
1726[20:06:53] <sney> this is pretty far off the mark of "simplified", but it might work, you are free to try anything with your computer
1727[20:07:09] <omarek> sney: What's e1000? Any models I should check?
1728[20:08:16] <sney> omarek: e1000 is the controller and the name of the kernel module. Intel usually sells them as "Pro 1000" or something like that. but really if you filter by intel and gigabit, any results you get should be good
1731[20:08:42] <omarek> My needs are pretty basic network-wise. I don't want to overspend. I stream movies sometimes and watch a lot of youtube, but not a netflix subscriber etc.
1732[20:09:14] <sney> this nic series has been around for a very long time, so it should be easy to find them cheap.
1741[20:14:17] <ax56234> omarek with static electricity its really hard to identify if something wennnnnt wrong, you would know if it happened, you would have felt static discharge.
1744[20:15:56] <sney> ESD is worst in dry environments, so if the humidity is around 50% or higher where you are, it's less likely to be a real concern.
1745[20:16:23] <sney> still, you can get a grounding strap to be extra sure while working on your computer. hardware repair shops usually require their techs to use one.
1746[20:17:35] *** Quits: chupawan (~chupawan@replaced-ip) (Remote host closed the connection)
1747[20:17:53] <omarek> ax56234: Never felt anything like that, no suspicious sound or spark. I have this hardware for 2nd day now I would remember.
1758[20:19:01] <ax56234> oh ok, yeah, it would have been something you felt, it just could be a bad card DOA , at this point, time to get dirty and spend that thirty :p
1761[20:21:15] <Delf> sney: I created a file system in the console and selected that drive and it seems to be working. But now I don't know if it was necessary. I'm going to try again but this time without creating a file system in console.
1764[20:26:58] <Delf> sney: It was necessary. But expert mode was not necessary. I just had to mkfs.btrfs /dev/sdX and get back to the installer and hit detect disks and it correctly detected the btrfs drive.
1765[20:27:28] <sney> you don't have to share the play-by-play, but feel free to ask if something breaks
1875[21:27:06] <n4dir> growly: cause that name is taken ... minus the hyphen though
1876[21:27:37] <TRS-80> Seems my colleagues in Armbian seem to be having bit of Debian specific question. I am embarrassed to say, though I have been using and promoting Debian for few years now, I have still not gotten around to delving deeply into all the internals and policies. And from my reading of the following thread, apparently the same from some of my colleagues (including guys much more knowledgeable than me
1877[21:27:38] <TRS-80> an Armbian internals). Anyway, if anyone could perhaps offer any insight, it would be appreciated: replaced-url
1887[21:35:55] <neilthereildeil> why is it better ot have a different partition for /tmp and /home, rather than adding everything to 1 root parittion?
1888[21:36:27] <TRS-80> you can wipe and reinstall OS partition if anything goes wrong, without disturbing your data
1889[21:36:34] <sney> neilthereildeil: /tmp can live in memory, and memory is faster than disk. it's nice to have /home separate for recovery purposes.
1890[21:36:46] <TRS-80> and that
1891[21:36:47] <sney> I don't know about universally "better" but there are good reasons to do it, depending on your environment
1892[21:36:57] <TRS-80> right, it's just an option
1893[21:37:20] <karlpinc> neilthereildeil: If (if) you use LVM, you can resize individual partitions and having separate ones is less of a hassle.
1894[21:37:49] <TRS-80> in Armbian for instance, we mount /tmp to RAM to also decrease wear on the flash memory (eMMC or sdcard)
1895[21:38:36] <n4dir> i just put it all on one partition.
1896[21:38:51] <jelly> what is "better" depends on the use case
1897[21:39:46] <jelly> we separate /opt and /var on servers, /home only on multiuser systems
1898[21:40:20] <karlpinc> (It's nice to have a shared home directory across multiple systems....)
1910[21:43:19] <n4dir> don't use such nasty words in here.
1911[21:43:23] <sney> I wouldn't share a full /home because of configs and dotfiles, but ~/Documents sure
1912[21:43:35] <jelly> I just rsync a minimal HOME setup to each new machine
1913[21:43:38] <TRS-80> Now, "reproducable systems" are nice, but I achieve that by keeping my dotfiles in a git repo wherein I have a Makefile that will create any number of defined system levels (basic shell only, full desktop, etc.) based off of a fresh Debian install as a base.
1914[21:43:51] <jelly> git would be an improvement.
1915[21:44:05] <greycat> shared $HOME is a traditional configuration that works well if you're careful, and don't run any Desktop Environments
1916[21:44:05] <karlpinc> Ansible, in git. :)
1917[21:44:12] <TRS-80> jelly: if you only knew how far behind I was on commits :D
1918[21:44:39] <jelly> greycat, anything can work well if you're careful
1919[21:44:48] <greycat> yeah, it's not something I would recommend to a novice
1920[21:44:53] <jelly> windows 95 works well if you're careful
1921[21:45:29] <TRS-80> careful + lots of knowledge (referring to GNU/Linux here /especially/ when you add a DE into the mix)
1922[21:45:34] * TRS-80 is still learning
1923[21:45:55] <greycat> a DE will probably choke on lack of file locking over NFS, etc.
1924[21:46:06] *** Quits: omarek (~slawuta@replaced-ip) (Quit: Lost terminal)
1926[21:46:35] <greycat> traditional WMs are probably less dependent on fancy Linux crap, having been developed when a shared NFS $HOME was a lot more common
1927[21:46:48] <TRS-80> yeah which is why I took and recommend above mentioned approach
1928[21:47:02] <TRS-80> but even then, it has been a slow and steady learning process
1930[21:47:44] <Strogg> I've got python3.8 and 3.9 installed on my system.. I guess 3.9 is used by default for python3.. but I ahve a lot of packages which are failing to install because of the changes to HTMLParser module. can I just uninstall python3.9 and use 3.8 by default? When I go to apt remove 3.8, apt wants to remove everything pyhon3 as well
1931[21:47:46] <mspe> like sharing firefox and thunderbird between more recent and older versions between the different distros is a recipe for trouble!
1958[21:50:08] <lastshell> Is there a way to use wpasupplicant with the new network name convention, im currently have a custom rc.local called for a systemd unit that runs this:
1959[21:50:12] <neilthereildeil> this has to be high performance
1960[21:50:14] <lastshell> ip link set wlp4s0 name wlan0
1961[21:50:14] <lastshell> ifup wlan0
1962[21:50:37] <n4dir> iirc encrypton minus lvm was a bit of a clusterfuck
1963[21:50:39] <TRS-80> Strogg: You are having problem with system packages? Or user installed?
1964[21:50:42] <jelly> Strogg, since 3.8 is gone, you'll have to acquire it some other way and put it maybe into /opt/python3.8
1966[21:51:47] <Strogg> TRS-80: this is just python modules that haven't made it into debian yet. I guess I'll have to setup a python environment for it, but that's a bit of a nuisance
1967[21:51:54] <Thete> don't break debian lulz
1968[21:52:31] <Strogg> I don't blame debian for this.. this is on python's head and removing functionality everyone was using. I was just wondering if debian had a neat way around the issue. you never know and can't hurt to ask. hehe
1969[21:52:38] <Thete> that's no fun
1970[21:53:28] <mspe> neilthereildeil: you can have encrypted data on a server
1971[21:53:35] <TRS-80> recalling the times I seem to have broken Debian, I seem to recall Python being involved (along mith my own disregard of above link) but I could be wrong and nowadays Debian seem to default pip3 install to --user which I think it a lot more sensible
1972[21:54:02] <neilthereildeil> nah i dont wanna encrypt anything
1973[21:54:27] <jelly> Strogg, if it's an often used module, you might want to file a bug report to get it reinstated if possible
1974[21:54:33] <karlpinc> TRS-80: Mixing package managers (pip + apt) is always a bad idea.
1975[21:54:46] <Strogg> TRS-80: oh yeah. same. anything I install, I put in my own ~/usr. but that doesn't help me if the module itself won't even install. hehe
1976[21:55:25] <jelly> karlpinc, which is why we do it in venv / virtualenv
1980[21:56:12] <TRS-80> as long as you don't sudo, pip3 seems to install to ~/.local/... which I find a lot more sane and easier than the myriad "now this is the one true way, no now it's this one, no now this one" sort if confusing nonsense I experienced in past with Python
1981[21:56:17] <karlpinc> I prefer venvs over docker.
1982[21:56:29] <Thete> can venvs inside containers tho
1983[21:56:37] <karlpinc> Of course they're not the same thing.
1984[21:56:42] <Thete> I'm teasin
1985[21:56:52] <karlpinc> ;)
1986[21:57:01] <Strogg> nah. Iv'e got a couple things installed in /opt (mostly proprietary toolchains), but anything in my system is debian
1987[21:57:14] *** Quits: gormenghast (~gormengha@replaced-ip) (Remote host closed the connection)
1988[21:57:35] * Strogg resumes kicking python3.9 in the shins
1989[21:57:58] <Thete> aww, what's wrong with 3.9, I like it
1994[22:04:37] *** Quits: Jerrynicki (~niklas@replaced-ip) (Remote host closed the connection)
1995[22:04:50] <jelly> Thete, keep at least the majority of your comments in this channel to constructive ones. Quips and non-support -> #debian-offtopic
2006[22:14:24] <Strogg> ok. I'm less annoyed at python3.8 now. apparently when I installed python3.9, it didn't also upgrade python3-setuptools. So the setuptools module was trying to use methods from python3.8 and failing. manually upgrading python3-setuptools fixed the issue
2029[22:23:42] <ax562> you have to add the non free repos and you could do "apt search xxx" or "apt-cache search xxx"
2030[22:23:43] <judd> No packages in buster/amd64 were found with that file.
2031[22:23:44] <n4dir> jaggz: i find version 9 and 11
2032[22:23:48] <greycat> apt-cache or apt will search whatever sources you've configured it to use
2033[22:23:48] <jaggz> oh, I thought libcudart10.1, on my system, came from non-free bpo, but now I see it's only listing its source as /var/lib/dpkg/status
2034[22:24:22] <jaggz> so sometime that got on here and stayed. interesting
2035[22:24:26] <greycat> ,v libcudart10.1
2036[22:24:27] <judd> No package named 'libcudart10.1' was found in amd64.
2037[22:24:41] <jaggz> tensorflow seems to require 10.0 of a bunch of these cuda libs
2038[22:24:52] <greycat> ,v libcudart10
2039[22:24:53] <judd> No package named 'libcudart10' was found in amd64.
2040[22:25:13] <greycat> *sigh* it is SO obnoxious trying to guess a library's package name
2041[22:25:48] <greycat> ,v libcudart9.2
2042[22:25:49] <judd> Package: libcudart9.2 on amd64 -- buster/non-free: 9.2.148-7
2043[22:26:01] <greycat> so buster has libcudart9.2 and bullseye has libcudart11.0
2046[22:26:53] <jaggz> what I resorted to was installing some python thing "conda install cudatoolkit=10.0", which installs those libs in my miniconda directories, then setting export LD_LIBRARY_PATH=/home/jaggz/miniconda3/pkgs/cudatoolkit-10.0.130-hf841e97_6/lib
2047[22:27:07] <jaggz> thanks
2048[22:27:24] <jaggz> but, is it .so obnoxious, greycat ?
2049[22:27:49] <jaggz> or sometimes .a.bnoxious
2050[22:28:08] * jaggz groans and decides he needs coffee.
2061[22:31:23] <greycat> jaggz: well, that library probably came from somewhere other than Debian, unless it was in testing or unstable for a brief time, in which case you could find it on snapshot
2062[22:32:54] <sney> yep, nvidia-cuda-toolkit 10.x was in testing starting 2019-07
2084[22:43:30] <ax562> jezbel the best way is to go to blutooth devices, select which device you want to send file to and then use the "send file" function.
2103[22:56:02] *** Quits: SergioCabral (~SergioCab@replaced-ip) (Remote host closed the connection)
2104[22:58:41] <neilthereildeil> i didnt see a place to force /boot to be bootable in the GUI installer, but i saw the EFI partition is automatically marked bootable. should /boot also be marked bootable?
2105[22:59:08] <greycat> Nothing cares about that except you.
2106[22:59:29] <greycat> Or maybe the old MS-DOS boot record.