rrq | and that file should be a table of static assignments, in format "mac,hostname,ip" | 00:00 |
---|---|---|
grillon | yes I got it | 00:01 |
grillon | and it's exactly what I want | 00:01 |
grillon | I mean it's filled as it should be good mac, ip, hostname | 00:02 |
grillon | in this order | 00:02 |
rrq | hmm that's different order from mine (which works) | 00:03 |
rrq | perhaps try reconfigure libvirt and use name= as ip and ip= as name, then restart | 00:04 |
rrq | (if that works, then someone worried about semantics should lodge a bug to debian/libvirt) | 00:05 |
rrq | but remember also that the clients try to persist their last assignment by proposing that in their request, and dnsmasq typically accepts that | 00:06 |
grillon | I think it's your last statement | 00:08 |
grillon | because as I said if I wait few minutes and do dhclient eth0 on the guest I get the good IP | 00:09 |
grillon | and sometimes I have two Ip when I boot (last given IP and IP I want) | 00:09 |
rrq | the client uses leases file... client:/var/lib/dhcp/dhclient.eth0.leases | 00:10 |
grillon | yes so I erased it but it's incrementing, instead of using IP it should | 00:11 |
grillon | it's like it's about time... | 00:12 |
rrq | have you tried with "corrected" file for dnsmasq? and the leases file deleted | 00:15 |
rrq | it might also be something with the macaddress; that it starts up with a random one and then the client sets it? | 00:16 |
rrq | (would be rather odd) | 00:16 |
rrq | do "pgrep -a qemu" to ensure that the macaddress is assigned as command line argument | 00:17 |
rrq | (that's on the host obviously) | 00:17 |
rrq | or even: pgrep -a qemu | sed "s/ -/\n-/g" | 00:18 |
rrq | or even even: pgrep -a qemu | sed "s/ -/\n-/g" | grep macaddr | 00:19 |
grillon | rrq here the pgrep qemu : https://controlc.com/6beb3ef8 | 00:32 |
rrq | ok it has a mac= assignment matching up with your intended devuan_front client | 00:33 |
grillon | before doing it I erased leases files and restart every thing(virtual bridge, libvirt including dnsmasq), started tcpdump and finally the vm | 00:35 |
rrq | so.. and you had changed the detail order for dnsmasq's conf .. you tried my order and it still didn't work? | 00:36 |
grillon | here my tcpdump : https://controlc.com/f81b35c8 | 00:36 |
grillon | here ip addr on guest: https://controlc.com/9e3ad991 | 00:37 |
grillon | as you can see I have dual IPs the good and another | 00:38 |
rrq | hmm first Reply is for 192.168.122.151 | 00:38 |
grillon | I did not changed dnsmasq conf yet | 00:38 |
grillon | yes and I have dual Ips. Sometimes it's dual ip, sometimes only bad one bad ip. to have only the good one I need to wait few minutes and do dhclient eth0 on guest | 00:39 |
rrq | hmm how many network management sub systems does the client have? | 00:40 |
grillon | do you still want I switch hostname and IP in dnsmasq conf? | 00:40 |
rrq | seems not necessary given that first reply is good | 00:40 |
grillon | it's a fresh devuan installation | 00:40 |
rrq | probably dnsmasq is flexible :) | 00:40 |
rrq | the client is "fresh devuan"... with which networking? | 00:41 |
grillon | yes client is a fresh devuan, what do you mean by which networking? | 00:43 |
rrq | let's see: is there an /etc/network/interfaces with a "allow-hotplug eth0" in? | 00:44 |
grillon | yes there is | 00:45 |
rrq | that indicates the client wants udev to set up the network; and this typically competes with eg wicd for configuring eth0 | 00:46 |
rrq | for your use case I would purge wicd | 00:47 |
grillon | ok how? | 00:47 |
rrq | so: apt-get purge wicd | 00:47 |
grillon | not installed so not removed | 00:48 |
rrq | ok; hmm network-manager? same action | 00:48 |
grillon | same result | 00:50 |
rrq | hmm now I see that your desired hostname "devuan_front" gets digested into the hostname "front" for the secondary dhcp request | 00:50 |
rrq | probably the client code takes offense to an underscore in the hostname | 00:50 |
grillon | no I think it's because I change the hostname | 00:50 |
grillon | it was front | 00:50 |
rrq | ok; and that causes a new dhcp request for the new hostname | 00:51 |
grillon | may be... | 00:51 |
rrq | which dnsmasq does not associate with the assignment ... it's a theory | 00:51 |
grillon | it's more starnge than that | 00:51 |
grillon | hostname is still front | 00:52 |
grillon | vm name is devuan_front | 00:52 |
rrq | yes, and dhcp asks for the assignment for "front" which there is none | 00:52 |
rrq | so, add that to the libvirt conf | 00:53 |
grillon | what? front? | 00:53 |
grillon | maybe I should I change my hostname in devuan_front to have devuan_front everywhere? | 00:54 |
rrq | yes, it seems important that the dhcp assignment agrees with the client about the hostname | 00:55 |
rrq | that's new knowledge for me; thank you for that. | 00:56 |
fsmithred | Host names may contain only alphanumeric | 00:57 |
fsmithred | characters, minus signs ("-"), and periods ("."). They must begin with | 00:57 |
fsmithred | an alphabetic character and end with an alphanumeric character. | 00:57 |
rrq | periods? | 00:57 |
fsmithred | that's right from the man page | 00:57 |
grillon | ok my mystake, hostname is front and vm name is devuan_front | 00:58 |
rrq | man page of what? meaning that hostname may be a FQDN I suppose | 00:59 |
fsmithred | man hosts | 00:59 |
fsmithred | yeah, they're talking about fields in the file here | 01:00 |
fsmithred | etc/hosts | 01:00 |
grillon | yes it's front | 01:00 |
fsmithred | that's fine | 01:00 |
grillon | front in /etc/hostname and /etc/hosts | 01:00 |
rrq | ok; anything after the first "period" would be "the network domain" | 01:00 |
grillon | devuan_front for libvirt | 01:00 |
rrq | yes: change it to "front" | 01:01 |
rrq | apparently libvirt conflates the guest name with the hostname of the guest | 01:02 |
grillon | I did and I obtain: https://controlc.com/a089b9a6 | 01:02 |
grillon | yes seems something like that | 01:02 |
grillon | maybe I should rename my vm front | 01:03 |
grillon | for libvirt I mean | 01:03 |
rrq | maybe delete the leases file to start from scratch | 01:03 |
rrq | yes | 01:03 |
grillon | ok I delete then restart every thing(bridge, libvirt) | 01:04 |
rrq | .. .and it's all strawberries and cream ... | 01:04 |
grillon | ok I have restarted everything without changing vm name if it does not work I'll change vm name before restart everything again | 01:06 |
rrq | meaning it says "front" in the static assignment I guess although "devuan_front" somewhere else in that tool | 01:08 |
grillon | dual ip :-( | 01:08 |
grillon | ok let's change the vm name | 01:08 |
rrq | does the static assignment say "front" ? | 01:09 |
grillon | yes | 01:09 |
rrq | perhaps add an assignment for the name "(noname)"; same mac and IP as "front" | 01:10 |
rrq | no: name "(none)" | 01:10 |
grillon | I think you are right | 01:11 |
grillon | still dual ips but this time the first ip is the ip I want | 01:14 |
grillon | tcpdump : https://controlc.com/96ef53f2 | 01:15 |
rrq | hmmm so your client's initrd doesn't have an /etc/hostname I guess | 01:15 |
grillon | ip addr guest + /etc/hosts + /etc/hostname : https://controlc.com/99395e29 | 01:17 |
rrq | yes, I wonder if it'd into initrd upon "update-initramfs -u -k all" | 01:18 |
rrq | it'd = it's captured | 01:19 |
grillon | the hostname was different on the begining, I do not remember what, do you think I need to update initramfs? | 01:19 |
rrq | worth a try ... otherwise you can also delay network configuration to post-pivot, by changing allow-hotplug to auto (+ update-initramfs) | 01:20 |
rrq | btw the dnsmasq format is documented in "man dsnmasq"; look for --dhcp-host= | 01:24 |
grillon | seems to work with auto + delete leases + update-initramfs | 01:25 |
grillon | tcpdump: https://controlc.com/61d477f7 | 01:26 |
rrq | in particular the IP address should be before hostname, which is what you have (not I) | 01:26 |
grillon | I obtain only one IP \o/ | 01:27 |
rrq | great ... maybe raspberries and cream, then :) | 01:27 |
rrq | or something vegan could be good too | 01:27 |
grillon | yes :-) | 01:27 |
grillon | where are you from? | 01:28 |
rrq | I live in Melbourne, AUS | 01:28 |
rrq | Sweden the first 30 years, then here | 01:29 |
grillon | haha on the other side of the world for me | 01:29 |
grillon | Sweden waw the opposite for you too | 01:29 |
rrq | got tired at moving snow | 01:30 |
onefang | I thought you where a New South Welshman rrq. | 01:30 |
rrq | nothng much wrong with them either | 01:31 |
grillon | how's life in melbourne? | 01:33 |
rrq | right now it's splendid! | 01:34 |
rrq | a bit warm of course, but it's summer | 01:35 |
grillon | hehe nice | 01:36 |
rrq | teh odd one-two days of 35-40 C is ok; typically below 25C in the night | 01:36 |
rrq | and off course, no virus; one of the handful countries | 01:38 |
n4dir | i think africa looks pretty good too. And that is more than just a handful of countries, iirc | 01:39 |
grillon | I knew for Africa but not for Autralia | 01:39 |
rrq | yeah.. not a competition of course .. and this topic is moving towards #devuan-offtopic | 01:40 |
grillon | haha you are right, I'm moving to my bed, it's almost two(am) here | 01:41 |
grillon | thank you for your help, have a good day I guess | 01:42 |
n4dir | even better. bed, that is :-) | 01:42 |
* rrq off to edit dnsmasq config | 01:42 | |
grillon | sorry I was curious, all ingredient are needed : same vm name as hostname, delete leases, auto eth0 and maybe update-initramfs | 01:55 |
grillon | good night | 01:55 |
hwpplayer1 | hi people! | 02:53 |
plasma41 | hwpplayer1: o/ | 02:53 |
hwpplayer1 | plasma41: Do you know me I was here maybe 3 years ago and I was joining Online Meetings | 02:55 |
plasma41 | hwpplayer1: I don't recognize your nick, no. | 02:55 |
hwpplayer1 | plasma41: No problem I am here for the good | 02:56 |
chomwitt | hi . | 09:16 |
chomwitt | whille installing from a netinstall iso is there a point when i choose my init system? | 09:17 |
chomwitt | ok. i got two choices | 09:19 |
chomwitt | why not runinit? | 09:20 |
* rrq gratuitous search result for "devuan why not runit" https://github.com/cloux/aws-devuan/issues/1 | 09:52 | |
chomwitt | i am reading it, but having enabled runit how i start/stop sshd? | 11:50 |
Joril | sv up/down ssh ? | 11:52 |
fsmithred | chomwitt, try '/etc/init.d/ssh start|stop|restart' | 13:10 |
fsmithred | only things that have runit init files by default are gettys. | 13:11 |
xinomilo | occasionally stuck windows/DE when using mouse in usb3 port.. could it be mouse hw fault or is it common? | 13:25 |
fsmithred | xinomilo, it's not a common problem | 13:26 |
xinomilo | ok, i see.. it gets unstuck when using trackpad, but after a while the same unresponsiveness.. so my guess is either power options(?) or faulty hw | 13:27 |
debdog | wireless mouse? | 13:28 |
xinomilo | wired | 13:28 |
xinomilo | never had issues when used on usb2 port | 13:34 |
fsmithred | is the plug tight? tail -f /var/log/messages and wiggle the plug. | 13:42 |
xinomilo | plug seems fine, normal disconnect/connect messages | 13:44 |
fsmithred | is it a toshiba? mine has wonky usb3 ports | 13:48 |
chomwitt | sv wont work for sshd | 13:48 |
fsmithred | chomwitt, try '/etc/init.d/ssh start|stop|restart' | 13:49 |
chomwitt | gives error: unable to change to service dir, file does not exist | 13:49 |
chomwitt | fsmithred: one moment | 13:50 |
xinomilo | had a thinkpad before with the same issue, but with usb2 port available never really cared.. now on fujitsu, usb3 ports only. | 13:50 |
xinomilo | will try another mouse to see if it's reproducible | 13:50 |
fsmithred | no init script? Is openssh-server intalled? | 13:50 |
chomwitt | /etc/init.d/ssh start|stop|restart worked | 13:51 |
fsmithred | runit uses the sysvinit scripts | 13:52 |
fsmithred | you can change that, but I don't know the details | 13:53 |
xinomilo | debian runit has only a few scripts in /etc/sv/ . most daemons use sysvinit scripts | 13:55 |
chomwitt | thanks. But i try to understand the runit. So if a service is not runit ''ready'' :-) then the sysvinit scripts will be used. | 13:56 |
fsmithred | runit scripts for gettys are the only ones I saw. I know how to set up console autologin. That's all. | 13:56 |
fsmithred | yes, sysvinit scripts are used with runit and also with openrc. | 13:56 |
fsmithred | there are probable sample runit scripts/files floating around the net | 13:57 |
fsmithred | probably | 13:57 |
chomwitt | so runit is ready but work must be done at the service side? | 13:59 |
fsmithred | yeah, I guess so. | 14:00 |
yanmaani | What's the most stable network configuration? I'm trying to use r8169 with wicd, it used to work half-decently, but now it only works if I set it to 10 Mbit, half-duplex, autoneg=off. | 17:55 |
gnarface | sounds more like a problem with the room or the router | 17:57 |
gnarface | change your router or router settings lately? or change anything about the environment? | 17:58 |
gnarface | signal problems aren't unheard-of, if you just upgraded kernels and can still go back to the old one it's worth checking | 17:58 |
gnarface | but honestly the most stable network configuration is one that uses wires | 17:59 |
gnarface | it's not impossible that another piece of electronics in the room is interfering, but if it is, you need to repair or discard whatever it is, because it's probably carcinogenic and a fire hazard | 18:00 |
gnarface | could be in a neighboring apartment too i suppose | 18:01 |
gnarface | leading causes for that type of malfunction are usually leaky microwaves and faulty fly-back transformers | 18:02 |
gnarface | sometimes ground wire faults in the building wiring | 18:02 |
gnarface | if it's actually a regression in the wifi firmware itself, i would expect complaints to up in a search | 18:05 |
gnarface | like i said, those are less common but also not unheard of | 18:06 |
gnarface | hmm, the autoneg=off thing makes me suspicious of jamming though | 18:12 |
gnarface | heh, interestingly enough, a few days ago, elsewhere on the internet i saw someone asking for advice on how to jam up their neighbor's wifi so their own router would stop channel-hopping. maybe that guy is your neighbor, yanmaani. | 18:14 |
fsmithred | yanmaani, wired or wireless? What chipset? I've got RTL8111/8168/8411 and wired is working ok with network-manager in beowulf. I have RTL8111/8168B on older box and wired kept dropping out on that. I got r8168-dkms package from ubuntu to get it stable. That was on debian wheezy. No idea if that package still exists. | 18:33 |
gnarface | oh, i assumed wireless, i guess i should have verified though | 18:35 |
fsmithred | oh, r8168-dkms does still exist and it's in the repo. | 18:44 |
fsmithred | yanmaani, see 'apt show r8168-dkms' for a list of hardware and some notes about it. | 18:46 |
fsmithred | it's in non-free | 18:47 |
yanmaani | gnarface: No, this is an ethernet cable. | 19:15 |
yanmaani | fsmithred: Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller (rev 15) | 19:15 |
yanmaani | fsmithred: How would I set up r8168-dkms? Just install it? | 19:16 |
gnarface | probably yes | 19:17 |
gnarface | dkms is a thing that triggers module rebuilds on-the-fly, i'd infer by the package name it triggers that automatically on install but if not, check the documentation that comes with it | 19:18 |
gnarface | it's in non-free so it probably contains some binary blackbox blob you need | 19:18 |
gnarface | did you upgrade to gigabit network switches recently? that could also explain a sudden change | 19:19 |
gnarface | some of these devices work fine without the non-free components as long as you don't need it to sync at gigabit speeds | 19:20 |
gnarface | although it's worth noting that if your ethernet cables are hand made or really old, it's possible they themselves can only stably link up at 10 megabit | 19:20 |
gnarface | (just due to the wiring order of the twisted-pairs not being optimal) | 19:21 |
fsmithred | and you need linux-headers | 19:24 |
fsmithred | oh yeah, good call gnarface. I can attest to the fact that dog-chewed cables offer inferior performance. | 19:25 |
gnarface | oh, i forgot, yea for dkms to work you'd need the linux-headers package for the corresponding kernel, though i'd also assume that it would complain about that first if you were missing it | 19:30 |
user___ | https://www.theregister.com/2021/01/25/use_firefox_fedora_chromium_maintainer/ maybe-relevant to devuan too | 19:39 |
user___ | (we may get a better browser in time) | 19:40 |
asdflkj | user___: great time to switch to FireFox or LibreWolf. Personally I ♥ Palemoon | 20:17 |
nemo | wish I could trust palewolf team to fix vulnerabilities in timely fashion | 20:30 |
nemo | but it's already once removed from upstream, and working on backporting stuff | 20:30 |
nemo | also ofc they are ever further away from "standard" web tech ☹ | 20:30 |
nemo | but wow. that's a depressing announcement | 20:31 |
user___ | No, it was to be expected. Apple Google Facebook are all "capitalizing" on past investments. | 20:31 |
user___ | They're not in the vaccin business so they are looking bad on the stock market, by comprison. | 20:32 |
nemo | you'd think google would be more cautious with antitrust investigations in progress | 20:32 |
user___ | Google is officially too big to fail since they removed the "do no evil" motto years ago and also removed the yellow cheese closeup moon view from goole maps. No sense of self-humor, no respect for others other than Saint Dollar and Power. | 20:33 |
user___ | Fartbook wants to index the private WhatsApp messages and attachments people have on their devices and make money off of bare naked children pictures, credit card and medical history data in there. | 20:34 |
user___ | Apple is also gettin greedy with non repairable devices and so on. | 20:34 |
user___ | Sounds like music to me, being on the open source software user side with just a little bit of development thrown in | 20:35 |
nemo | my FOSS contributions dropped off a lot after the kids were born. but maybe in another year or two | 20:37 |
user___ | Today I got a new USB keyboard, slightly less bad than the one it replaces. It has a Power button. And I promptly switched my Beowulf off pressing it. Editing /etc/acpi/events/power-button had me put in the "%e" event detail per acpid manpage, then in the action script, I matched the power button event whole line, instead of just the event head button/power. It's more like button/power PWR 000... 0000... | 20:38 |
user___ | K when coming from the USB keyboard, and exactly the same without the 'K' when coming from the PC's hw power button. So I made sure the 'K' version was ignored and now the power button does nothing. | 20:38 |
nemo | heh | 20:39 |
nemo | the family living room computer has 3 usb and one wireless keyboard attached, but no bluetooth yet. | 20:39 |
nemo | might be a bit more secure than the wired one | 20:40 |
nemo | er. wireless | 20:40 |
user___ | My bt kbd is a source of cussing reasons. No chording no key rollover. Type "legato" and it "eats" characters. | 20:42 |
user___ | I only use it for android. | 20:42 |
user___ | Even as bad as it is, it beats the on screen torture keyboard. | 20:43 |
user___ | Speaking of new browsers, and Chrome "divorce": now is a good time to push non Google standards. They have ceased contributing progress, they now contribute opt-in nd onboarding useless features imo. | 20:44 |
nemo | I do love my bluetooth keyboard when working in linux on my phone. I use a folding aluminium one. it does sometimes eat characters, but works a lot better than on-screen keyboard. | 20:45 |
user___ | So websites etc should be coded to stick with published possibly frozen standards and not follow every crazyness foisted upon up from G. | 20:45 |
nemo | I've always been a fan of that. Matomo instead of Google Analytics. Local font hosting even if the font was generated by google. | 20:45 |
user___ | I am still miffed by the "security" policy implemented recently which prevents AJAX loading from file: domains. | 20:45 |
nemo | but ofc there are other options too for fonts | 20:45 |
user___ | nemo: what brand/model bt kbd? | 20:46 |
user___ | https://forums.theregister.com/forum/all/2021/01/25/use_firefox_fedora_chromium_maintainer/ the comments are as always golden on thereg. | 20:47 |
nemo | user___: ummm not a particular brand. it's a folding aluminium keyboard. folds in 3 pieces - I like that one because middle section is stable on my leg when typing. it has magnets for a carrying case that doubles as a standard, and due to magnets works well on like hood of my car. there are dozens of companies selling knockoffs | 20:47 |
nemo | usually between $10-30 | 20:47 |
nemo | I've bought a few over years due to losing them, having them stolen, or accidentally destroying them by leaving out in the rain | 20:48 |
nemo | but overall like how it handles | 20:48 |
user___ | just link me a pic or such. Thanks. | 20:48 |
nemo | https://www.amazon.com/Bluetooth-Keyboard-Arteck-Smartphone-Rechargeable/dp/B015LSEQ28/ um. like this one I guess - but seems to be missing the case and unusually pricey. hunting | 20:51 |
nemo | fits in a jacket pocket though | 20:51 |
user___ | ok, I know that style. Thanks. | 20:51 |
user___ | It does not integrate a pointing device at all, right? | 20:52 |
nemo | https://www.amazon.com/Plugable-Foldable-Compatible-Full-Size-Bluetooth/dp/B019PIXO78/ here | 20:52 |
user___ | Aka mouse clit mouse etc? | 20:52 |
nemo | 's one with case and an insanely high price | 20:52 |
nemo | there's various ones with touchpad, I never bought them because note 4 had a stylus with hover and right click. | 20:52 |
unixbsd | you asked me about the pdf editor, so now, it is available: https://gitlab.com/openbsd98324/xpdfedit (first vers, is bit complicated to use, but it will be improved). | 20:53 |
nemo | this is probably wildly offtopic tho | 20:53 |
user___ | yeah, thanks and let's stop. There's #devuan-offtopic for this | 20:53 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!