8[00:11:27] <longshot> Is there a better way I can control the before/after dependencies for systemd?
9[00:12:22] <longshot> I need a strict hard requirement for my service to run *serially* before another serivce. I've added some debugging lines to my script and I can see that the other service that MUST run later is already running when my service starts up.
10[00:12:49] <longshot> What are my options for *forcing* systemd to do what I want?
11[00:13:01] <ratrace> to do what exactly?
12[00:13:23] <longshot> I need to 100% guarantee that another service does not begin starting before my service has concluded
13[00:13:24] <ratrace> you frist have to very, very precisely describe the condition and relationship between services
25[00:16:04] <ratrace> so then it should behave exactly as you expect it. systemd considers Type=oneshot services "started" when the main process exits. that satisfies the Before for dependency
26[00:16:20] <longshot> It's not though.
27[00:16:37] <longshot> The Before= services that I listed are already done when my script starts.
38[00:19:55] <longshot> That doesn't seem like the right thing at all.
39[00:19:56] <ratrace> no, why? you can have local overrides in /etc
40[00:20:17] <ratrace> but I'm really convinced that this works as expected and you're maybe missing something or not looking at it correctly
41[00:20:43] <jmcnaught> Maybe telling us what the services are or sharing the units in a pastebin would help clarify then.
42[00:20:47] <ratrace> I know because I have such services. Type=oneshot, Before=somethingelse.service and somethingelse.service doesn't start until this oneshot exits
43[00:21:14] <longshot> My service is already in /etc/systemd/<service I want to run later>.requires/
45[00:21:46] <longshot> ratrace: That's why I'm asking. I'm trying to figure out wtf I'm doing wrong.
46[00:22:23] <longshot> I'm tring to preempt zpool import
47[00:23:02] <ratrace> longshot: do you have RemainAfterExit=yes in your "before" service?
48[00:23:07] <longshot> I need to do somethings before the pool is imported. The first line of my script is `zpool list ; zfs list`, and the pool and all datasets are already imported.
49[00:23:13] <ratrace> longshot: preempting zpool import, to do what?
50[00:23:24] <longshot> I have RemainAfterExit=false
51[00:24:03] <longshot> I'm building an appliance type of system. So if there's a bootparameter passed in I want to destroy the pool
52[00:24:18] <longshot> this is the factory reset mechanism
53[00:24:19] <ratrace> sounds like maybe a task for an initramfs script
54[00:24:52] <ratrace> because really... are you sure the pool is imported from the systemd service and not from initramfs?
55[00:24:53] <longshot> So maybe this should be in the live-bottom?
56[00:25:49] <longshot> ratrace: It might be imported from the initramfs.
57[00:25:52] <ratrace> you mean init-bottom? I don't know what you're doing there exctly
66[00:28:13] <ratrace> but for initramfs-tools, you have "scripts" which execute in stages on boot, for regular debian installation.
67[00:30:50] <longshot> OK.
68[00:30:56] <longshot> Yeah, I think this is what I need to do.
69[00:31:20] <ratrace> the initramfs-tools(7) manpage explains the "scripts" stages
70[00:31:32] <ratrace> then again I don't know what your "factory reset" really does
71[00:31:45] <longshot> It may not seem like it, but I think this definitely helped. Digging through some of the other stuff I've got, I think I know where I can insert these things.
72[00:32:06] <longshot> So it's a live image and data is stored on the pool.
73[00:32:36] <longshot> If the pool doesn't exist it will automatically go into setup that prompts for config params, creates the pool, writes the config and reboots.
74[00:33:30] <longshot> and that stuff is all working, but I can't destroy the pool while I'm using it, so the factory reset is triggered from a boot menu option
85[00:38:06] <longshot> At least now I know I'm not crazy with the systemd stuff being out of order. It probably is in the order I expect, but the initfs is doing the import, not zfs-import.target so I was already way too late anyway.
87[00:39:22] <ratrace> longshot: you can put an ExecStart=sleep 10 in the unit, above the regular ExecStart. and then look at the timestamps when each service has run. yours and zfs-import.service's
129[01:28:29] <neilthereildeil> from the debian wesite
130[01:28:31] <neilthereildeil> amd64
131[01:28:58] <neilthereildeil> the boot messages said starting SDDM was OK
132[01:29:10] <neilthereildeil> its hanging at starting hostname service
133[01:30:17] <neilthereildeil> however startx works
134[01:30:26] <neilthereildeil> why isnt startx getting called automatically?
135[01:31:58] <abrotman> have you considered it's a video driver problem?
136[01:32:05] <neilthereildeil> but startx works
137[01:32:19] <neilthereildeil> abrotman: and ur right, theres no driver installed
138[01:32:22] <neilthereildeil> so its using VESA
139[01:32:24] <abrotman> then look at the sddm logs?
140[01:32:27] <neilthereildeil> which is software rendering
141[01:32:29] <neilthereildeil> which is OK for noiw
142[01:32:44] <neilthereildeil> but it should at least show X automatically on boot
143[01:32:46] <abrotman> you can type a whole sentence on a single line, we swear it won't hurt ..
144[01:32:49] <neilthereildeil> it shouldnt be booting to the cmdline
145[01:33:52] <abrotman> Which graphics card do you have? Does it require non-free drivers? Did you install those?
146[01:34:00] <neilthereildeil> its ATI
147[01:34:36] <neilthereildeil> and yes, it requires nonfree drivers. but again, even VESA software drivers should work for now to simply get X to automatically display with SDDM.
148[01:35:36] <neilthereildeil> abrotman: my goal is to just get X with VESA automatically starting
149[01:37:22] *** Quits: securethemews (~securethe@replaced-ip) (Remote host closed the connection)
150[01:37:28] *** Quits: ax56234 (~NickServ@replaced-ip) (Read error: No route to host)
151[01:37:35] <abrotman> tried installing XDM to see if it behaves differently?
175[01:46:24] <dpkg> Debian-Installer is able to load additional <firmware>, by including it within installation media or supplying on removable media (e.g. USB stick, floppy). See replaced-url
176[01:46:28] <abrotman> no .. not that
177[01:46:30] <neilthereildeil> ill just say it: debian is broken out of the box
178[01:46:31] <abrotman> dpkg: firmware images
179[01:46:31] <dpkg> There are <live> system and <netinst> and DVD images containing non-free Debian <firmware> packages available from replaced-url
223[02:33:38] <ASDX> we've had a secondary interface not come up on restart (have to manually 'ifup' it). it's in /etc/network/interfaces and in the same subnet as the primary one. does anything special need to be done to ensure it comes up on restart in debian 10?
331[04:53:48] <PMT> Hi all, I'm having an inconvenience where I'm trying to see if an up to date system from X years in the past would have a bug I'm encountering for bisect purposes, but trying to run "apt update" in the chroot with its repository pointed at snapshot.debian.org/archive/debian-ports/... is erroring out because "The following signatures were invalid: EXPKEYSIG 06AED62430CB581C Debian Ports Archive Automatic
332[04:53:54] <PMT> Signing Key (2018)" - is there a way to downgrade this error to a warning, since this is purely a testbed, or am I going to be stuck hand-installing packages?
408[06:36:26] <mtcdood> I'm trying to rewrite incoming HTTP requests for / to /index.php?page=1 but the following isn't having any effect: replaced-url
409[06:36:33] <mtcdood> Does anyone have advice on what to do?
410[06:38:57] <PMT> It has been awhile, but I think you want a condition like RewriteCond %{REQUEST_URI} ^/$ then RewriteRule ^(.*)$ [...]
466[07:59:05] <wlxmhls> hello! Does anyone use openscenegraph? I've installed libopenscenegraph. But it does not contain osgProducer. What else should I install for osgProducer?
507[08:48:50] <PMT> In this case, it's a sid box, I just need sid from 2018 for investigating a kernel bug, since installing ancient kernel packages on modern systems results in a mess (if you want to compile modules against it), and actually recompiling on this system is painful.
513[08:52:18] <PMT> nkuttler: yes, the whole reason for this discussion was using snapshot, but the GPG signatures on the repo I wanted to use were with expired keys.
514[08:52:52] <jelly> a fancy sparc not 20 years old!
556[09:51:15] <g0zzy> I set up a headless minimal buster using the wireless interface. That is not connecting and so i thought i'd use Ethernet. The docs say ifup should do it but ifup enp2s0 simply results in an unknown interface message. How do i do it manually?
557[09:53:07] *** Quits: tejr (~tejr@replaced-ip) (Remote host closed the connection)
740[13:28:26] <solrize> if you're trying to implement a somewhat complicated processing chain, is there wisdom on whether it's preferable to use python scripts vs bash scripts? right now i'm using bash but python seems more precise, if that means anything
748[13:40:32] <otisolsen70> I have just installed kvm and am trying to create a kvm VM and I get this error: ERROR Host does not support domain type kvm for virtualization type 'hvm' arch 'x86_64'
749[13:40:33] <otisolsen70> - what am I missing for this to work?
750[13:41:38] <otisolsen70> Do I need to reboot after installing kvm for it to work?
752[13:45:03] <otisolsen70> Hmm.... Reboot did not help
753[13:47:59] <Walex> solrize: Shell and Python are not similar languages: Shells have scripting languages, Python is a programming language.
754[13:48:20] <solrize> yeah i know that, i wonder if complicated bash scripts shouldn't really exist
755[13:48:44] <Walex> solrize: the main entities in shell language are processes, pipes, process statuses, files; Python's are lists, strings, dictionaries.
756[13:48:45] <solrize> i've written a lot of python and not that much bash
757[13:49:06] <solrize> well it's not too hard to set up pipes and stuff in python
758[13:49:07] <petn-randall> solrize: bash is fairly limited if you want more than simple if/then constructs. Really depends on what your goal is. What is your "somewhat complicated processing chain"?
759[13:49:21] <Walex> solrize: so if your task is mostly process scripting, use shell; if your task is mostly data structures, use Python
760[13:49:48] <solrize> well i'm using bash and i keep running into stuff that seem real clunky and would be easier in python
761[13:50:00] <Walex> solrize: put another way, if data processing in your case is done by chains of processes, and not by the shell, use shell.
763[13:50:21] <solrize> like i can say "if -f filename" to check if a file exists, but i want to handle it differently if its size is > 1MB and there's not a real simple way to do that afaict
764[13:50:26] <petn-randall> bash, or rather POSIX shell, has the advantage that it'll run pretty much on any OS in the last 20 years.
765[13:50:37] <solrize> there is that, true petn-randall
766[13:50:54] <Walex> solrize: you are then using it wrong as your example shows.
767[13:51:02] <petn-randall> solrize: Can you tell us about your specific case you're working on?
768[13:51:12] <solrize> i'm trying to convert a huge pile of epub files to text files
769[13:51:24] <Walex> solrize: to check the size of a file you can use a command that does that.
770[13:51:46] <Walex> solrize: where the epub conversion is done by a tool?
771[13:51:47] <petn-randall> solrize: If it's your personal use, go with whatever is more comfortable for you, in that case Python.
772[13:52:19] <petn-randall> I'd only pick sh if I'd have to support Mac OS X, Redhat 6, and Debian 10 with it.
774[13:52:43] <solrize> Walex, is there a convenient command to get the size of the file? i mean other than parsing ls -l output through awk or sth awful like that. yea, epub conversion is by pandoc
775[13:53:20] <solrize> well i thought i would expand my horizons a little by using bash but now i'm wondering if it's worth it
776[13:54:15] <Walex> solrize: 'stat'
777[13:55:31] <Walex> stat -c '%s" "..."
778[13:56:10] <Walex> solrize: but that is probably not the way to do it in shell language.
779[13:57:03] <solrize> walex ah interesting thanks i didn't know about that
780[13:57:17] <Walex> solrize: probably the right way is not to check the size of each file individually, but to use 'find' to generate lists of files below or above a size
781[13:57:48] <solrize> yeah i thought of that but that's a little messy too
782[13:58:14] <solrize> anyway thanks i have to sign off for now, might ask some more about this later
787[14:03:51] <Walex> another way to check the size of a single file is: case $(find "$FNAME" -size -10M) in ?*) ...;; '') ...;; esac
788[14:03:54] <otisolsen70> When I use virt-isntall to install a vm with kvm it fails if I am a regular user and works if I am root. What do I need to make it work for a regular user? Should I add the user to group 'kvm'?
795[14:06:04] <Walex> otisolsen70: if you don't want to you can run virtual machines fully in user mode by using QEmu without the KVM accelerator (but things will be slow)
796[14:06:07] <otisolsen70> Walex, it is just that replaced-url
797[14:06:23] <otisolsen70> Walex, I want kvm accel
855[15:16:14] <bodqhrohro> How is autoclean supposed to work? I have 3 versions of teamviewer in /var/cache/apt/archives/: teamviewer_14.4.2669_amd64.deb teamviewer_15.10.5_amd64.deb teamviewer_15.17.6_amd64.deb, shouldn't autoremove delete at least two of them? All seem to share the same name "teamviewer", not "teamviewer14"/"teamviewer15" or so.
856[15:16:22] <wknapik> hi
857[15:17:26] <wknapik> just registered at salsa.debian.org, but can't log in - "Your account is pending approval from your GitLab administrator and hence blocked". anyone here that could approve?
858[15:17:50] <bodqhrohro> *shouldn't autoclean delete at least two of them
859[15:19:44] <tarzeau_> bodqhrohro: what does dpkg -l |grep teamviewer say?
860[15:20:12] <tarzeau_> just teamviewer and then the version separate it appears, just checking to be sure though
861[15:20:22] <bodqhrohro> tarzeau_: ii teamviewer 15.17.6 amd64 Remote control and meeting solution.
934[16:28:37] <sigint> elios, on desktop I have notifications "Software Update Available" coming from time to time but I'm not sure how this got setup, it was there by default
958[16:55:00] <deadrom> Hi. raid5, 3 disks, /dev/md0p1 with xfs. added one disk. ran a full reshape. ok. partition is still size of 3 disks. gparted updated GPT to reflect new size but I can't get it to expand the p1 partition. help?
960[16:55:33] <factor> Does anyone know of a joystick mapping program that will map an AXIS value to a button. Say negative value to Left and positive value to right?
971[17:01:14] <Walex> deadrom: perhaps it is just my limitations, but are your sure you have a good conceptual model of block devices?
972[17:01:29] <deadrom> Walex: yes.
973[17:01:56] <Walex> deadrom: then it is must be my own limited understanding :-)
974[17:03:01] <deadrom> Walex, well, it does ntohing more than reshape the pattern of the checksum stripe distribution, so that's not the issue. I think it's more about the xfs filesyystem on it and the tooling around xfs that are troublesome
975[17:04:03] <Walex> deadrom: remember you said that you are sure you understand block devices...
976[17:04:39] <deadrom> Walex: yes. if what I say makes you think otherwise stop the innuendos and please tell me.
980[17:07:11] <factor> Walex, not really I am trying to play DIRT4 which only accepts buttons for left or right , while the wheel I have is analog. I just want to convert a positive or negative number to left or right.
982[17:08:46] <factor> Walex, Would be cool to write my own program to do this for the joystick and send a specific positive value or negative value as left button or right button.
983[17:09:25] <Walex> factor: sorry no idea whether that is possible at the input subsystem level, I'd buy a gamepad, as to writing a little daemon there are some potential examples
984[17:09:37] <factor> Walex, Can I pragmatically enable a joystick button vi /sys/input or /dev/input/js0
992[17:11:40] <cybrNaut> would it make more sense to install bullseye now?
993[17:11:42] <deadrom> cybrNaut: install 10 and dist-upgrade to 11 once it's out - why not?
994[17:11:49] <Walex> deadrom: Those values might say something to you, I have no idea whether those are the expected values.
995[17:13:17] <deadrom> Walex: it's 4x12TB in raid5 so the the 70.... number refelcts a usable capacity of 35TB from 4x12TB/R5 quite well, and the 46.... is pretty much the old capacity of 22-23TB from before
996[17:13:18] <cybrNaut> deadrom: i've been burnt by dist upgrade. I prefer to do fresh installs. My situation is quite complex.. i have a few abandoned fragile pkgs that I keep on life support
997[17:13:59] <Walex> deadrom: there is your answer.
1007[17:15:47] <Walex> deadrom: you cannot "grow" partitions, you can delete the partition table entry, recreate it larger being careful to have it start at the same offset, and then resync the partition tables.
1008[17:16:01] <deadrom> if you know another tool to grow it I'd appreciate a hint.
1010[17:16:46] <Walex> deadrom: that applies to partitions on all block devices, whether physical disks or virtual disks like 'md0'.
1011[17:16:48] <deadrom> well, yes, that is what happens under the bonnet, I'd still like to have a tool do it than manually. where is why I look at gparted, but it won't offer the option.
1015[17:19:52] <Walex> deadrom: as I wrote, one cannot "grow" partitions, because "partitions" don't exit, they are not "entities", it is partition table entries that must be redefined in place
1017[17:20:32] <alkisg> gparted has an option to "grow partitions" (which of course means to rewrite the partition table), but I've no idea if it can handle md
1018[17:20:37] <Walex> deadrom: after that is done you can resize the filesystem to the new larger size of the block device.
1019[17:20:44] <alkisg> It works fine for ext4, ntfs etc though
1020[17:21:01] *** Filohuhum_ is now known as Filohuhum
1026[17:26:46] <Walex> alkisg: that depends on the version of 'gparted'; the one I got here has "Resize/Move" which is basically the same as 'gdisk
1027[17:26:53] <Walex> alkisg: that depends on the version of 'gparted'; the one I got here has "Resize/Move" which is basically the same as 'gdisk' or 'cfdisk'
1041[17:30:05] <heydrickx> My router is a mini-pc with debian. It has two lan interfaces and wifi
1042[17:30:06] <deadrom> heydrickx: and just describe the issue right away, no need to ask if it's ok to ask
1043[17:30:19] <heydrickx> deadrom, Sorry
1044[17:30:58] <heydrickx> So I've installed everything from scratch, starting with debian, then configured everything I needed and installed hostapd for wifi
1045[17:31:57] <deadrom> and?
1046[17:31:58] <heydrickx> Everything works, but the wifi just stops from time to time. When it happens, the lan network keeps working and can access the internet. But the wifi is unavailable and the router is not responding to SSH. It still continue to route the lan network
1052[17:33:01] <heydrickx> My only solution is a hard reboot of the router. I don't really know where to look. After some searches I thought it was due to transmission-daemon running on an other machine. But even when it's stopped, it keeps happening
1053[17:33:07] <Walex> heydrickx: unloading and reloading the driver usually fixes it for me.
1054[17:34:11] <Walex> heydrickx: sometimes there is some indication in 'dmesg', but sometimes it just "stops working".
1055[17:34:30] <heydrickx> Walex, and the router wasn't responding to SSH neither when it happened to you ?
1056[17:34:51] <deadrom> heydrickx: is adding another rj45 interface an option? in that case I'd actually recommend to get an AP and let the firmware do it's proprietery firmware shenanigans in it's Shenzen box and let you do something better with your precious lifetime. that's how I tackle wifi issues and it works a charm.
1057[17:35:08] <Walex> heydrickx: SSH over WiFi or Ethernet?
1058[17:35:16] <heydrickx> Walex, Ethernet
1059[17:35:42] <heydrickx> deadrom, sorry but what is an AP ?
1060[17:35:45] *** Quits: ska (~ska@replaced-ip) (Read error: Connection reset by peer)
1061[17:35:54] <Walex> heydrickx: if SSH over Ethernet stops working, there is some kernel bug (probably within the WiFi driver) that causes the kernel to get "stuck".
1065[17:36:58] <Walex> heydrickx: Either get a dedicatd AP as "deadrom" suggests, or change the WiFi chipset, or change the setup parameters until the bug is no longer triggered.
1066[17:36:58] <heydrickx> Walex, so re-installing the driver may work ?
1067[17:37:30] <heydrickx> Walex, What is an AP ?
1068[17:37:32] <Walex> heydrickx: it could be also a problem with the filesystem/storage driver, or overheating, etc.
1073[17:38:37] <Walex> heydrickx: I have found that many USB WiFi sticks have a fairly well debugged "generic" driver. Most Intel WiFi chips seem fairly reliable.
1074[17:38:55] <deadrom> heydrickx: which wifi type, 11ac, 11ax..?
1076[17:39:34] <heydrickx> deadrom, PC is this one : replaced-url
1077[17:39:44] <deadrom> heydrickx: what does lsusb oder lspci tell you about the hardware ID and hat does google tell you from the hardware ID what chipset maker and model it is?
1078[17:39:46] <Walex> heydrickx: in my current laptop, which I use as an AP too with 'hostapd', the Realtek RTL8822BE seems fairly reliable.
1079[17:40:40] <heydrickx> deadrom, in hostapd, I specify hw_mode=g so that's what I actually use
1081[17:40:48] <Walex> heydrickx: welcome to the realization that drivers, firmwares, etc. are not bug free, and system integration is a difficult task.
1082[17:41:17] <deadrom> "WI-FI: IEEE 802. 11b/g/n/ac 2. 4G+5G" that can be anythign and nothing. look in lspci or lsusb and see what it says there
1083[17:41:47] <heydrickx> deadrom, I'm using 802. 11b/g and 2.4G only
1084[17:42:03] <deadrom> it's fairly possible that debian loads the best fitting module to drive it but the chip still new.
1085[17:42:19] <heydrickx> Walex, I get : Network controller: Intel Corporation Wireless 7265 (rev 59)
1086[17:42:55] <deadrom> heydrickx: the mode you use is not the issue: it's an 11ac chip and ac chips are not that well docuemtend these days, so no matter what .11 mode you drive, it could still be a fundamental code issu in how the chip talks to the system
1087[17:43:16] <deadrom> Walex: xfs grown to new size, all fine now
1088[17:43:22] <heydrickx> deadrom, so there's nothing I can do except replacing the chipset ?
1093[17:46:56] <deadrom> heydrickx: well, the 7265 is officially supported, the internet says but wifi comes with a grain of lback magic. even if the driver is out there it's possible that the firmware has issues that we don't know about or this NUC based motherboard has some whacky chipset. I saw soem strange issues on NUC form factor boards, even those by intel. since it is intel's spanking newest chip I'd say have a look at what /var/log might hold regarding error
1095[17:46:58] <deadrom> messages and feed them to you favorite search engine, see if something comes up, and in a pinch... try a newer kernel series
1096[17:47:04] <deadrom> heydrickx: debian 10?
1097[17:47:45] <Walex> heydrickx: eventually the Intel guys get to debug the driver for newer chipsets, but it is not a huge priority for them. Intel also maintains the Linux GPU drivers for their chipsets, but those are fairly buggy too, even if much improved over the past few years.
1101[17:48:02] <Walex> heydrickx: you could also try backported kernel
1102[17:48:22] <deadrom> Walex: I didn't expect cfdisk to offer the reisze feature, in the past it did not. would have gone there straight away. everyday's a school day...
1103[17:48:47] <heydrickx> Walex, In order to get the driver I installed firmware-iwlwifi. Is it correct or do I miss something ?
1104[17:49:28] <deadrom> heydrickx: debian 10 comes with a 4.19 kernel out of the box and I know there have been *huge* wifi improvements in the 5. kernel series, so yes, backported kernel as Walex suggested
1105[17:49:38] <Walex> heydrickx: that's also needed. But the driver is in one of the kernel packages.
1111[17:51:42] <heydrickx> I'm going to try what you said. But "backported kernel" doesn't mean anyting to me. I'm not familiar with kernel manipulations
1112[17:52:11] <epsilon> heydrickx: do you want to run wifi on that box as client or access point?
1113[17:52:18] <deadrom> heydrickx: the distinction between formware and driver is somewhat tricky. firmware is a dreaded blackbox, always has been, reason we don't like it. you could have the latest driver and all and still firmware is flaky, but still, the kernel module plays a large part, and with wifi issues my first go-to is "latest kernel", in this case a 5 series backpported from debian/testing
1114[17:52:24] <heydrickx> epsilon, as an access point
1119[17:54:01] <deadrom> the bot here will tell you, too, I forgot how to invoke it
1120[17:54:21] <heydrickx> I understand I will get more up-to-date kernel but I may lose in stability... right ?
1121[17:54:38] <avu> !bpo kernel
1122[17:54:38] <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))'».
1159[18:05:23] <shtrb> Walex, that's work for me (that what I do ) I just don't see why the change does not do the job for me
1160[18:05:31] <Walex> phogg: editing SONAMEs is a great crime :-)
1161[18:05:54] <patroxos> hey. i will get debian 10 netinstaller. should i pick the free or non-free (with firmware) iso? my laptop needs firmware for a couple of hardware so i am going to install firmware anyway in the end
1162[18:05:56] <shtrb> Walex, I'm little bit hazed so there might be some incorrect thing I do there
1163[18:06:05] <patroxos> does the non-free install every firmware by default?
1164[18:06:05] <phogg> Walex: I just meant by way of a symlink. Pretend it's the same, see what happens (probably a link error)
1165[18:06:10] <patroxos> or only ones i use
1166[18:06:20] <shtrb> patroxos, it depends if you wish that hardware to work or not :p
1167[18:06:28] <Walex> phogg: the SONAMEs have to match :-)
1168[18:06:38] <patroxos> i will install firmware after installation (if i get the free iso)
1169[18:06:39] <phogg> Walex: you make a compelling argument
1170[18:06:57] <patroxos> but isn't that the same as getting non-free? or non-free installs every other firmware i will also not use
1171[18:07:14] <Walex> bah! That SONAME business is very hard to get around
1172[18:07:41] <Walex> shtrb: not all route changes are allowed, some are ambiguous and are disallowed.
1173[18:07:46] <phogg> time to build your own, old openssl! And then dump it in a place the rest of the system can't find it
1174[18:08:05] <patroxos> does anyone know if the non-free iso installs every available firmware? (which i don't desire)
1175[18:08:12] <Walex> shtrb: ambiguous or whatever the 'iproute2' authors decided was not going to be supported
1176[18:08:19] <shtrb> lol
1177[18:08:20] <shtrb> thanks
1178[18:08:25] <patroxos> and in that case i should go with the free iso and manually install the firmware i need only?
1179[18:08:28] <shtrb> patroxos, it does not infact install everything
1180[18:08:46] <petn-randall> patroxos: If you install with the libre iso, I'd use that.
1181[18:08:50] <petn-randall> *can
1182[18:09:03] <patroxos> petn-randall, i then have to install firmware manually
1183[18:09:10] <petn-randall> patroxos: But most wifi needs non-free firmware, and if you need to use the wifi to access the internet you'll be stuck.
1184[18:09:18] <patroxos> if non-free autodetects firmware missing, and installs ONLY that it would be ideal
1185[18:09:32] <patroxos> i can install with the libre one
1186[18:09:32] <petn-randall> patroxos: Then just use the non-free installer and install/remove only the firmware you want.
1187[18:09:39] <Walex> BTW as to proprietary, I am very impressed with how many games are supported by Steam on Linux, and IIRC Debian Developers get it for free.
1188[18:10:11] <Walex> or at least all the Valve games for free.
1206[18:15:32] <petn-randall> patroxos: Maybe if you explain what you fear from the non-free installer we might be able to guide you better.
1207[18:15:53] <patroxos> i fear that the non-free installer will install firmware for a myriad devices i dont actually need
1208[18:16:25] <patroxos> while the libre installs nothing and i can pick manually after installation the 3 packages i need. linux-firmware, amd64-microcode, and another firmware for wifi
1209[18:16:26] <petn-randall> patroxos: but you can remove all those quite easily, even if it were the case.
1210[18:16:47] <patroxos> yeah i would rather having just enough
1211[18:16:56] <petn-randall> AFAIR there's some heuristics that it will only install some of those.
1213[18:18:00] <shtrb> patroxos, at most you can put some extra deb files on a usb get the libre one , and plug and manually install missing packages during process
1214[18:18:24] <patroxos> i read this. he says instead of installing debian non-free and getting every firmware, just get the libre iso, and install the missing firmware from usb stick
1224[18:20:43] <patroxos> yeah never done that before
1225[18:20:49] <patroxos> usually get non-free iso for laptops
1226[18:21:13] <patroxos> (which i have to add isnt so easy to find on the debian ftp server)
1227[18:21:13] <sney> patroxos: the non-free iso will install extra firmware files, yes. they sit in /lib/firmware and if you don't have the associated hardware, the drivers will not load, and the files will simply be on your disk doing nothing
1228[18:22:05] <patroxos> sney, i see
1229[18:22:18] <patroxos> so i might use them in the future with some external device maybe
1230[18:22:31] <patroxos> so its plug and play
1231[18:22:50] <sney> yes, that's a side benefit as well, if you ever have to use some wifi dongle etc
1232[18:23:00] <patroxos> yes thats what i thought
1247[18:30:20] <hid3> Greetings everyone. Before I start a certain service, I need to prepare configuration files and start multiple instances. For example, I need to prepare X configs, where X is the number of /dev/ttyACMX devices. Is there any guidance achieving sth similar using systemd?
1262[18:45:55] <imMute> hid3: templated service units will be useful. and you can write udev rules that match each device node and cause systemd to start a unit for each.
1263[18:46:33] <imMute> hid3: SYSTEMD_WANTS in the udev rule
1303[19:23:54] <deadrom> set smbpasswd -a -U joe on deb10. other client machine: smbclient -L server asks for password then says tree connect failed: NT_STATUS_ACCESS_DENIED
1305[19:24:26] <deadrom> what am I missing? no domain controller, just want a simple share with simple passowrd protection that I can access from other windows and linux machines
1306[19:25:14] <epsilon> does the user exist on the host, not just in smbpasswd?
1307[19:25:27] <deadrom> that user does exist
1308[19:26:30] <epsilon> you have cifs-utils installed? mount.cifs //serverpath /localpath -o user=username
1309[19:27:21] <epsilon> and there must be a share fmatching or that user in etc/samba/smb.conf
1331[19:40:23] <deadrom> in 99 I asked Epic Games if they will be porting more games to Linux since they did with Unreal Tournament, and they even replied: no, easier to port the entire game than just extract the server part for multiplayer hosting
1333[19:41:46] <[Kraken> so if I'm seeing kernel panic in normal boot And recovery mode (4.19.0-16-amd64) ; though -14 boots fine ; what would be suggested
1334[19:42:03] *** Quits: kristijonas (~kristijon@replaced-ip) (Remote host closed the connection)
1337[19:44:52] <deadrom> now this is strange deluxe: I tried the old passwd from a long running samba server on my network that will decomission once all transfers are complete, and that worked. I nowhere told samba to ask other servers for anything. do they sync once they see each other..?
1343[19:51:59] <[Kraken> sney: the panic message, says 'try passing init= option to kernel'
1344[19:52:04] *** Quits: kristijonas (~kristijon@replaced-ip) (Remote host closed the connection)
1345[19:52:28] <[Kraken> maybe If I wait long enough the next kernel will work, and I can just remove -16 for now ?
1346[19:52:46] <sney> [Kraken: init= takes /bin/bash as an option, but the initramfs debug will tell you more.
1347[19:52:50] <sney> [Kraken: sure, that's an option.
1348[19:53:45] <sappheiros> Firefox is resuming, perhaps from sleep or restoring after shutdown, with window offset in the wrong position unable to click-drag to change ...
1350[19:54:23] <sney> [Kraken: you can 'zless /usr/share/doc/linux-image-4.19.0-16-amd64/changelog.gz' to see if any of the changes/fixes in -16 are relevant to your system. they may not be.
1351[19:54:38] <[Kraken> when and where would I enter ; rootfstype=<rootfstype>
1354[19:55:24] <sney> [Kraken: you need to read the description and instructions, not just the code blocks. that says "to fix wrong fs recognition" and is probably not what you need. start at the part that says "rescue shell"
1355[19:56:33] <sney> [Kraken: you can see the same changelog at replaced-url
1356[19:57:04] *** Quits: kristijonas (~kristijon@replaced-ip) (Remote host closed the connection)
1380[20:20:00] *** Quits: sauvin (sauvin@replaced-ip) (Remote host closed the connection)
1381[20:22:15] <[Kraken> all a bit over my head sney, can't even figure out how to get a rescue shell
1382[20:23:03] *** Quits: kristijonas (~kristijon@replaced-ip) (Remote host closed the connection)
1383[20:23:08] <sney> [Kraken: then refer to the part about "it is probably safe to just stay with -14" - the wiki page was merely an option for if you want to investigate the cause of the panic.
1390[20:27:46] <happy_potato> Hi everybody! One question: do you guys know if I can find a specific channel for the setup of Secure Boot within debian stable?
1391[20:27:53] *** Quits: emOne (~emOne@replaced-ip) (Remote host closed the connection)
1392[20:27:58] <Delf> Tried to install nvtop, then there were dependency problems. Now dpkg gets stuck on "Setting up nvidia-installer-cleanup (20151021+9)"
1393[20:28:32] <sney> Delf: pastebin apt's whole output and link here
1396[20:29:10] <sney> happy_potato: most documentation is here, replaced-url
1397[20:29:39] <rr123> not really debian specific, for priorities, i have realtime 1~99 mapped to -2~-100 under /proc/pid/stat, then I have 0~39 for non-real-time processes, what about -1?
1399[20:29:42] <sney> happy_potato: some EFIs, particularly early revisions, are buggy with non-windows OSes though, so make sure yours is up to date before trying debian.
1412[20:32:01] <happy_potato> @sney, right, the problem is that after enrolling the certificate into MOK, and so, the system boots (SB validation is still disabled) and I get messages like "PKCS#7 signature not signed with a trusted key"
1413[20:32:05] *** Quits: tocka (uid198544@replaced-ip) (Quit: Connection closed for inactivity)
1414[20:32:08] *** Quits: kristijonas (~kristijon@replaced-ip) (Remote host closed the connection)
1416[20:33:03] <sney> Delf: this only appears if you have used the nvidia installer, rather than the debian packages. nvidia-installer-cleanup is a specialized package to undo the damage it does to a debian system. if you enter N here, what happens?
1417[20:33:07] <happy_potato> This is coming from the modules (zfs-related) that I have built via DKMS. Seems I need to register my private key, the one used to sign those modules, in the kernel?
1422[20:35:30] <Delf> So now I cannot install/remove anything because it cannot complete this
1423[20:36:29] <sney> happy_potato: aiui the 4 steps on that wiki page are comprehensive, you generate a key, enrol it, sign the kernel, and then sign the modules. if you did all 4 steps and there are still problems, maybe just do them again to make sure and watch for specific error output during that process
1427[20:39:05] <sney> Delf: ctrl-c out of that and try running 'nvidia-installer --uninstall' directly, the script is supposed to run it non-interactively with '--no-questions --no-runlevel-check --no-x-check --ui=none' but maybe nvidia changed the command line options
1448[21:01:17] <happy_potato> @sney, I have just repeated (a number of times) the process, and I get the same situation. Something weird is that I see my certificate if I do mokutil --list-enrolled... but if I run "dmesg | grep cert", I do not see any mention to it (and in the wiki from debian says I should see something)
1482[21:36:11] <happy_potato> @sney, also: if I use sbverify --list, I see that the kernel has been successfully signed.... But I still get those PKCS warnings about signature not signed with a trusted key
1492[21:55:19] <happy_potato> @sney, got it: I was getting that error because I had still not enabled secure boot in the EFI. Once enabled, there are no warnings and the system successfully boots. Thank you!
1594[23:57:03] <ndroftheline> what's the most minimal gui i can install on a server
1595[23:57:27] <ndroftheline> sorry
1596[23:57:30] <ndroftheline> desktop environment
1597[23:59:29] <ratrace> well, if you just want a clicky-clicky gui, then a WM like openbox maybe. if you want a bit more elaborate environment, xfce or mate