libera/#devuan/ Monday, 2024-11-11

AlverstoneSometimes, when I try to bash complete a non-existent file (basically make a typo), it blocks for a minute or so, becoming completely unresponsive, then suddenly goes back to normal. This really annoys me but there are so many variables involved I can't even make it into a bug report. Anyone experienced something similar?12:37
rktaprobably a some network share or unresponsive mount12:39
fsmithredAlverstone, I see the exact same thing sometimes. Also see 'ls' take a long time to respond if there are a lot of files in the directory. (a lot could be 100, nn 10,000))12:43
fsmithredno/not12:43
fsmithredsame if I open thunar. It takes a long time to show me the files in my home dir.12:44
Alverstone_rkta, nope, it's just a HDD drive. The last few times it happened that I remember were with one of my USB drives. They're were responsive just fine, but still the console would block completely.13:03
Alverstone_fsmithred, any way in which I can reproduce the bug with lots of files?13:03
Alverstone_for ; do touch ; done ??13:03
* Alverstone is annoyed with random internet reconnections!13:04
freem<Alverstone> Sometimes, when I try to bash complete a non-existent file (basically make a typo), it blocks for a minute or so, becoming completely unresponsive, then suddenly goes back to normal. This really annoys me but there are so many variables involved I can't even make it into a bug report. Anyone experienced something similar?13:05
freemI have that as well (except for zsh), especially with my slow HDD or with git. It can be cancelled with ctrl+c though.13:05
djph^13:06
djphI've seen it with older/slower computers where it takes forever to try to figure out what it is I meant ...13:06
Alverstonefreem, I wish. In my case it ignores all input. At first I thought I pause it accidentally (you know, ^S and ^Q sequences), but nope. Not even ^C works13:07
freem(and that HDD is slow because hardware is aging and didn't enjoyed all the travels. I should just migrate data, but I say that since 2 or 3 years already...)13:07
AlverstoneMy drive isn't slow though. It's only a few years old13:07
AlverstoneIt can write its legitimate 100-200 mb/s just fine.13:08
Alverstone5600 rmp iirc13:08
Alverstonerpm13:08
freemI suspect my drive to have damaged sectors, actually. Which the kernel does not exactly enjoys, so that makes bunches of failed I/Os, which can't help with speed13:08
Alverstonefreem, smartctl -A ?13:08
Alverstonetouch {1..10000}; ls13:09
Alverstoneworks fine to me13:09
freemhttps://p.mort.coffee/g2H13:10
freemyeah, I guess it's on the end life13:10
Alverstonelooks fine though?13:11
freemreally? The old-age and other stuff are not problems?13:11
Alverstonesee smartctl manpage about -A13:12
AlverstoneEach Attribute also has a Threshold value (whose range is 0 to 255) which is printed under13:12
Alverstone              the heading "THRESH".  If the Normalized value is less than  or  equal  to  the  Threshold13:12
Alverstone              value,  then  the Attribute is said to have failed.  If the Attribute is a pre-failure At‐13:12
Alverstone              tribute, then disk failure is imminent.13:12
freemI also have this kind of files which spawn in my /tmp: https://p.mort.coffee/awv13:12
Alverstonefreem, run an extended self test13:13
freemhow? I have doing that stuff on TODO-list since years, but never actually put enough time to do more than getting smartd running (which is useless since I don't understand it's results... but is a reminder that I need to dig)13:14
Alverstonesmartctl -t long /dev/sdX13:15
Alverstonewhen it completes, `smartctl -a /dev/sdX`13:15
AlverstoneAlso, see `smartctl -c /dev/sdX`, under offline data collection capabilities, it should say 'Suspend Offline collection upon new command.'13:16
freemthat's going to take litterally 8H according to the command...13:16
AlverstoneI never measured to be honest, but yes, usually several hours.13:16
freem"Please wait 483 minutes for test to complete." guess I should ensure my tea/beer storage is not empty...13:17
AlverstoneThis test is the only reliable way to see what's up with your disk though. I don't do it often, but nonetheless do it just to prepare my ass in case of an imminent error.13:17
freemsounds like a good idea13:19
AlverstoneI think all modern disks allow suspension of the test when a new command arrives. But if it doesn't, then you must not invoke smartctl until the test completes, and how you know about its completion is... a mystery :)13:19
AlverstoneAlso to be honest I don't really know why I run smartd either :) I guess just in case or whatever13:21
AlverstoneAnd finally, `smartctl -c /dev/sdX` IS the way to monitor the progress of the test13:21
freemwhat's certain is that I don't see any disk activity13:21
freemah13:21
Alverstoneahh13:22
Alverstone'-c' to check if it's still running. If it is, leave it alone. It seems to idle one second and throttles you IO completely the next13:22
freemkind of looks like this stuff have no actual end... :13:23
freemrecommended polling time:  (   2) minutes.13:23
freemExtended self-test routine13:23
freemrecommended polling time:  ( 483) minutes.13:23
AlverstoneWhat do you quote?13:23
freemthe "xtended self-test routine" is placed as if in the case you continue the test longer, it improves13:23
freemthe output of smartctl -c13:24
freemthat's just guessing though13:24
freemin 8H... I'll probably *really* want to sleep so I don't think I'll try :)13:24
AlverstoneI don't actually know what it means. I only know that self tests are not infinite and that they can be aborted using '-X' option.13:26
AlverstoneAnyway it requires a bit of patience, I usually run these tests before going to sleep :)13:27
morenonaturalhey, y'all ... probably a long shot, but... how should I monitor a specific gamepad being connected/disconnected? currently doing `ls /dev/input/js0` on a 1s sleep loop, `inotifywait` didn't work13:29
freemI would probably do something akin to "dmesg -w | grep /dev/input/js0"13:30
fsmithredtail -f /var/log/syslog13:30
* Alverstone sighs13:33
Alverstonegrep Xorg log13:33
fsmithredmaybe some way with udev rules.13:33
Alverstoneor go wild as a true linuxoid and abuse udev. You can even abuse libudev to do it as a regular user, I remember writing something ugly like that a while ago for fun:)13:34
morenonaturalthat scaled quickly13:34
AlverstoneIf you find some python/go bindings it isn't even going to be that complex. 50-100 lines at worst13:35
morenonaturalin all fairness, I'm probably gonna use such script until this week... the replacement PCB for said gamepad arrives this week13:35
AlverstoneIf you hardcode gamepad info instead of implementing configs it's basically trivial13:35
AlverstoneBut honestly, write a udev rule and abuse notify-send13:35
morenonatural👍🏾13:36
Alverstonemorenonatural, also, `man udevadm`, see monitor section. I bet it's trivial to write a wrapper shell script for that, gonna be by far the easiest way to go13:37
AlverstoneMonsoonSecrecy, look, I made a 10-minutes masterpiece for you https://paste.debian.net/plain/133517914:12
Alverstone:D14:13
Alverstonemorenonatural,14:13
AlverstoneMonsoonSecrecy, sorry, accidentally tagged you instead of morenonatural14:13
morenonaturalAlverstone, ohmysystemd thank you! sorry to trigger that OCD, too14:15
Alverstonemorenonatural, can't help that :P I'd even go as far as to say it's good it got triggered for something simple like this, sometimes it derail my brain for several days14:16
Alverstonecan derail*14:16
morenonaturalwell, that's what's so interesting to me about Emacs: you can always re-charge energies by going into a fine-tuning void for several hours before going back to mundane repetitive work14:18
AlexLikeRockwhats alternativo to :  gksu ,  kdesudo , pkexec17:07
debdogsu-to-root [-X]17:10
debdogsu-to-root - A simple script to give an `interactive' front-end to su.  It can be used in menu entry commands to ask for the root password17:10
debdog-X     The command is a X11 program that does not require a terminal.  This is to be used with menu entries that declare needs="X11".17:10
AlexLikeRockwhere can find ?17:15
AlexLikeRock its not at synaptic17:15
debdogpackage "menu"17:18
n4dirit is part of "menu", so you probably have it installed17:18
AlexLikeRockheheh17:18
AlexLikeRock yes, its install it !17:18
AlexLikeRockwhere are the path ?17:19
n4dirin case it comforts you: i never heard of su-to-root myself17:19
AlexLikeRockyes17:19
AlexLikeRockinever heard of su-to-root  my self17:19
AlexLikeRockXD17:19
* debdog created and maintains his own menu so he's required to know17:20
n4diri got no display-manager like gdm running, so i think i usually can just type su in a terminal and start gui apps from there.17:20
n4dirseldom have to, so not sure17:20
n4dirwell: am pretty sure17:21
AlexLikeRock¬_¬17:22
AlexLikeRockusage: /usr/sbin/su-to-root [-X] [-p <user>] -c <command>17:31
AlexLikeRockyes ,17:31
AlexLikeRockits working  , by terminal17:31
AlexLikeRockthanks debdog17:31
debdognw17:31
Alverstone>gksu, kdesudo, pkexec19:39
* Alverstone is acutely pained19:39
AlverstoneHonestly just use doas or `su -`19:40
AlverstoneBy the way19:40
Alverstoneis making cgroup tasks file world readable a good idea? It's just a CPU cgroup to set lower priority to processes. `nice -n 19` isn't enough for me19:41
Alverstoneworld writable*19:42
AlexLikeRockhoe to reboot with hardward too ?20:56
AlexLikeRockinit 6  : just restart the kernel20:57
AlexLikeRock, i need shutdown and boot again20:57
fsmithredalt-sysrq-R... S... U... B20:57
AlexLikeRockhahahhaha20:58
AlexLikeRockyou forget  "I"20:58
AlexLikeRockand  "E"20:58
fsmithredE and I do nothing20:58
fsmithredand what I typed might not work with the default settings - I'm not sure20:59
fsmithredin /etc/sysctl.conf20:59
fsmithredreplace B with O for powerOff21:00
FatPhilshutdown -r will reboot21:01
AlexLikeRockok21:01
rwpIf you are on the hardware vt console with a keyboard then Control-Alt-Delete will trigger a graceful "shutdown -r now" action.21:05
Alverstonesysrq + B works like a breeze21:05
rwpOn a hardware X console Control-Alt-F1 will switch to the hardware vt console and then Control-Alt-Delete will trigger the reboot.21:06
AlexLikeRocknot work reboot -r21:06
rwpAnd then if that fails to work then using SysRq is the best way.21:06
AlexLikeRockmy laptop not have  "SysRq"21:06
AlexLikeRocksorry21:06
rwpWhat would "-r" do with "reboot" on a GNU/Linux system?  That's not a valid option here.21:06
Alverstonealt+print screen21:06
Alverstoneis the sysrq combination21:06
AlexLikeRock they have , but  take a screen shot21:06
Alverstonealt + print screen + B21:07
Alverstoneyou'll like it21:07
rrqyou might try: echo b > /proc/sysrq-trigger21:07
AlexLikeRockhahahah21:07
rwpI often need to look it up on random keyboards but I think it is Alt+Fn+PrtSc on my current keyboard.21:07
AlexLikeRockthanks  i like so much  rrq21:07
rwpIf the system is running enough to do those commands then it is probably running enough to run "shutdown -r now" then too.21:07
AlverstoneI wish I could break my system like that21:09
AlexLikeRockwhat stop to you?21:09
AlverstoneLately whatever I do it still works somehow and the worst part it works more or less like I intended21:09
AlexLikeRock do it!21:09
AlverstoneKinda takes all the fun from tinkering21:10
AlexLikeRockhhahha21:10
AlverstoneI am actually on the verge of encrypting my drive, but somehow I still have enough sanity to resist it21:17
AlverstoneOn the other hand, if I understand it correctly, with su disabled, when the root is encrypted, I can disable password for some users (I mean the agetty's feature to log you in without requiring password), since it doesn't add any meaningful protection anymore21:18
AlverstoneThough if somebody has physical access...21:19
AlverstoneI just can't get over having to type one more password at boot21:19
rrqdoesn't encryption bring a "single-byte failure" problem for the whole ?21:21
Alverstone:)21:22
AlverstoneLUKS header can actually be backed up. But in general encryption is a lot of things too keep in mind and maintain and I do not have any data to protect against theft. Critical passwords are encrypted already.21:25
AlverstoneHe finally managed to reboot, ehh21:25
AlverstoneWhich password encryption do you use? Bookworm had sha512 with default amount of rounds (or it is leftovers from buster or bullseye), but now I see it has changed to YESCRYPT. Is YESCRYPT better?21:26
djphAlverstone: 'yes'.  it's even in the name :P21:27
djphAs with all cryptography options, it's likely better in some respects, parity in others, and worse in others.21:28
* rrq keeps protected stuff in a gocryptfs partition that I unshare mount21:28
Alverstonehttps://github.com/openwall/yescrypt sounds good to me21:31
Alverstonerrq, I use gocryptfs too, but instead with pam_mount, because I don't want to enable allow_others in fuse.conf21:31
Alverstonegocryptfs is a very handy thing21:31
rrqyeah I do unshare mount to disallow others :)21:32
Alverstoneunshare mount ? :)21:32
rrqwhile accessing through a single session shell21:32
Alverstoneyou have an entire partition for gocryptfs?21:33
rrqmmm actually I changed it to be a directory ... mounting gocryptfs within "unshare -m"21:35
rrqthat makes the gocryptfs mount unavaliable to other sessions21:35
rrq(root obviously may do "nsenter")21:37
AlverstoneBut what is the point? Uh, can you describe the bigger picture? Mine is simple, I do pam_mount with gocryptfs and then simply stuff whatever I want to encrypt in there as a regular user. Easy, handy, whatever. Another benefit is that it's trivial to make a shell script to notify you if mounting has failed21:39
rrqby unsure I lock out all others from peeping into the decrypted tree, including same user in other sessions. So it's "stronger" lock-out.. not actually needed from a practial perspective, but it ended up like that while I learnt about unsharing.21:42
rrq"unsure" => "unshare"21:42
rrqI stepped away from partition encryption so as to the avoid "single-byte error" problem for my "secret" data.21:44
rrqotoh I need to keep some junk in the files to make them large enough to hide the data with encryption21:47
Alverstonesize shouldn't matter?21:56
AlverstoneAlso, `unshare -m` requires root privileges!21:57
rrqyes unsure requires root. Not knowing the datum size makes it a harder nut to crack.21:58
Xenguy_I'm unfamiliar with gocryptfs; does it resemble an encryption scheme like 'veracrypt' at all?21:58
Xenguy_Meanwhile I've been using KeePassXC for encrypting passwords, and 'ccrypt' for one-off file encryption operations21:59
AlverstoneIIRC the file size isn't obscured so if the have the plain text file they can with high probability say that you have it too, but even if they know the size of a file with passwords, even if they know the syntax, it would be pretty damn hard to brute force it.22:00
Xenguy_(I also use LUKS at boot time)22:00
rrqgocryptfs does directory tree encryption by mapping to a decrypted tree, preserving the directory structure whilst scramblilng the names as well as file contents.22:01
rrqthe decrypted tree is a fuse filesystemm22:02
Alverstone>KeePassXC22:03
Xenguy_rrq, Putting it on the todo list to look into further22:04
AlverstoneI symlinked my clipboard manager's directory with the history file to the gocryptfs mount point and don't care about anything anymore22:04
rrqright. I only use my vault transiently.22:05
Alverstonerrq, ?? mount, write, unmount?22:05
AlverstoneThere is just no point in using stuff like KeePassXC when there is such a beautiful thing as gocryptfs :)22:06
Xenguy_Sounds like a better mousetrap, hrm22:07
rrqyes; mostly mount+read+unmount .. scripted of course22:07
rrqso "read" is basically starting a shell to do whatever (aka run emacs :)22:08
rrq(emacs-nox)22:08
Alverstoneemacs :( gimme a break22:08
* Xenguy_ solved one religious war by running both Vim and Spacemacs : -)22:09
Xenguy_I've come to appreciate emacs fans more since I took up using Spacemacs22:09
rrqif I could learn and remember how to make and run keyboard macros in vi or whatever, I could probably use that/those too22:10
Xenguy_rrq, I use Vim macros all the time at work.  I usually just press 'qq' to record a macro, then '@q' to replay it (assuming that's what you are talking about)22:11
rrqmmm sounds easy enough... now on my reading list; thanks22:12
Xenguy_my pleasure22:12
AlverstoneThe problem with vim is that while it seems easy, its documentation is a black hole in which you can effortlessly waste a lifetime22:19
Alverstoneemacs is pretty much the same in this regard22:19
Xenguy_Alverstone, I think of it as the 'beautiful mystery of Vim':  the idea that the editor is so vast that one person could never learn it completely in a lifetime is fascinating to me...22:21
AlverstoneIt's what happens when software is developed for a lifetime of an entire team of developers22:21
rrqI don't really mind the tool; just that I'm most efficient using emacs.. been using it since 198022:22
Alverstone>198022:22
AlverstoneNow isn't you gray beard heavy these days?22:22
Alverstoneyour*22:22
Xenguy_Because Vim is so vast, it almost guarantees those 'mystery moments' when you discover a new feature and think 'why did I only learn about this amazing feature now?'  : -)22:23
rrqmy first terminal editor.. teco emacs on mips-20.. the step after shuffling cards22:23
djphrrq: great OS, all it needs now is a goo editor :P22:23
djph*good22:23
Xenguy_rhymes with pico22:23
djph... er, oops, this isn't offtopic :(22:24
Xenguy_Oh right22:25
Xenguy_Well, on the topic of text editors, pico was my first in a *nix environment22:26
Xenguy_Nowadays that's nano I suppose22:27
rrqAlverstone: some leasure time reading .. https://en.wikipedia.org/wiki/TECO_(text_editor)22:33
rrqah; it was TOPS-2022:34
AlverstoneI can't believe people actually wrote something in TECO22:37
AlverstoneI don't have capacity to take recreational computing to this level of magnificence22:39
rrqyeah, today we look for other conveniences than "saying a lot with few characters"22:55
FatPhilAPL was pretty whack.22:57
rrqindeed .. and I looking for a way to have my aarch64 h/w boot an aarch32 kernel.22:58
onefangWell off topic, but APL was my first language, and I have programmed in TECO, coz that's what the system I took over was written in.23:37

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!