| gnarface | Ghost2: it might work, but installing stuff from newer releases into older releases is not a supported. you're supposed to recompile it. | 00:01 |
|---|---|---|
| gnarface | *not a supported action | 00:01 |
| gnarface | why, you're welcome. of course i'll help you rebuild it. using the packaging tools it should be easy! | 00:04 |
| debdog | hehe | 00:04 |
| hacksenwerk | Hi. How can I find out why dhcp takes a long time at boot? The routers are on and it is a raspberry p400. In #raspberrypi they only trolls to use systemdeath for that task... | 13:15 |
| hacksenwerk | It takes at least a minute or so. | 13:15 |
| djph | use wireshark to watch the network | 13:25 |
| hacksenwerk | djph: o0 | 13:27 |
| hacksenwerk | That's a monster is that really the only way? | 13:27 |
| hacksenwerk | I mean it is not a big deal, I was just wondering why it takes so long and looked at the boot output why connecting my headless pi to a monitor. | 13:29 |
| hacksenwerk | *while | 13:29 |
| djph | hacksenwerk: I mean if you want to know _WHY_ DHCP takes so long ... watch the network for the handshaking | 13:32 |
| djph | also, are you connecting with a wired connection, or wifi? | 13:33 |
| buZz | if dhcp takes -very- long and you dont know why; i've found that 'send dhcp hostname' often causes it | 13:54 |
| buZz | you might be able to make your dhcp client send its hostname along with the request, and see it get a lease faster | 13:54 |
| buZz | something like 90% of the cases i've debugged , this was the case | 13:54 |
| hacksenwerk | sorry was afk | 14:45 |
| hacksenwerk | djph: I use wired connection | 14:45 |
| hacksenwerk | buZz: How do I make this? | 14:47 |
| buZz | hacksenwerk: add 'hostname' on a line in dhcpcd.conf | 14:49 |
| buZz | according to https://manpages.ubuntu.com/manpages/trusty/man5/dhcpcd.conf.5.html | 14:49 |
| buZz | (i dont know what dhcp client you're using) | 14:49 |
| hacksenwerk | buZz: I have isc-dhclient-4.4.3-P1 on that p400 | 15:12 |
| buZz | cool | 15:12 |
| hacksenwerk | buZz: There's no dhcpcd.conf bit a /etc/dhcp/dhclient.conf | 15:14 |
| buZz | try 'man dhclient.conf' then ;) | 15:14 |
| buZz | and find the relevant option | 15:14 |
| hacksenwerk | and there's this line: send host-name = gethostname(); | 15:15 |
| buZz | alright, and its got a unique hostname? | 15:15 |
| buZz | not the 10th 'raspberrypi' on your lan? :P | 15:15 |
| hacksenwerk | the pi? yes. | 15:15 |
| hacksenwerk | no | 15:15 |
| buZz | oh, another thing might be having ipv6 enabled | 15:15 |
| hacksenwerk | it has a unique one | 15:15 |
| buZz | especially on some SystemD setups that causes high delays | 15:15 |
| hacksenwerk | buZz: ... tehre's no systemdeath... | 15:16 |
| buZz | (on networks without proper configged ipv6) | 15:16 |
| hacksenwerk | I use devuan | 15:16 |
| buZz | alright | 15:16 |
| hacksenwerk | We are in #devuan ;P | 15:16 |
| buZz | then i'm out of ideas, check traffic or check logs :) | 15:16 |
| hacksenwerk | /var/log/boot ? | 15:17 |
| buZz | dhcp client logs, dhcp server logs | 15:17 |
| buZz | btw ; you can completely ignore this issue and just give your device a -static- ip ;) | 15:17 |
| buZz | s/can/could/ | 15:18 |
| hacksenwerk | man dhclient "In either case, the client logs messages using the syslog(3) facility | 15:21 |
| hacksenwerk | syslog is full of dhcp stuff from the last decades... | 15:21 |
| hacksenwerk | There's this at /var/log/boot: https://paste.debian.net/plainh/7862ed11 | 15:21 |
| hacksenwerk | As you can see from here on: Wed Jun 14 13:06:23 2023: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 6 it does nothing for a minute | 15:22 |
| buZz | it got DHCP within 1 second there | 15:22 |
| buZz | from 13:07:31 | 15:22 |
| buZz | oh,Wed Jun 14 13:07:31 2023: Internet Systems Consortium DHCP Client 4.4.3-P1 | 15:22 |
| buZz | thats the first line | 15:23 |
| buZz | it didnt get any dhcp OFFER on the physical network before | 15:23 |
| buZz | if it worked for 'decades' (weird, its a pi400, not even 10yo) and this SUDDENLY started, replace the network cable | 15:23 |
| hacksenwerk | buZz: but that's weird, caus it also does this, when I had a connection via ssh before and the reboot from ssh without touching the routers or something | 15:24 |
| hacksenwerk | buZz: the cable? | 15:24 |
| buZz | sure, intermittent connections are intermittent | 15:24 |
| hacksenwerk | wait! | 15:24 |
| buZz | ah lol , i see | 15:24 |
| hacksenwerk | could it take that long because the cable is very long? | 15:24 |
| buZz | -first- it tried eth0 | 15:24 |
| buZz | -second- it tried eth1 | 15:24 |
| buZz | eth0 didnt give a link | 15:25 |
| buZz | eth1 did | 15:25 |
| hacksenwerk | buZz: yes both interfaces are in /etc/network/interfaces | 15:25 |
| buZz | which 'both' , pi400 has a single ethernet port | 15:25 |
| hacksenwerk | Just in case my ethernet-to-usb adapter breaks at some point in the future | 15:25 |
| hacksenwerk | buZz: I use an adapter | 15:25 |
| buZz | eth0 is not connected | 15:25 |
| buZz | yet its doing dhcp over it | 15:26 |
| buZz | which indeed never gets a ip | 15:26 |
| hacksenwerk | eth0 is the ethernet port, eth1 is a usb port | 15:26 |
| buZz | eth0 doesnt give DHCP | 15:26 |
| buZz | eth1 does | 15:26 |
| buZz | read your own log again | 15:26 |
| hacksenwerk | or reverse i dunno | 15:26 |
| buZz | its doing dhcp on BOTH | 15:26 |
| buZz | yet eth0 isnt working | 15:26 |
| hacksenwerk | and it doesn't matter cause I always use the adpater | 15:26 |
| buZz | cool, so you found the issue now, enjoy :) | 15:26 |
| hacksenwerk | :/ | 15:27 |
| buZz | TLDR ; disable doing dhcp on interfaces that arent working | 15:27 |
| hacksenwerk | buZz: no | 15:27 |
| hacksenwerk | but how about this: | 15:27 |
| hacksenwerk | placing eth1 above eth0 in /etc/network/interfaces ? | 15:28 |
| hacksenwerk | Is this file red from top to bottom? | 15:28 |
| hacksenwerk | So it would first try eth1 | 15:28 |
| buZz | i dont understand what you're trying to achieve, ALL interfaces with dhcp enabled will do dhcp | 15:28 |
| buZz | it would still be hanging trying to get a ip for the interface that you configured as dhcp | 15:29 |
| hacksenwerk | buZz: so it would bound on eth1 and then do that "waiting" | 15:29 |
| hacksenwerk | ok | 15:29 |
| buZz | it will still be delaying | 15:29 |
| hacksenwerk | The problem is, that that adapter is always eth1 and if it will break one day, I will not be able to connect via the ethernet port | 15:30 |
| hacksenwerk | So I need to keep that in interfaces | 15:30 |
| hacksenwerk | buZz: But thanks for your help, I've learned something again. :) | 15:31 |
| buZz | <3 | 15:32 |
| buZz | hacksenwerk: everything 'breaks one day' , i dont see why you need to burdon today just to safe tomorrow :P | 15:32 |
| buZz | i dont downclock my desktop cpu to 'make it work more days' either :P | 15:32 |
| buZz | btw; pi4 ethernet isnt a USB device, but a PCIe device | 15:36 |
| buZz | so its faster | 15:36 |
| hacksenwerk | buZz: It is not about "faster", it is about firmware | 15:37 |
| hacksenwerk | The adapter runs on the usb 2 port | 15:37 |
| hacksenwerk | https://libreboot.org/faq.html#nic-ethernet-controller | 15:38 |
| hacksenwerk | buZz: further I never felt a difference in speed :) | 15:39 |
| hacksenwerk | It has 100Mbps that's more than I pay for ;) | 15:39 |
| buZz | oh your only network usage is internet? | 15:40 |
| buZz | yeah likely doesnt matter much then | 15:40 |
| hacksenwerk | buZz: yes and local network | 15:40 |
| hacksenwerk | I ssh to that pi with my laptop | 15:40 |
| buZz | hacksenwerk: btw, USB network cards have firmware too | 15:40 |
| buZz | you just cant read the firmware | 15:40 |
| buZz | its internal on the device | 15:40 |
| hacksenwerk | buZz: I know, but it for example has no dma, at least 2.0 | 15:41 |
| hacksenwerk | usb 4.0 has it and 3.0, or 3.1 I'm not sure | 15:41 |
| buZz | yeah lol, i dont understand avoiding dma :D | 15:41 |
| hacksenwerk | It is just the traffic goes over your cpu first | 15:41 |
| buZz | no, DMA is -direct- memory access | 15:42 |
| hacksenwerk | yes | 15:42 |
| buZz | doesnt go through CPU | 15:42 |
| buZz | all USB traffic does | 15:42 |
| hacksenwerk | not-dma goes over cpu | 15:42 |
| buZz | indeed, so makes it slower -and- higher load on cpu | 15:42 |
| hacksenwerk | buZz: you will not notice it trust me | 15:42 |
| hacksenwerk | and I'm on an old machine here :D | 15:43 |
| buZz | hacksenwerk: -i- use a network to capacity | 15:43 |
| buZz | not 'just my internet' | 15:43 |
| hacksenwerk | buZz: I have problems understanding that | 15:43 |
| hacksenwerk | buZz: Can you rephrase it please? | 15:44 |
| hacksenwerk | Look: I try to avoid as many unfree hard- and software as possibel. So when most of my peripherals run over the same firmware, like usb, then I do not have to care so much about firmware in other interfaces, like cdrom devices, hdds and ssds and so on. | 15:48 |
| hacksenwerk | And you really don't have to worry about having slower tasks, you can not notice that if that is the case, or maybe only on some machine that is even much older then mine. | 15:49 |
| hacksenwerk | But only as long as you use a usb specification that has no dma,wich is why I avoid everything over usb 3 (the pi has it but i only use it for dvd ripping and nothing more). | 15:52 |
| hacksenwerk | *over usb 2 | 15:52 |
| blizzow | I'm futzing about with the Raspberry Pi 4 image. I imaged a USB stick with the daedalus image and promptly changed /etc/apt/sources.list to ceres and ran 'apt full-upgrade' | 16:16 |
| blizzow | A few things: I wanted to encrypt my root partition so only boot was unencrypted. Anyone have an idea on how to do this? | 16:18 |
| blizzow | Also, I wanted to make sure opengl hardward acceleration is available. The typical boot partition layout and files for pis is not present in /boot. How might I enable opengl? | 16:19 |
| blizzow | Finally, I only see firefox-esr in ceres. Is there a repo to add to get regular or edge firefox in ceres? I don't understand why unstable only has the long term release of firefox and the release (daedalus) has the latest firefox. | 16:20 |
| blizzow | nevermind about that last point, I see the non-esr in the repo. | 16:21 |
| ted-ious | blizzow: What does full-upgrade to ceres do from daedalus? | 16:25 |
| ted-ious | And how is it different from full-upgrade from chimaera? | 16:25 |
| blizzow | ted-ious, I'm not sure how it's different. I was just operating off a page I found about how to 'upgrade' to ceres. | 16:38 |
| blizzow | ceres being an unstable rolling-release. daedalus being a point in time release. | 16:38 |
| ted-ious | Right but daedalus was just made out of ceres so I was wondering what upgrading does. | 16:39 |
| blizzow | I think it's a grand total of 20 or so packages that get upgraded. | 16:40 |
| rwp | Note that Sid/Ceres has been frozen in preparation for releasing Debian 12 Bookworm. That has happened and now Sid and Ceres are zooming forward. I've had 79 packages upgraded in the first 24 hours and 73 today already. | 18:17 |
| rwp | Ceres will be a wild ride over the next month or months as everything that had been held back by the release freeze will now be allowed through. | 18:17 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!