libera/#devuan/ Thursday, 2024-05-23

systemdleteI am seeing unprintable characters in the output of some commandline programs.  E.g., wmctrl -l lists all my windows and displays firefox with three black diamond question marks in front of the word "firefox"02:38
systemdleteI know, it's unicode and all the issues around "wide" characters and yada yada.  Is this going to be cleaned up, or maybe wide characters get implemented in Gnu/Linux, or whatever?02:39
systemdleteI've seen this in the output of other commands also (not many, but...)02:40
systemdlete(can't recall which other ones atm)02:40
systemdletethe main reason I ask is because if I ever need to, say, grep through some program's output, it can be tricky, and stripping out unprintables is might not always be a guarantee of accurate results.02:41
systemdletes/ is //02:41
systemdletewith regard to my experimentation with 3D VMs, it looks like daedalus is not the issue: It just froze the desktop withOUT 3D.  So the problem is something else.02:44
systemdletechimaera has not frozen once, at least up until now.02:45
XenguyI'm running Chimaera now, and I get no such artifacts with that command02:46
onefang_I'm experimenting with apt-cacher-ng today.  What was your basic issue with it again systemdlete?02:46
* onefang_ removes my tail.02:46
fsmithredno black diamonds in daedalus, either. I tried xfce4-terminal and xterm02:47
XenguyMATE Terminal here02:47
systemdleteonefang, clients intermittently fail to get packages (Ign instead of Hit)02:47
gnarfacehmm... systemdlete, the unprintable characters thing seems like either a mis-configuration where you've got the wrong terminal or environment variable settings or else a bug in that build of the program where it's ignoring your environment02:49
onefangFunnily enough I had that problem hit during initial tests, but seems fine now.  My main problem right now is it doesn't get along with apt-listbugs.02:50
gnarfacethe description of the problem also matches potential video ram hardware failure, but when that happens usually the corrupted glyphs seem randomized, and not just isolated to the terminal02:50
systemdleteok, nvm about the unprintables--that seems to be only related to ff windows/tabs running thruk02:51
onefangapt-cacher-ng also doesn't get along with HTTPS, but that's the problem with HTTPS, it's designed to shut out the MITM, but proxies work by being the MITM.  I can live with that for now.02:51
systemdletesorry for the noise about those unprintables...02:52
* systemdlete hits themselves hard on the head for not being more thorough02:52
onefangAt least you are not printing unprintable noise.02:52
systemdlete(might be better if I did sometimes)02:53
onefangAny clues on the apt-cacher-ng / apt-listbugs mismatch?  It's telling me it can't contact the bug tracker server.02:54
systemdleteonefang, I continue to look at the cacher problem, but I've been busy with other stuff here for the past few weeks.02:54
onefangThat's my problem in general, been busy with other stuff since beginning of last year.  lol02:55
systemdleteit's like, every time I try to investigate one problem, it leads me down some rabbit hole or another where I encounter yet another issue, and ...02:55
systemdlete(y'know)02:55
onefangI was about to dive into the problem for the day, but since you where here being unprintable, I figured I'd ask you first for clues.02:56
systemdleteI SOOO want the cacher to work right.02:56
XenguyIs there a simple way to constrain an application from running at 100% CPU ?02:56
XenguyKind of like a 'nice' at the application level?02:57
systemdleteXenguy, that's a problem I've encountered frequently.  Supposedly, limit(1) is supposed to allow you to restrict resources like, say, cpu for processes.02:57
systemdleteBut sadly, I found it didn't really work too well.02:57
systemdleteonefang, am I really all that unprintable?02:58
onefangapt-cacher-ng is at least doing a good job of catching the things that mmdebstrap avoids the local cache for in the early steps.  So that helps with testing my install script over and over..02:58
Xenguysystemdlete, Appreciate the tip, I'll have a look02:58
systemdleteYou are making me feel like The Pentagon Papers, the way you say that...02:58
systemdleteXenguy, let me know if it works for you.  I tried, but for some reason, it just didn't seem to have any effect.02:59
XenguyWill do03:00
systemdletestrangely, Xenguy, I haven't run into that problem in a long time.03:01
Xenguysystemdlete, This new(er) version of qbittorrent seems to be running quite hot, and pushing the CPU to 100%03:08
systemdletewhat happens if you try setting cpu limit?03:09
systemdleteand then running qbittorrent in the same environment (I think the environment has limits set inherited from its parent(s))03:10
systemdletebeen a while since I looked into that one03:10
XenguyI'll let you know when I get around to it03:12
systemdleteoh, ok.  I thought you were trying this right now03:13
systemdletebtw, the cmd line is ulimit03:13
onefanghttps://bugs.debian.org/cgi-bin/bugreport.cgi?bug=775126 seems relevant.  "apt-cacher-ng fails to let apt-listbugs retrieve bugreports when used as proxy".  I'll try the work around in the other bug report linked from that one.03:14
systemdlete(it's actually "prlimit" now?  What?)03:14
onefangSeems SOAP is a bit too slippery.03:15
gnarfaceXenguy: for a bittorent client it may be easier to constrain the CPU load by constraining the concurrent connection count... assuming there's some setting for that03:16
Xenguygnarface, Makes sense, thanks for the idea03:17
rrqonefang: seems apt-listbugs has bugs.debian.org stringly hardcoded in its ruby sources, but youc can have an apt.conf setting like "Acquire::http::Proxy::bugs.debian.org "DIRECT" to avoid it using proxy03:29
rrq.. documented in the ruby script /usr/bin/apt-listbugs03:30
onefangThat's the work around I mentioned, I'm trying it now.03:30
onefangJust don't hold your breath, it takes over half an hour to run this test script.  "That work around works" means "No problems with listbugs reported".  Then I can go down the next unprintable rabbit hole.03:34
onefangapt-cacher-ng is working in general.  "Fetched 17.1 MB in 0s (65.8 MB/s)"  B-)03:38
onefangDamn, didn't help.03:49
onefangAh works fine if I don't override the config file at the end to redirect the proxy from localhost to the qemu host.03:52
systemdleterrq, onefang: just fyi, I am looking at apt-cacher-ng problem atm.  I am running tcpdump (and maybe wireshark later on)04:05
systemdletewhen it failed a few minuts ago, it looks like it was hitting m10k.jp.http04:06
onefangGot an IP for that?  Name doesn't ring a bell.04:07
systemdleteI was just about to ask how you want the hosts, so from now on, I'll give you IPs04:07
systemdlete160.16.137.15604:08
systemdletebtw, I got the list of IPs from "nslookup deb.devuan.org"04:08
onefangdevuan.m10k.jp is the name of that mirror.04:12
onefangapt-panopticon says there's nothing wrong with that mirror as of a few minutes ago.04:13
systemdleteso there is something wrong with me then04:14
systemdleteheheheh04:14
systemdleteI /think/ that was the one... I was just starting to investigate at that moment, and I may have mis-matched the packets I was seeing to the invocation of apt04:14
onefangI'm using my own mirror during today's testing.  At least I can read it's log files if that'll be helpful.04:16
systemdleteI think I will switch over to my cacher system.  bbs04:16
al1r4dhttps://kelar.org/~bandali/blog/pacify.html04:20
al1r4dPulseAudio with changing machine-id04:20
al1r4dfor devuan04:20
systemdletegive me a momemnt to get my wireshark configured...04:21
olivia-mayHey, I can't get devuan Daedalus 5.0.1 to work with ventoy. When it gets to the boot preamble, mount has an error. "mount: mounting LABEL=DEVUAN501 on /cdrom failed: No such file or directory" "mount: mounting UUID= on /cdrom failed: No such file or directory". It tries this 4 times and then starts the emergency shell. I tried this with Debian on Ventoy and theres no issue. I'm using libreboot's SeaBios Payload to load Ventoy on my04:42
olivia-maythinkpad-x230.04:42
onefangI think this was discussed before olivia-may.  Hopefully someone that was paying more attention than me at the time is paying attention now and can help you soon.04:48
systemdleteI've had varying amounts of success with ventoy.  It seems they have to tweak ventoy a bit for each distro and release (i think)04:53
systemdleteI also used another similar tool as ventoy, but don't recall the name atm04:55
systemdleteone of them seemed to work ok for me.04:55
olivia-maysystemdlete: MultiBoot?04:56
onefangapt-listbugs with apt-cacher-ng sorted.  Next rabbit hole!  rrq's suggested work around that I was already trying did the trick.04:59
systemdletethat might have been it.  I tried a few of these.04:59
rrqdevuan preamble expects the ventoy partition to be labelled "Ventoy"04:59
rrqexpects=requires05:00
rrqolivia-may: if that's a hurdle for you there's another hands-on option05:13
olivia-mayrrq: gparted shows the first partition (/dev/sdb1), the partition with the .iso files, as labelled as "Ventoy".05:23
rrqright. the preamble also expects it to be an iso9660 filesystem... and probably then expecting this of the whole ventoy device (/dev/sdb)05:27
olivia-mayrrq: its showing as an exfat filesystem05:28
rrqprobably a bug in the preamble there... so it may need the special hands-on route05:28
olivia-maywhats the hands-on route?05:29
rrq1. at the boot splaces, push TAB to see the boot command line, and add EMERG to it05:30
rrqdoing so will make the preamble start an early "emergency shell"05:30
systemdleteSo I just tested ventoy with devuan daedalus.  The net-install doesn't seem to boot (with the same errors as olivia-may reported), yet chimaera net-install boots ok05:30
systemdleteI can also boot daedalus live install05:30
systemdletemaybe this should be reported to ventoy devs05:32
rrq2. use "sed" to change line 27 of /sbin/unpack to use exfat for partition type rather than iso9660 (for the Ventoy partition)05:32
rrq3. exit the early emergency shell05:33
rrq... if that's still fails, there's also the "excessive hands-on route" to try05:35
olivia-mayoh i forgot to mention, i was using the devuan_daedalus_5.0.1_amd64_desktop.iso.05:40
rrqthat's fine; it's the same install s/w on all ISOs; when it works, the ISO (as ventoy image file gets) is set up at a loop device06:15
systemdletethey are all the same, yet daedalus live install booted but daedalus net-install did not06:23
systemdlete"when it works,..."06:24
systemdlete(maybe I misunderstand)06:24
systemdleteminimal-live booted but netinstall did not (fixing spelling)06:25
systemdleteNot saying you are wrong, but this doesn't comport with "same install s/w"06:25
systemdleteolivia-may, I haven't tried with desktop iso06:26
systemdlete(just fyi)06:26
systemdleteit turns out, btw, that I have been using ventoy, not multiboot (but I did try that one also)06:27
rrqlive is not the installer06:27
rrqall the installer-iso ISOs have "the same" installer s/w06:28
rrqthe main differences are in the on-disk package pools06:28
rrqventoy only works when the booted filesystem is fully in the initrd or it includes special software to find the media as image file06:30
systemdletechimaera netinstall boots06:39
rrqbut the installer will need access to the media for loading modules which will fail06:42
rrqthe daedalus installer(s) also boots (the preamble)06:43
rrqand they also need access to the media (for loading the "actual installer")06:43
rrq(which then again needs access to the media for loading modules... and possibly the package pool)06:44
systemdleteso whatever difference(s) exist, must exist after the preamble?06:53
rrqyes the differences between netinstall, server and desktop are in the iso pool content (and identifications); the installer s/w that is packed up into a subsequent initrd is the same07:06
rrqand the preamble (initrd) is the same07:07
olivia-mayrrq: so what you're saying, is that the installer doesn't work with Ventoy?07:23
rrqright; the "actual installer" requires the media be mountable, i.e. be a device or partition. Ventoy only holds it as an image file.07:46
rrqthe preamble solves that be setting up a loop device for the image file07:47
rrqbefore switching to the actual installer07:48
rrqiow it mounts the ventoy partition, finds the media file and sets it up as loop device, then switches to the installer which now can find the media as a device (the loop device)07:54
rrq(in practice it's a little bit more involved since the installer also "thinks" that it is the boot filesystem)07:56
rrqyou'll find the particulars in the /init and /sbin/unpack scripts at the emergency shell filesystem (the preamble)07:58
rrqor you can join Luke at https://git.devuan.org/devuan/installer-iso08:02
s|oO|0o|gHi, not a Devuan specific question, but that's the OS I'm typing this from: is duckduckgo failing to show results for other people than I?09:06
systemdleteI tried searching for "furkin" on ddg a few minutes ago, and I got an empty page09:17
systemdletes|oO|0o|g, yeah, everything I search on ddg comes back bupkas now09:18
s|oO|0o|g"Sorry, we ran into an error displaying these results. Click here to try again."09:19
systemdleteyep, yep09:19
systemdletesame here09:19
s|oO|0o|gok ty, I'm not too insane yet09:19
s|oO|0o|gsame on my Lineage mobile phone (though this one fails to get results from startpage.com as well, wth is going on)09:21
systemdleteisitdownrightnow.com says ddg is up09:21
s|oO|0o|gwell, the site is technically up, just functionally useless09:22
systemdleteseems like the whole net is slow09:25
systemdleteor the web anyway09:26
systemdletestartpage.com is working for me09:33
systemdleteand I can get to duckduckgo.com, but the searches are useless09:34
s|oO|0o|gI asked elswhere, bing seems down too, doesn't ddg use bing?09:39
ted-iousI believe microsoft invested in duckduckgo before they stopped pretending that they didn't modify search results so it seems reasonable that bing and ddg sites are related somehow.09:44
ted-iousOther people have mentioned bing problems elsewhere recently.09:45
gnarfaceduckduckgo is powered by bing, they've admitted that in an interview before at some point09:49
systemdleteWhat about startpage?  or maybe others?09:50
gnarfaceno idea09:50
s|oO|0o|gstartpage is ok AFAIK, they source from google09:51
s|oO|0o|gmaybe their images are from bing09:51
ted-iousStartpage was purchased by an advertising firm several years ago.09:52
s|oO|0o|gstartpage image search fails09:52
s|oO|0o|goh?09:52
s|oO|0o|gted-ious: right, thanks for the heads-up09:58
s|oO|0o|g"System1 operates an industry-leading responsive acquisition marketing platform (RAMP) powered by AI & machine learning." yay09:59
ted-iousI wish I had better news but knowledge is the beginning of progress.10:00
systemdletegnarface,rrq,onefang: https://dpaste.com/DUEL9RXEB (sorry, debian paste thinks I am spamming it12:17
systemdleteIt looks to me--and I am sort of an amateur with this--that the request comes in to the cacher, responds with a 416, but then returns the info anyway in a later response.  Notice the long line of dots at the end... unprintables?12:18
systemdletethe timeframe here matches an apt failure on the client side.12:20
systemdleteclient is mx-linux, not sure what version12:20
systemdleteI got this dump from wireshark by following the HTTP sequence12:21
* systemdlete still wondering what happened to ddg...12:22
systemdletethose dates seem to be all over the place!  some are GMT, but some don't indicate any TZ, and seem to be local time (9:33 would be 7 hours ahead of us on west coast USA, so that matches)12:33
gordonDrogonsystemdlete, it seems there is a big outage at Microsoft - which ddg, etc.  uses...12:44
systemdletethanks gordonDrogon13:07
systemdleteone thing I notice is that mx-linux wants to update chrome from google's repos.  I don't see any links or references to google in the apt-cacher-ng configuration, but maybe I am missing it13:08
systemdletemaybe I have to manually create it?13:09
gordonDrogonhttps://www.bleepingcomputer.com/news/microsoft/microsoft-outage-affects-bing-copilot-duckduckgo-and-chatgpt-internet-search/13:09
gordonDrogonI don't use chrome, so can' help there...13:09
systemdletegordonDrogon, that was aside from ddg outage13:10
systemdlete(I'm sorting out a problem with the apt-cacher)13:10
Unit193Yes, today I learned DDG is just a theme for Bing.13:10
gordonDrogonsure. from what I recall when I did use chrome was that the package was just a stub that then got the full thing directly from google though...13:10
systemdletemx's apt config has chrome hardcoded for dl.google.com13:13
rrqsystemdlete: apt-cacher-ng creates cache areas for all repository hosts it is asked about; no manual intervention needed13:13
systemdleteThere is no cache area for google/chrome13:14
systemdleterrq:  For openwrt, i followed instructions on their wiki and added it to the cacher--it works.  But I had to manually set up for openwrt.13:15
rrqmmm sounds like apt-cacher-ng doesn't have write access to it cache root dir13:16
systemdleteI see it create caches for kali and gentoo (not sure why; maybe I was experimenting with them, idr now)13:17
gnarfaceafaik if you don't register domains in the config it just caches to the cache root dir and they can theoretically clobber each other13:17
gnarfacei think that's how it was working anyway...13:18
systemdleteI don't see any directoreis or files that look like "strays"13:18
gnarfacehmmm13:18
systemdleteone thing I note--and it is probably harmless--is that some of the links in /etc/apt-cacher-ng go to directories under /var/lib/apt-cacher-ng and others go to /usr/lib/apt-cacher-ng.13:20
rrqit's the CacheDir setting in /etc/apt-cacher-ng/acng.conf13:20
systemdleteah, there I see stuff that looks familiar13:21
systemdletebut nothing for google... not that it matters I guess13:22
rrqthen further down may be some "Remap" rules (I comment out those) that "reorganise" the cace a bit13:22
systemdleteit still looks like a blessed mess to me...13:24
systemdleterrq, have you looked at the link I posted above?13:25
rrqwithout remap rules apt-cacher-ng would want to create a directory in its CacheDir named by the host13:25
* rrq looking13:26
systemdleteI had one of the errors we've been discussing at 02:33, and there were some packet sequences at that time in the wireshark pcap I have been gathering13:26
systemdleteI should probably check just before and after that sequence13:27
rrqthe cache gets polluted by *.head files that sometimes cause issues when some download has failed13:27
rrqyour log suggests that apt-cacher recovered(?) .. that the InRelease file eventually was 44069 bytes long13:32
systemdleteand the cacher does not catch all of these errors?13:32
rrqI think there's a timeout setting that it should trust its cache for some 30 seconds13:33
systemdleteI mean, I expect that the cacher would check return codes (or equivalent) and clean up its temp files and such if a download fails.13:34
systemdleteBEFORE those files become part of that trusted cache13:34
systemdleteyou know, 2-phase update sort of thing13:34
systemdleteI'd be surprised if it isn't at least trying to do that13:34
rrqyeah I think there's room for improvement13:35
systemdleteso would you agree that the problem isn't with your servers?13:35
systemdleteor do you still think there might be an issue there (e.g., what creates those sometimes download failures)13:36
rrqthe servers may have hiccups, but sometimes the cacher locks in to keep the bad file13:36
systemdleteeven if the rr servers have hiccups, the cacher should be able to deal with it.  Hey, this *is* the Internet, after all.13:37
systemdleteit sounds like you are familiar with the cacher code?13:37
rrqit's also a problem if a server updates its meta file(s) before content files13:37
systemdletetrue.  But it should update them atomically together, right?13:38
rrqthat triggers caching badness.. though the cacher could probably be improved to recover better13:38
systemdleteok, so there are problems at both ends13:38
rrqservers should add content files first, then upgrade the meta files13:38
rrqmost use plain rsync though13:39
rrqI believe pkgmaster itself updates in the good way13:40
systemdlete"servers" -- that could mean devuan's, debians, etc ?13:42
rrqyes both devuan mirrors and debian mirrors13:42
systemdleteand openwrt's and kali's etc also13:43
systemdletemx ubuntu yada yada yada13:43
systemdletethere are too many external pieces to try to address the problem that way.13:43
systemdleteI think the cacher side should get attention first, for now.13:44
rrqand afaik there is no update syncronisation so plenty of race windows to think about13:44
systemdletejust fix it so that it catches those intermittent errors and handles them appropriately13:44
systemdleteI really like the cacher approach because it saves bandwidth and allows me to "prime" the local cache before updating systems, some of which might have to shut down in order to do the updates, and maybe not have net access at that time13:46
systemdletewhich is the case here.13:46
rrqyes it's good. Though apt-cacher-ng (and maybe apt-cacher) as well don't interpret the files at all, and it apt that detects inconsistencies13:48
rrqthe cacher(s) is just an http proxy, and there's no control protocol for apt to tell it that a file is bogus13:48
systemdleteIf it would merely handle meta file and package file updates together, as a database-like transaction, I think that might suffice.13:49
systemdleteWould it need to get into the specific weeds for every distro, release, etc?13:49
systemdleteI really don't know13:50
rrqit certainly would need to understand "Release" files13:50
systemdleteI've never gotten that deep into the package hierarchy architecture etc13:51
rrqor one could superimpose a control protocol via bad range request... which then apt would need to know about and use13:51
rrqbut being http there's no "identitiy connection" between client and proxy13:53
systemdleteso for now, I guess I am going to have to handle it from my end (not my cacher's end) and run the maintenance when a failure occurs.13:58
systemdletewhich is exactly how I've been dealing with this13:59
systemdletebut I'm talking about automating the process a bit13:59
systemdletewhich is a separate topic13:59
onefangThought there was a cron job already?14:00
onefangAh, you mean fix it when it happens.14:00
systemdleteyes14:00
systemdleteok, time to go14:08
systemdletethanks to all who contributed to this... analysis?14:08
systemdletenot sure what to call it14:08
onefangWelcome to the Weird Wild Web.14:10
nomiashould i install zfs-initramfs or zfs-dracut if i am installing from debootstrap?16:55
e3d3fsmithred: can I asked you again for help with refractasnapshot ?19:39
e3d3Because of your previous advice I replaced (on a new Daedalus install) refractasnapshot 10.3.0 with 10.4.0, from temporary enabled Ceres repository.19:41
e3d3Using it ended in closing the window without any message, and without created iso, idem when trying again by using option "re-squash"19:43
e3d3I didn't remove the default installed micro-code because (if I understood you correct last time) this should not be necessary with version 10.4.019:44
e3d3Did I something wrong here ? Should I have downloaded a 10.4 version deb-file to install ?19:45
e3d3fsmithred: never mind. I forgot to use sudo :( .Trying again.19:58
AlexLikeRockdamn it!!! .   Has anyone installed HPLIP on deadalus???  that damn hp file, they only compile it in December.  ¿How is it possible, if I buy a printer in May, I have to wait until December for the damn driver to be able to scan?21:36
AlexLikeRock Help me notice that the package is not updated21:36
AlexLikeRockhttps://developers.hp.com/hp-linux-imaging-and-printing/hplip-user-survey21:37
AlexLikeRockplease team , wrote anyting21:37
ted-iousAlexLikeRock: I would take it back and tell them I'm going to buy a canon instead.21:40
AlexLikeRockyes yes yes !!!21:40
ted-iousThen I would buy a brother or an epson ecotank.21:41
golinuxMaybe this will help? https://dev1galaxy.org/viewtopic.php?id=649221:41
golinuxHP usb printer not recognized by HPLip21:42
ted-iousAny company that can't manage to put an ethernet port in their printer and make it work with generic postscript and sane drivers doesn't deserve your money.21:42
golinuxOh . . . and it's dAEdalus btw21:42
golinuxted-ious: OT is not helpful.21:43
rwpI will never buy anything but a network printer ever again.  I am done with USB drivers.21:43
rwpMeanwhile...  Is it possible to say that this new printer is enough like an old printer to set it up as an older printer?21:44
rwpI had to do that recently and after I was clued in as to what older printer to say that it was then things worked great.21:44
AlexLikeRocki try to SCAN by USB port21:54
AlexLikeRockany one have a blog or tutorial to SCAN  from devuan ? .  maybe found somting new21:54
AlexLikeRockyeah , always the olds printers work better thant a new21:55
ted-iousgolinux: It looks like alex agrees that my comments were helpful.22:18
rwpAlexLikeRock, I was suggesting using an older printer driver definition for the newer printer.  I agree older printers were better but here I was suggesting only using the configuration not an actual older printer.22:30

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