| rrq | mmm and to avoid conflict .. there could be an acme-nameservers attribute handled by due scripting | 00:00 |
|---|---|---|
| rrq | and then maybe soonish also a new useful package "ifupdown-acme-nameservers" ? | 00:02 |
| onefang | There are several dig like tools, though some of them are just wrappers around dig. Meh, got more important things to do today. | 00:05 |
| systemdlete | why stop there? ifupdown-acme-nameservers-systemd | 00:06 |
| systemdlete | :p | 00:06 |
| rrq | yes, maybe some slight dissonance btw acme "minimalism" and "systemd" | 00:08 |
| systemdlete | turns out I might not need to update the resolv.conf file after all... sorry for pulling you all into my config issues with that, although it seems like it did spur some constructive discussion. | 00:13 |
| systemdlete | and my network config is working!!! | 00:13 |
| onefang | Yay! | 00:14 |
| systemdlete | yes, having one interfaces file is much, much simpler | 00:14 |
| onefang | Now I gotta figure out what's up with repo.jing.rocks. Did Debian update their key even more recently than I did in my test scripts, but no one else has removed the old key from a few months ago? | 00:15 |
| systemdlete | thanks to rrq, rwp, fsmithred, and onefang (and any other fang, if such exists) | 00:16 |
| onefang | Ah, could be, other mirrors are reporting the same, and looks like it's all the ones that have their own Debian mirror, instead of redirecting to Debian. | 00:17 |
| onefang | Well, since you asked, onefang refers to my 63 year old baby tooth being so tiny, it looks like I have only one fang, but there's an adult tooth above it, so I have three actually. | 00:18 |
| rrq | (not much can be said about that without close-up pictures :) | 00:19 |
| systemdlete | here's the web page that talks about dns-nameservers, syntax of interfaces, yada yada: https://wiki.debian.org/NetworkConfiguration | 00:22 |
| systemdlete | (sorry for not posting that earlier when you asked) | 00:23 |
| onefang | Yep, three months after their last key change, Debian has changed their key once more. | 00:27 |
| onefang | One of these days I'll get around to making apt-panopticon cleverer about this sort of thing. It's on the TODO. | 00:28 |
| rwp | We had a huge storm dump a massive amount of rain and now I have had a water incursion in my basement that I am trying to clean up. In 30 years this is the first time the basement has gotten wet. It was a huge downpour though. | 04:55 |
| rwp | Oops. Mistaken recall posting on that one. Please ignore. | 04:56 |
| xnovax | The following signatures were invalid: BADSIG 94532124541922FB Devuan Repository (Primary Devuan signing key) | 05:14 |
| xnovax | how to fix? | 05:14 |
| adhoc | a simple way to test, might be to a DNS lookup for the hostname of the repo, pull out the IPs and pin them in your /etc/hosts file and try again. rotate through IPs until you find the one giving this erro. | 05:16 |
| adhoc | er, unless that is the packages at install time? | 05:17 |
| rwp | Is your clock set correctly? Clocks wrong causes signatures to fail. | 05:17 |
| rwp | If not that then I would select a single mirror and use it instead of the deb.devuan.org round-robin or other that is the default. | 05:19 |
| fluffywolf | on that topic, I ran into a wrong clock on my daedalus install, which is likely a bug, if anyone wants to take an irc bug report... even though I selected Pacific in the installer, the installed system was set up for UTC. had to reconfigure it after installing. | 05:20 |
| fluffywolf | whoops, excalibur install, not daedalus install. | 05:20 |
| rwp | That's another one where I just say that's pretty odd. The timezone setting stuff is pretty basic. | 05:26 |
| adhoc | rwp: how much does it need to be out by? | 06:29 |
| adhoc | do not believe i've had that issue, everything has ntp configured =) | 06:29 |
| rwp | The problem is very apparent on Raspbery Pis which do not have a hardware clock. | 06:30 |
| rwp | Booting a RPi (without the package that saves the data to a file and sets it back at boot time) will set the date to 1970 and that's old enough that all certs are invalid as too far in the future. | 06:31 |
| adhoc | ew | 06:31 |
| rwp | Most Let's Encrypt certs will be valid for 90 days. So any clock which is 90 days in the past or future will think the cert is not valid. | 06:32 |
| adhoc | indeed. | 06:32 |
| rwp | But on average certs will be seen with variable amounts of time left. So 45 days on average in that case. Or maybe even a couple of hours if one is "lucky" to hit things just right. | 06:32 |
| rwp | We never heard back from xnovax so don't really know why they got a BADSIG there. The repository seems okay to me at this time here. | 06:33 |
| rwp | A mirror though could also be mangled. Or there might be a proxy in the middle that is mangling things. | 06:34 |
| rwp | There are almost an infinite number of ways for something to fail. But there is only a limited number of ways for things to succeed. | 06:34 |
| adhoc | is this a server /mirror that has not updated correctly yet ? | 06:34 |
| onefang | Entirely possible. | 06:34 |
| rwp | Look in the scrollback to an hour ago at :14 for a question from xnovax who was getting a BADSIG. | 06:35 |
| onefang | I suggest anytime there's something that might be a package mirror problem, get the IP so I know which one to investigate. | 06:35 |
| adhoc | when i maintained my own mirror at home, i would get this whe nthe mirror script timed out | 06:35 |
| adhoc | the repo is in an inconsistent state | 06:35 |
| onefang | Though at the time I was out shopping. | 06:36 |
| rwp | Most often though the problem with mirrors and proxies results in a single package failing the hash check fingerprint on individual deb files. | 06:36 |
| onefang | Which is why apt-panopticon tests several packages, including specifically picking a new one / update. | 06:37 |
| onefang | Can't test them all though. | 06:37 |
| rwp | Also many of us might remember, a while back, when the signing key was allowed to expire. That created some problems. Which are thankfully in the past now. | 06:38 |
| adhoc | well, you can walk all the records in the DNS RR response. | 06:38 |
| onefang | I dunno, found out that Debian has changed their signing key twice in the last few months. Dunno why yet. | 06:38 |
| onefang | Can't test all the packages. Walking the DNS-RR to get all the mirrors is exactly what I do. | 06:39 |
| adhoc | ah nice | 06:39 |
| adhoc | how often do mirrors get updated ? | 06:39 |
| onefang | apt-panopticon does a bunch of tests on package mirrors. | 06:40 |
| onefang | http://veritas.devuan.org/apt-panopticon/results/Report-web.html includes that info in the Updated coloumn. I try to get them updated every 30 minutes, but some admins want to run slower. | 06:40 |
| onefang | Also in that column you'll see some have a few warnings. Coz they are still within their expected update window, but haven't fully updated yet. | 06:41 |
| onefang | Then there is pkgmaster, not tested for updates, coz that's the source of everyone else's updates. | 06:43 |
| adhoc | wow | 06:43 |
| paculino | I have made an alias for commands to change both the screen temperature and the backlight, as this provides finer control than the gui. But I've noticed that the automatic screen dimming from inactivity will reset to what the GUI is set to upon activity. Is there a workaround to get the screen dimming to return to the actual brightness rather than what the gui says? | 06:44 |
| onefang | So if someone's reporting a problem that might be on a package mirror, if I know which mirror, I can start looking there and dig deeper as needed. | 06:45 |
| adhoc | onefang: got it. | 06:45 |
| adhoc | is there a database data visualiser tool in devuan ? | 06:52 |
| adhoc | something i can look at the table views of postgresql or mariadb. | 06:52 |
| onefang | Adminer is what I use for poking at databases. It's a web thing, but runs on it's own port. | 06:56 |
| onefang | Though if I recall I had to create a startup script for it, it only came with a systemd unit. | 06:56 |
| onefang | Simple one liner that basically copied the relevant bit of the unit did the trick for me. | 06:57 |
| adhoc | ok, thanks. | 06:58 |
| freaxeh | ok | 07:00 |
| freaxeh | how do i install amd graphics drivers | 07:00 |
| freaxeh | for my rx580 | 07:00 |
| freaxeh | i ran l4d2 and it just stutters and has <1fps | 07:01 |
| freaxeh | more like 0.0001 fps | 07:01 |
| freaxeh | not in the menus tho strangely | 07:02 |
| freaxeh | nm they are already installed | 07:14 |
| freaxeh | takes like 10+ minutes to launch l4d2 too, and the mouse freezes while its doing its thing | 07:18 |
| gnarface | freaxeh: make sure you have the package "firmware-amd-graphics" | 07:19 |
| freaxeh | not exactly a stellar impression | 07:19 |
| freaxeh | roger | 07:19 |
| freaxeh | i tried installing that earlier and for some reason the mirror is down... | 07:19 |
| freaxeh | at least i couldn't access it | 07:19 |
| freaxeh | deb.devuan.org | 07:19 |
| onefang | Which IP? | 07:19 |
| gnarface | it's a dns round-robin, if you retry it should work. | 07:20 |
| onefang | Coz that's a DNS-RR with a bunch of servers behind it. | 07:20 |
| freaxeh | roger | 07:20 |
| onefang | And as I said above, if I know which of those servers is the problem one, I can investigate. | 07:20 |
| gnarface | freaxeh: make sure you have these mesa packages too https://paste.debian.net/1326450/ | 07:20 |
| gnarface | (ignore libgl1-mesa-swx11, that's cruft) | 07:20 |
| freaxeh | does apt have a way to display the ip address of the domain its connecting to? | 07:22 |
| freaxeh | or should i just ping it | 07:22 |
| gnarface | not sure, it's easy to just sniff the traffic though... | 07:22 |
| freaxeh | whats the command to resolve a hostname if there is one | 07:23 |
| gnarface | there's a few | 07:23 |
| freaxeh | i'm only familiar with wireshark | 07:23 |
| gnarface | wireshark would work fine | 07:23 |
| gnarface | jnettop or tcpdump would also work | 07:23 |
| onefang | Hmmm, no such thing as libgl1-mesa-swx11 on my desktop frankenDevuan. Cruft that has been removed, or added? | 07:23 |
| gnarface | onefang: "rc" < something that should have been removed | 07:23 |
| gnarface | onefang: (and possibly from a sid install before devuan jessie, not something to worry about) | 07:24 |
| * onefang goes back to worrying about other things. There's plenty of those. lol | 07:24 | |
| gnarface | freaxeh: keep in mind that whatever you get on one lookup try would be randomized again for the next lookup try. you could use "nslookup deb.devuan.org" to see the whole list, but you'll have to actually get your test on a run that fails to know for sure which one is down | 07:26 |
| freaxeh | so far multiple tries fail | 07:27 |
| onefang | Have you tried apt-get, apt, and aptitude? | 07:30 |
| freaxeh | i've tried apt | 07:31 |
| freaxeh | and synaptic | 07:31 |
| yeti | just pinging the dns aliases I see 100% loss with 160.16.137.156 and 147.78.194.22 | 07:44 |
| gnarface | freaxeh: maybe you have a DNS issue locally or at your ISP... make sure the contents of /etc/resolv.conf are right | 07:46 |
| adhoc | ping is less useful, have you tried tracroute or tracepath or mtr ? | 07:47 |
| gnarface | hmm, yea they're not all up but they're not all down either | 07:47 |
| yeti | adhoc: if you know better, just do it. | 07:49 |
| adhoc | tokyo and amsterdam, not like theuy are co-located | 07:49 |
| adhoc | my devuan box at home is offline for some reason =/ | 07:49 |
| adhoc | ok then, the traceroute to both those IP address fail from my vps in japan | 07:51 |
| gnarface | traceroute -T succeeds for me to 147.78.194.22, could it just be blocking ping? | 07:51 |
| yeti | so down to 1 potential problem \o/ | 07:52 |
| freaxeh | how do i update resolv conf | 07:53 |
| freaxeh | it gets overwritten on reboot | 07:53 |
| adhoc | gnarface: good point, but no root to run it there =/ | 07:53 |
| gnarface | freaxeh: if you're using dhcp it's probably set by your dhcp server. if you're using a gui for static network config it's probably set in there instead | 07:53 |
| freaxeh | right | 07:54 |
| gnarface | freaxeh: a couple of them might be down, but you shouldn't be getting 100% failure rates so it's probably a problem with your network one way or another... maybe your ISP's DNS servers just went down? | 07:55 |
| adhoc | some servers do not RR the DNS entries much. | 07:56 |
| adhoc | some times I get stuck with the same IP on a server that is down for a few hours, | 07:56 |
| freaxeh | i just switched to quad9 and it fixed it | 07:56 |
| adhoc | so I pull one of the other IPs and shove it in the /etc/hosts files | 07:56 |
| adhoc | after doing a traceroute to find the closest one | 07:56 |
| gnarface | freaxeh: if it turns out to be a persistent problem with your ISP, you can easily just run a local DNS server instead | 07:57 |
| adhoc | quad9 used to have some privacy concerns | 07:57 |
| adhoc | not sure what their policies are now | 07:58 |
| onefang | If you want to pick a close mirror, https://pkgmaster.devuan.org/mirror_list.txt lists their countries at least. | 07:58 |
| gnarface | freaxeh: let me know if those packages fix it; you should be getting like 300FPS in l4d2 with that hardware | 07:58 |
| gnarface | freaxeh: (native or proton, but i recommend native) | 07:58 |
| adhoc | BBL | 08:00 |
| onefang | Both of those mirrors have been passing the apt-panopticon tests. Though the 147.78.194.22 (mirror.ungleich.ch) has a complication with the way they set up their DNS. The tests are run from two different locations, soon to be three again. | 08:04 |
| freaxeh | gnarface: about 6 packages in that list are missing from my repo https://paste.debian.net/1326450/ | 08:07 |
| gnarface | freaxeh: well, maybe they're deprecated ones, this is an upgraded system... | 08:10 |
| freaxeh | ok | 08:10 |
| gnarface | you're not missing mesa-utils, or any of the ones with the version "22.3.6-1+deb12u1" are you? | 08:10 |
| freaxeh | gnarface im launching l4d2 now | 08:10 |
| gnarface | (exact versions not really important, i haven't upgraded in a few weeks, the important part is you get the latest version of the ones l4d2 needs, and i'm not sure) | 08:11 |
| freaxeh | ayeee it works | 08:12 |
| freaxeh | high fps | 08:12 |
| gnarface | cool | 08:12 |
| gnarface | putting settings on max shouldn't even make a dent | 08:12 |
| gnarface | in fact i recommend you turn on vsync to avoid overheating your video card unnecessarily | 08:12 |
| onefang | Yay! | 08:12 |
| freaxeh | dont think so gnarface | 08:16 |
| freaxeh | so far so good, its more stable than linux mint, hasnt crashed once yet | 08:28 |
| onefang | Even more yay! B-) | 08:28 |
| onefang | Or, um, give it time. | 08:29 |
| freaxeh | she crashed | 08:53 |
| gnarface | proton or native? | 08:59 |
| gnarface | i'm telling you though... try vsync, it might be overheating | 09:00 |
| gnarface | you got lm-sensors installed? | 09:00 |
| xnovax | rwp: sorry for the late response, ended up finding the issue, was on my end, with the caching proxy. | 09:02 |
| xnovax | just did some rm'ing of the bad cache | 09:03 |
| freaxeh | gnarface: vsync is enabled so she aint overheating at least i dont think she is, i'll install lm-sensors and check. | 12:25 |
| freaxeh | how the heck do i use lm-sensors on this system | 12:46 |
| CueXXIII | freaxeh: start "senors" - if that outputs nothing, you might need to run sensors-detect | 13:03 |
| stripinska8 | Your advice from yesterday was very helpful. I deleted the file in etc/apt/apt.conf.d/* then purged apt-listchanges and proceeded with apt --fix-broken install, and to reinstall any missing packages. | 14:38 |
| stripinska8 | for the most part this worked well but I have the following packages which can't be configured it seems because they depend on python3 and python3 can't be configured due to an problem in the file. | 14:39 |
| stripinska8 | Unconfigureable packages python3 python3-setuptools-scm | 14:40 |
| stripinska8 | python3-packaging python3-typing-extensions | 14:40 |
| stripinska8 | emacs-gtk python3-pluggy | 14:40 |
| stripinska8 | python3-trove-classifiers python3-hatchling | 14:40 |
| stripinska8 | sane-utils python3-pathspec | 14:40 |
| stripinska8 | python3-lib2to3 plasma-desktop-data | 14:40 |
| stripinska8 | python3-hatch-vcs python3-pkg-resources | 14:40 |
| stripinska8 | kwin-data emacs | 14:40 |
| stripinska8 | python3-setuptools plasma-desktop | 14:40 |
| stripinska8 | python3-tomli python3-gdbm:amd64 | 14:40 |
| stripinska8 | python3.12-full python3-debconf | 14:40 |
| stripinska8 | python3-tk:amd64 kwin-common | 14:40 |
| stripinska8 | kwin-wayland kde-plasma-desktop | 14:40 |
| stripinska8 | plasma-workspace-wayland idle-python3.12 | 14:40 |
| stripinska8 | kde-standard task-kde-desktop | 14:40 |
| stripinska8 | kwin-x11 plasma-workspace | 14:40 |
| stripinska8 | plasma-widgets-addons kinfocenter | 14:40 |
| stripinska8 | sddm-theme-breeze sddm-theme-debian-breeze | 14:40 |
| stripinska8 | First of many error messages when apt --fix-broken install tries to configure python3 | 14:41 |
| stripinska8 | Setting up python3 (3.12.4-1) ... | 14:42 |
| stripinska8 | running python rtupdate hooks for python3.12... | 14:42 |
| stripinska8 | Traceback (most recent call last): | 14:42 |
| stripinska8 | File "/usr/bin/py3clean", line 210, in <module> | 14:42 |
| stripinska8 | main() | 14:42 |
| stripinska8 | File "/usr/bin/py3clean", line 196, in main | 14:42 |
| stripinska8 | pfiles = set(dpf.from_package(options.package)) | 14:42 |
| stripinska8 | In spite of these errors the box boots and runs fine. I assume some of this will resolve automatically at some time and if anything gets too annoying I will dig into it another day to see if I can find the actual problem. | 14:44 |
| stripinska8 | Thank you all for your help. | 14:44 |
| golinux | Not a good idea to post multiple lines at once. You may get kicked. Post elsewhere and link | 16:26 |
| rwp | xnovax, Are you using apt-cacher-ng? I had to stop using it in order to avoid problems. | 16:52 |
| dvbst | hello, how do i install devuan with the installer but use my own partitions? (the catch: they are encrypted) | 19:39 |
| dvbst | and also, do i just leave the partitions as they are or do i put some filesystem there before i let the installer run the rest of the show? | 19:43 |
| fsmithred | dvbst, I think if you unlock the encrypted volume(s) first and then run the partitioner, it will see them and you can select them for use. (/dev/mapper/whatever) | 19:44 |
| fsmithred | I'm not sure if you have to go to another console to open them manually or if you can run the advanced boot to rescue or whatever it's called and then back up into the installer. | 19:45 |
| dvbst | okay but will it make the filesystems itself or does it just expect to dump files on whatever i provide? | 19:46 |
| fsmithred | when you set up which partitions will be used, you can select whether to format it or not | 19:46 |
| dvbst | i see | 19:46 |
| plasma41 | dvbst: [Obligatory warning to make backups first] | 20:10 |
| dvbst | theres nothing to backup | 20:10 |
| plasma41 | So you encrypted... empty space? | 20:14 |
| dvbst | yea, im about to | 20:33 |
| fsmithred | Are you using an installer iso or a live iso? | 20:35 |
| dvbst | netinstall | 20:40 |
| dvbst | i mean, i made the partitions on this other live iso cuz for the love of god i cannot use the shell in the netinstall | 20:41 |
| dvbst | that thing doesnt even have lsblk | 20:41 |
| sgage | dvbst: you can manually install to the existing partion(s) | 21:04 |
| sgage | of your choice with the manual option in the partition-selection | 21:05 |
| sgage | phase of the install. Make sure to NOT check the format box! | 21:05 |
| sgage | and your existing setup will be all good. | 21:06 |
| dvbst | what would the format box do | 21:06 |
| sgage | Probably mess up your encryption scheme. No need to format if | 21:07 |
| sgage | you're planning on installing into an empty partition. | 21:07 |
| dvbst | but do i need to put a file system on there? | 21:08 |
| sgage | Oh, I thought you already had given it a fs. Use gparted or something | 21:10 |
| sgage | and install an fs. To be honest, I'm not sure how it all works | 21:10 |
| sgage | with encryption. | 21:11 |
| dvbst | damn | 21:12 |
| sgage | I'm not really sure what the installer would do to your | 21:15 |
| dvbst | how do i list devices without lsblk | 21:16 |
| sgage | encryption if you formatted it. I think that if you do choose to | 21:16 |
| sgage | have the installer format it, it asks for a filesystem to format to. | 21:16 |
| sgage | Not sure what would happen to the encryption - not something | 21:17 |
| sgage | I've played with. | 21:17 |
| xnovax | rwp: yes acng. | 21:26 |
| xnovax | rwp: rarely do i ever have issues with it across multiple devuan/debian hosts. 99% of the time it works fine, just a random fluke here and there | 21:27 |
| dvbst | so i made the partitions on the other iso, but didnt format them, now i opened the encrypted drive, and i want to mkswap the swap partition but mkswap says it cant open it | 21:29 |
| dvbst | but lvdisplay shows both things correctly | 21:30 |
| fsmithred | Oh, it's encrypted lvm? Why won't you use the installer to do that? | 21:33 |
| dvbst | because it falls apart | 21:34 |
| dvbst | didnt i say that before? | 21:35 |
| fsmithred | not sure | 21:35 |
| dvbst | so thats what i wanted to do before too | 21:36 |
| dvbst | and i selected the disk that i want to format | 21:36 |
| dvbst | all went nice | 21:36 |
| fsmithred | I always get lost doing encrypted installs. | 21:36 |
| dvbst | but for some reason it formatted that disk, and also the live iso | 21:36 |
| dvbst | and as you know, it wasnt really able to do much else | 21:37 |
| fsmithred | oh yeah, you mentioned that yesterday, I think | 21:37 |
| dvbst | but the installer was never good | 21:37 |
| dvbst | it worked, and the system it made was working fine | 21:37 |
| dvbst | but every single time, it makes a sda1 for boot, (thats good), a 1bk sda2 with nothing on it, then some empty space that lsblk doesnt list and sda5 is the encrypted partition with lvm on it | 21:38 |
| dvbst | like how do you even fuck it up? | 21:38 |
| dvbst | who wrote this? | 21:38 |
| dvbst | but the upside is that it does in fact work | 21:39 |
| dvbst | and the daedalus installer doesnt | 21:39 |
| fsmithred | I always use manual partitioning | 21:39 |
| dvbst | yea thats what i shouldve done too | 21:39 |
| dvbst | but i noticed it months after i installed it on both of my computers | 21:40 |
| dvbst | and was too lazy to reinstall | 21:40 |
| dvbst | nvm, the mkswap is good, its just that the lvs werent active | 21:43 |
| dvbst | wow, even with the manual installation it wants to tamper with the install media | 21:51 |
| dvbst | unable to install grub in dummy | 22:11 |
| dvbst | excecuting 'grub-install dummy' failed | 22:11 |
| dvbst | this is a fatal error | 22:11 |
| dvbst | what now? | 22:11 |
| plasma41 | dvbst: What disk are you attempting to install grub onto? | 22:14 |
| dvbst | what exactly are you asking for? | 22:16 |
| dvbst | its a m.2 sata ssd | 22:17 |
| dvbst | if thats what you mean | 22:17 |
| plasma41 | What is 'dummy'? The argument to grub-install should be a device such as /dev/sda. | 22:18 |
| dvbst | i have no clue | 22:24 |
| dvbst | thats what the installer did | 22:24 |
| dvbst | ok so it says that this step depends on the partitioning step | 22:36 |
| dvbst | and im pretty sure ive set up my /boot partition correctly | 22:37 |
| dvbst | the /dev/sda1 is a primary part, has 1gb of space there, use as ext2 file system, mount point /boot, defaults, bootable flag on | 22:38 |
| dvbst | is there something i am doing wrong? | 22:38 |
| rwp | xnovax, apt-cacher-ng for reasons not completely understood (but I have suspicions) does not work well with the rsync mirror update of Devuan mirrors. I still use it for Debian mirrors. But I stopped using it for Devuan mirrors. | 23:18 |
| dvbst | i just checked, and im using the same parameters as the boot partition on a system where the installer managed to successfuly install all the stuff with encrypted lvm | 23:19 |
| dvbst | i have no idea what could the problem be, the mountpoint is specified and all that | 23:19 |
| onefang | I've had no problem with apt-cachec-ng, but only used it during testing. | 23:20 |
| onefang | On the other hand, my testing involved "install Devuan to a test VM, test it, repeat", so it was being hit hard. But I was only using my own mirror, so perhaps it's other mirrors? | 23:27 |
| rwp | So when I have had the time to debug particular files with hash mismatches everything always seemed perfectly correct about apt-cacher-ng following all of the rules that it had been told. But the files were stale regardless. | 23:29 |
| rwp | In particular a mirror serving the files with a web server has set a TTL on the file. It is still within the cache TTL time. But if I fetch from upstream I can see the file has been updated. Which changed the file on disk through rsync out from under the web server's cache time. | 23:30 |
| rwp | I didn't spend the full time needed for me to write a congress and feel confident presenting it under cross examination though. So I haven't really dug into it more. But I think things like that are the problem. | 23:32 |
| rwp | Another thing that makes this inconsistent is that every mirror sets their own web server up uniquely different. Round Robin DNS means files are being pulled from here, from there, from another place, and each web server sets their own cache time on the files. | 23:34 |
| onefang | I have a TODO item to investigate the stuff Debian uses for updating their mirrors. | 23:34 |
| rwp | Life and time is what keeps things from happening all at once. | 23:34 |
| rwp | The fallback is to not use a caching proxy. And pretty much so far as far as I can tell that always works. | 23:35 |
| onefang | Crisis after crisis is what keeps me from getting shit done. lol | 23:35 |
| rwp | It's been like that for me too. Travel. Family stuff. Volunteer stuff. Storms draining being overwhelmed and making my basement wet. It's a problem! | 23:36 |
| onefang | Keep me informed if you find out anything more about these apt-cacher-ng problems. I really only installed and tested it coz you where having problems, but I didn't spot them myself. | 23:38 |
| rwp | If I can find the time I have this idea as a way to verify a test case. I will certainly tell you if I find anything. | 23:42 |
| onefang | Cool. | 23:42 |
| rwp | But you can't deny that there are many reports with that combination reporting problems. I think we would say, where there is smoke there is fire. | 23:42 |
| onefang | Which is why I have TODO items for these things. | 23:43 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!