| systemdlete | So I'm still getting the cacher errors, even with that DNS parameter changed. | 01:31 |
|---|---|---|
| gnarface | i'm not surprised, it was a long shot for DNS to be really involved here | 01:31 |
| systemdlete | I re-ran apt update and now it is having no problems. I did not do the maintenance this time. | 01:32 |
| systemdlete | Just the apt update. | 01:32 |
| gnarface | so, not sure about this, because i was drunk THE WHOLE TIME, but i seem to actually remember a prior employer getting bugged by this and going through extensive troubleshooting, digging into code with the intent to patch, only to come to the conclusions 1) it's a transient repo issue that is not the fault of apt-cacher-ng directly and 2) the only way to fix it would be to set up a periodic cron job that basically does | 01:34 |
| gnarface | the equivalent of that maintenance button + checking all boxes, and 3) it wasn't worth it | 01:34 |
| gnarface | still though, i'd want a solution to the mystery, but i think it might end up leading us back to convergence with onefang's issue about cleaning up the order of how files are actually updated in the repos when they update internally | 01:35 |
| * systemdlete has to restart desktops... | 01:35 | |
| gnarface | but i distinctly remember that employer personally tearing it apart extensively and at the end deciding apt-cacher-ng couldn't actually be improved | 01:35 |
| systemdlete | wouldn't apt itself be vulnerable to the same problem? I just don't experience that, though (at least not often) | 01:36 |
| gnarface | not necessarily, or if it is it might manifest itself in different ways | 01:37 |
| gnarface | occasional failures for apt/apt-get without a proxy have been a thing for me in fact in debian and devuan | 01:37 |
| gnarface | more often with devuan, but aiui that's just because debian has hundreds of mirrors and we only have a dozen or so | 01:38 |
| systemdlete | It is interesting that all I had to do (this time) was re-run apt update on the client | 01:38 |
| systemdlete | gott go... those desktops won't restart themselves | 01:39 |
| systemdlete | bbs | 01:39 |
| gnarface | later | 01:39 |
| * rrq also saw "FreshIndexMaxAge: 27" option in acng.conf | 02:01 | |
| rrq | .. and "ResponseFreezeDetectTime: 60" and "ReuseConnections: 1" | 02:03 |
| gnarface | hmm, this employer i am thinking of was really obsessed with the bandwidth savings, so it's conceivable he may have discovered a solution but merely discarded it because it cut into theoretical bandwidth savings | 02:04 |
| gnarface | (in the end he just decided clicking the maintenance button manually whenever the problem occurred was the best option, but that doesn't mean there's no other way) | 02:04 |
| rrq | yeah, and as the cacher is just a transparent proxy the protocol doesn't include a way for apt to say "this file needs clean download" | 02:10 |
| rrq | (I guess) | 02:12 |
| rrq | .. so that signal has to go via the keyboard monkey | 02:13 |
| gnarface | yea, i think this guy even figured out a way to make a minor edit to apt-cacher-ng so it could be done via cron job without manual intervention, but he quickly decided that it wasted more bandwidth than it was worth. (whether that's true or not in practicality i don't know; i wasn't the one paying for it) | 02:14 |
| gnarface | heh, actually now i recall that that business folded rather quickly and afterwards when tracing the dots between several conspicuous errors in judgement i got the distinct impression they might have scuttled it on purpose as a tax dodge, so frankly maybe this anecdote shouldn't be trusted | 02:20 |
| rrq | :) | 02:20 |
| onefang | Do I have to add all the popular proxies in front of all our apt-panopticon instances? Just to check more failure cases. | 02:23 |
| systemdlete | so on one client I just upgraded I had to apt update twice, and even then, it failed the first apt upgrade. The second one worked though. | 02:24 |
| systemdlete | so all my "fixes" (rerunning maintenance, or rerunning apt update) don't seem to really fix anything. | 02:24 |
| gnarface | onefang: i don't know for sure if that's necessary, but it might be helpful. apt-cacher-ng might end up being a very good canary. | 02:25 |
| onefang | I just apt upgraded / updated / rebooted coz new libc. All good so far. Almost down to ten minutes to log back into everything. lol | 02:25 |
| * onefang adds one more thing to test with my test VM, apt-cacher-ng. One thing that annoyed me about mmdebstrap is that it ignores the pre loaded cache I tell it to use for the first few things, but starts using my on disk cache from then on. | 02:26 | |
| * onefang quotes gnarface. | 02:30 | |
| onefang | I hate multipliers for apt-panopticon. Using up more bandwidth for each mirror, for each test, from each tester, ... | 02:34 |
| onefang | Maybe just a single instance testing behind a proxy, and that proxy being apt-cacher-ng? Then throw the coal mine at it. A good job for my desktop super computer and qemu. | 02:36 |
| onefang | https://sledjhamr.org/mantisbt/view.php?id=524 | 02:37 |
| gnarface | heh, nice | 02:37 |
| nsprra | do any of you run an amd igpu for your display output and an nvidia card for compute, with no display on it? | 02:38 |
| nsprra | when i stick in the nvidia card, boot hangs with 'waiting for /dev/ to be populated' | 02:39 |
| nsprra | i don't have any nvidia drivers installed so .. it's kernel trying to be smart or something | 02:39 |
| gnarface | nsprra: hmm, rings a bell but i'm not having luck coming up with specifics. you made sure eudev is installed, right? typically it's not a problem for a nvidia card to sit there idle without drivers but make sure your system BIOS(/EFI) isn't doing something dumb like forcing it as the primary by default. | 02:41 |
| nsprra | i will look into possible eudev issues thanks | 02:42 |
| nsprra | this is my first EFI system. avoided EFI up to now. | 02:42 |
| gnarface | nsprra: i only mention it because that's what populates /dev, and i did have a recent issue where an upgrade from an older release had removed udev automatically without installing eudev in its place, and it actually succeeded at the upgrade and the first reboot, but then choked on the second reboot. | 02:43 |
| nsprra | boots fine with just the integrated gpu though... hm. | 02:44 |
| nsprra | amdgpu kernel module gets loaded. amd_firmware is installed. | 02:44 |
| gnarface | nsprra: yea, make sure the EFI interface isn't set to use external GPU as the primary by default, then it shouldn't care about /dev/ (that one card anyway) | 02:44 |
| nsprra | right | 02:45 |
| systemdlete | rookie question: How to grab the edge of a window for resizing? I found ancient posts about this, c. 2010. Solutions (at least then) were ALT-middleclick or ALT-rightclick; I tried those but they don't work. | 02:52 |
| systemdlete | sometimes borders are very thin and hard to "catch" | 02:52 |
| systemdlete | hexchat has generous borders, and for some reason, I am able to resize firefox (with skinny borders) without trouble. | 02:53 |
| onefang | That's up to your window manager. | 02:53 |
| systemdlete | ok, thanks. I'll go ask him. | 02:54 |
| onefang | lol | 02:54 |
| gnarface | alt+middleclick+drag is what works here on enlightenment, but i know it was mentioned they purposefully widened the target area to make it easier to grab | 02:54 |
| gnarface | there's no actual requirement for the WM to even support this feature | 02:54 |
| * systemdlete tries to talk to his window manager, but they are in a meeting atm. lol | 02:54 | |
| gnarface | (though the ones i've used typically do) | 02:54 |
| rrq | Alt+F8 it says in my Window Manager settings | 02:55 |
| systemdlete | alt-middleclick-drag doesn't work in my xfce | 02:55 |
| gnarface | alt+F8 is that KDE? | 02:55 |
| rrq | I think its xfce | 02:55 |
| onefang | I map mine to that Windowskeythingy+middleclick+drag on the corners of the windows. | 02:55 |
| gnarface | systemdlete: i don't use xfce much, try alt+f8 | 02:56 |
| rrq | xfwm4 | 02:56 |
| gnarface | oh, hmm, that's older | 02:56 |
| systemdlete | alt-f8 and then what? | 02:57 |
| rrq | that's what my lxqt-session starts ... with xfce installed :) | 02:57 |
| onefang | Whatever window manager I use I map all that window manglement stuff to that WindowsMangledKey. Make it useful. | 02:57 |
| rrq | pushing alt+f8 moves the mouse to the bottom right corner in window-resize mode | 02:58 |
| rrq | for me | 02:58 |
| onefang | Ah, that key is sometimes called "Super". Might be worth poking at it. | 02:59 |
| systemdlete | ah, works on host, but not vm | 02:59 |
| systemdlete | thank you | 02:59 |
| systemdlete | now I just have to figure out how to map that for vbox | 03:00 |
| onefang | I'm still working on remaping Ctrl-Alt-F1 and friends in qemu, but that's low priority, the work around is fine for now. | 03:01 |
| nsprra | transitioning to ceres, sysvinit-core_3.09 fails to install a /sbin/init. | 03:56 |
| nsprra | unbootable situation now. better fix it. | 03:56 |
| nsprra | is there some problem with initramfs-tools and sysvinit-core 3.09 ? | 03:57 |
| nsprra | cause i have no /sbin/init now | 03:57 |
| nsprra | well i don't need a package manager to put init there :) | 04:04 |
| gnarface | nsprra: it's unstable, you know they break it for fun, right? | 04:12 |
| nsprra | yea | 04:12 |
| gnarface | i would just roll that package back to the previous version and wait for debian to fix it | 04:13 |
| gnarface | obviously there's some problem but since it's unstable there's no particular reason to believe they don't already know | 04:14 |
| gnarface | hmm | 04:14 |
| nsprra | thanks for the correction. it's not noteworthy i guess. | 04:14 |
| nsprra | just a new oopsie to me. :) | 04:15 |
| rrq | usrmerge | 04:15 |
| gnarface | is that what's doing it? | 04:15 |
| rrq | well I'm guessing that it installs as /usr/sbin/init which requires a usrmerge for the kernel to find it | 04:16 |
| gnarface | ah | 04:16 |
| gnarface | nsprra: ^ can you check this? | 04:16 |
| nsprra | ohh it does! | 04:16 |
| gnarface | probably simple to fix with a symlink | 04:17 |
| gnarface | rrq: symlinks work, right? | 04:17 |
| nsprra | for init? hmm... hmm... | 04:17 |
| rrq | yes, all of bin sbin lib lib32 and lib64 should be symlinked to usr/$DIR | 04:17 |
| rrq | (note the relative pathname, rather than /usr/$DIR; it is "safer" as it works in bind mounts and such) | 04:20 |
| systemdlete | Istr that, on some version of unix or another, that terminal session directories were restored when a new desktop sessoin started. So if term1 was cd'd into /tmp and term 2 was cd'd into /etc when I shut down, then 2 terminals would be started at session startup (startx, e..g) with one in /tmp and the other in /etc. | 06:50 |
| systemdlete | Currently, xfce does restore the windows with the terminals, but it does not automatically change the directories for me. | 06:50 |
| systemdlete | What I recall did NOT require me to do anything special to, say, my .bashrc. | 06:51 |
| systemdlete | This is not a huge issue, and I can figure out how to emulate that behavior, but it would be nice if someone knew what I was talking about and how it was able to work. | 06:51 |
| systemdlete | (the terminal program and the program running inside of it are not necessarily talking to each other, usally, esp. if it is a shell, which it most commonly is) | 06:52 |
| systemdlete | *usually | 06:52 |
| systemdlete | It could have been some custom environment someone created at one of the places I worked at, idk. | 06:53 |
| systemdlete | I just seem to recall it was more recent than that, possibly on some linux system I worked on here at home. | 06:53 |
| gnarface | systemdlete: doesn't KDE do that by default? | 08:25 |
| systemdlete | that might have been it, actually. I haven't used KDE in ages, so that might be my recollection. Thanks. | 08:44 |
| gnarface | it might not be the only one | 08:45 |
| joerg | KDE alt+rmb = resize, grabbing the wondow border the cursor is closest to. alt+lmb = drag window. alt+mmb sends window to back or something like that | 14:20 |
| nsprra | who removed X from dpkg X (to extract package files) | 17:02 |
| nsprra | so the debian/devuan libseat maintainer sets up a seat socket owned by user:video | 17:24 |
| nsprra | but user is not added to video group | 17:24 |
| plasma41 | nsprra: I'm not sure I follow you. Are you talking about 'dpkg --extract'? | 18:20 |
| plasma41 | s/dpkg --extract/dpkg-deb --extract/ | 18:23 |
| nsprra | dpkg-deb can do it plasma41 | 19:34 |
| nsprra | but i used to be able to do it with dpkg X | 19:34 |
| fsmithred | ar x | 19:48 |
| nsprra | so wrangled myself out of usrmerge and init mess. made a backup of semi working system. the solution to booting with nvidia + integrated AMD card was adding nvidia-drm.modeset=0 to /etc/default/grub. | 19:50 |
| nsprra | the wayfire (wlroots) problems with seatd are semi-ironed out; wayfire can start and render a screen on amdgpu, but something is still broken getting keyboard input, and it exits back to shell with "broken pipe" after 1 minute | 19:51 |
| nsprra | but mouse worked. kust 'foot' terminal couldn't get keyboard input. | 19:51 |
| nsprra | what on devuan sets-up /var/run/user? this looks like a creation of systemd. And wlroots seems to require /var/run/user | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!