| Besnik_b | Hello! Forwarding in a track in VLC using jack introduced a 1-2 second long noise, I did not hear before in the previous system, but I do now in Devuan. Any idea? | 10:19 |
|---|---|---|
| gry | is this for any ttack. can you check another file | 10:33 |
| gry | any track | 10:33 |
| Besnik_b | gry, yes. Is not related to just one track. | 10:34 |
| Besnik_b | (I asked in #lau too. Linux Audio Group.) | 10:34 |
| gnarface | Besnik_b: which devuan release? | 12:07 |
| Besnik_b | gnarface, Daedalus | 12:12 |
| gnarface | Besnik_b: does it happen in any other media players? | 12:13 |
| Besnik_b | gnarface, I have to check that. | 12:15 |
| Besnik_b | gnarface, I don’t hear it in alsaplayer + jack. Confirmed. | 12:22 |
| gnarface | Besnik_b: does it happen in vlc without jack? | 12:25 |
| Besnik_b | Let me check it. | 12:25 |
| Besnik_b | The noise disappears with pulseaudio | 12:27 |
| gnarface | odd | 12:28 |
| gnarface | seems like an issue with vlc or jack then, but hard to say much more | 12:28 |
| Besnik_b | Idem with my USB Rubix direct without any convertions | 12:29 |
| gnarface | did you mix in any 3rd party repo stuff like deb-multimedia or other devuan release versions or the like? | 12:29 |
| Besnik_b | Since alsaplayer + jack removes it, what makes you point jack? | 12:30 |
| gnarface | because you said vlc and pulseaudio also removes it | 12:30 |
| gnarface | you could try vlc and just bare alsa as a control case... | 12:31 |
| Besnik_b | and alsaplayer and jack does it too :) | 12:31 |
| Besnik_b | Here are the repos I have in my sources.list: | 12:32 |
| gnarface | use paste.debian.net | 12:32 |
| Besnik_b | Here you are: http://paste.debian.net/1300615/ | 12:35 |
| gnarface | Besnik_b: the culprit could be something you got from daedalus-backports, but i'm pretty sure daedalus-contrib and daedalus-non-free don't even exist | 12:37 |
| gnarface | you probably meant "main contrib non-free" at the end of all the other ones | 12:39 |
| gnarface | since daedalus, you probably also want the new one, non-free-firmware to those | 12:39 |
| gnarface | so, like this: deb http://deb.devuan.org/merged/ daedalus main contrib non-free non-free-firmware | 12:40 |
| Besnik_b | gnarface, I could not install vlc with just what vanilla installation had to offer | 12:41 |
| gnarface | i assure you if you do this it will include everything | 12:42 |
| Besnik_b | Thank you! | 12:42 |
| gnarface | you shouldn't have any problems installing vlc in a vanilla install though, that's a sign something else was probably already wrong by then | 12:43 |
| gnarface | maybe just that sources.list though | 12:44 |
| Besnik_b | I dd-ed the image in a USB stick and then installed in a dedicated hard disk | 12:44 |
| Besnik_b | After the installation, without adding or removing anything, I could not installed vlc and other programs I use daily | 12:45 |
| Besnik_b | Then I started to check ways and modified the sources.list, tried and found that the change worked. | 12:46 |
| gnarface | well, it had errors though so it's still suspect | 12:47 |
| yo9fah | hi all | 15:30 |
| yo9fah | I tried to install devuan on a laptop. I downloaded the iso image: devuan_daedalus_5.0.1_amd64_netinstall, but it does not boot. The monitor remains black, no error, nothing. The laptop is a Lenovo B50-80. | 15:33 |
| fsmithred | yo9fah, did you try to boot from dvd or usb, and how did you prepare it? | 16:41 |
| fsmithred | Did you see the boot menu? | 16:41 |
| fsmithred | is it uefi or legacy bios? | 16:42 |
| yo9fah | It works from the DVD, it's ok! My bios is uefi/gpt. The image is written in rufus. | 16:54 |
| yo9fah | I don't see any boot menu? | 16:56 |
| Besnik_b | yo9fah, I’d check the DVD, perhaps in another machine to see if it’s no faulty. And probably try writing the installation image in a USB. It’ much easier with a DVD. I’ve seen warnings about writing a iso in a DVD. | 19:09 |
| golinux | Maybe use dd to create the install media . . . | 19:18 |
| Besnik_b | He mentioned rufus, so he might not have a dd capable machine… | 19:19 |
| golinux | uefi/gpt requires special handling which I am not conversant with | 19:23 |
| fsmithred | If secure boot is enabled, disable it. | 19:31 |
| fsmithred | Maybe try the desktop-live iso. | 19:32 |
| yo9fah | Secure boot is disable. | 19:39 |
| yo9fah | Thank you for your help! I will try again to see if I can solve the problem... | 19:40 |
| yo9fah | I solved! Rufus is stupid. I rewrote the same iso image with Balena Etcher. It works perfectly. | 20:17 |
| yo9fah | Thank you all for your help and advice! ;) | 20:18 |
| Besnik_b | yo9fah, Good! Now that you’re under Devuan have a look at the dd command. Much easier and much better results. një gjë të tillë. | 20:20 |
| Besnik_b | - Rregullimet për Optimizoni Figura përdorin, si parazgjedhje, madhësi dhe cilësi optimale. | 20:20 |
| Besnik_b | - Temat instalohen si duhet për sajte nën plane Biznes dhe Tregtiri | 20:20 |
| Xenguy | yo9fah, Glad to hear it | 20:20 |
| Xenguy | yo9fah, BTW is 'dd' available to you? | 20:20 |
| yo9fah | Yes, it also works with the "dd" command! Thanks! | 20:24 |
| yo9fah | I wonder what Rufus is missing? | 20:25 |
| yo9fah | But anyway, it doesn't matter anymore... | 20:26 |
| Xenguy | My short notes on using 'dd': | 20:31 |
| Xenguy | https://paste.debian.net/hidden/17228600/ | 20:32 |
| Besnik_b | Ouch! Sorry for erroneous paste :( | 20:37 |
| sfox | https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1057843 | 21:38 |
| sfox | Don't update to the latest kernel it has a silent corruption bug. | 21:38 |
| gnarface | whew, thanks for the warning sfox | 21:39 |
| gnarface | not that i was using ext4 to begin with, (because it is low quality) but good to know anyway | 21:39 |
| sfox | your welcome. | 21:40 |
| sfox | For those that aren't so lucky, you have backups in place... right? | 21:45 |
| brocashelm | i've learned it's best to not use any kernel newer than 6.3 | 21:45 |
| sfox | gnarface, what filesystem are you using? | 21:45 |
| brocashelm | in fact, pretty much set just using 5.19 (would have gone back to 5.15, but nvtop needs it) | 21:45 |
| sfox | that you consider not low quality. | 21:45 |
| brocashelm | ffs2! | 21:46 |
| sfox | I haven't heard of it. isn't that a BSD filesystem? | 21:47 |
| gnarface | sfox: xfs | 21:47 |
| brocashelm | yeah, it's default for openbsd | 21:47 |
| gnarface | sfox: (mostly, i still have some machines using reiserfs for legacy reasons) | 21:47 |
| sfox | gnarface, what makes you consider it higher quality then ext4? | 21:47 |
| brocashelm | though hammer from dragonfly sounds interesting and would be cool to try it on linux | 21:47 |
| sfox | In my testing, xfs had problems with large amounts of files on busy webserver | 21:48 |
| sfox | s | 21:48 |
| sfox | plus it's backed by redhat is it not? | 21:48 |
| sfox | the same company who pushed systemd, pulseaudio, and gtk3 | 21:48 |
| sfox | brocashelm, I didn't know Devuan could use ffs2 | 21:49 |
| gnarface | sfox: you had an old version of xfs that didn't set appropriate defaults for the mount options. it should never have been slow. my opinion of ext4's quality is based just on the amount of times bugs like this have cropped up since it has existed, which happens to be several times more than the other two i mentioned. | 21:50 |
| gnarface | sfox: heh, xfs was created by SGI well before RedHat execs were even out of diapers | 21:50 |
| brocashelm | nah, not possible to use, but i use it on openbsd | 21:50 |
| gnarface | sfox: xfs is also the only multi-threaded filesystem, so it's a no-brainer for anyone selling big iron | 21:51 |
| gnarface | sfox: (i thought RedHat was actually pushing ext4 these days though, but i haven't really been paying close attention - but if they're pushing xfs that's why) | 21:52 |
| sfox | gnarface, it's true I was using an old version, but that's because I was forced to use CentOS 6 at the time. But I did not use the defaults and spent a lot of time fine-tuning mount and filesystem parameters and tracing with blktrace | 21:57 |
| sfox | what I ultimately found was the time looking up metadata, not the actual file was causing the webservers to slow down and time out | 21:58 |
| sfox | even with configuring the kernel to use most of it's ram keeping them in memory so they didn't have to be looked up from xfs | 21:58 |
| sfox | switching to ext4 solved this almost no tuning required | 21:58 |
| sfox | gnarface, has this changed? | 21:58 |
| sfox | the amount of files i'm talking about is enough to make rm -r take over a day | 21:59 |
| sfox | spread across multiple block devices. | 21:59 |
| sfox | also, about how often have you experienced bugs like this cropping up with ext4? | 22:00 |
| * blockhead ponders taking a day to rm files, and faints :o | 22:01 | |
| gnarface | sfox: so like, xfs has never caused me trouble once, and in the past 30 years there's been like one data corruption bug in it like this, and in the past 10 ext4 has had 3 and 2 of them didn't miss me. | 22:04 |
| sfox | several decades of multiple company's emails, images of products from several hundred company's online stores including resize and cropped varients, several hundred websites, etc. | 22:04 |
| sfox | that adds up to a lot of tiny files. | 22:04 |
| gnarface | sfox: yes, that's changed, and to my best recollection of the time frame, centos6 may have been the last one with slow xfs defaults | 22:04 |
| sfox | no matter how many SSDs I threw at XFS, it's metadata lookup just slogged. | 22:05 |
| sfox | well thanks for the advice. I'll give xfs a try again | 22:05 |
| sfox | I've since moved onto ZFS, but I still use ext4 in some places where I haven't had a need to change and it's historical. | 22:06 |
| gnarface | sfox: basically the key i remember was there were two mount options for log/journal buffer settings, and the default was unacceptably low for modern hardware, which caused a bottleneck. if you manually set both these values to their maximum setting of 32k the performance problems mostly evaporated. deletes were still slower than with ext4 but not days slower. and reads then are faster on any cpu with more than 2 cores. | 22:06 |
| sfox | like my encrypted laptop. | 22:06 |
| sfox | Would you suggest migrating all my ext4 filesystems to XFS now? | 22:06 |
| gnarface | sfox: but none of that should be needed anymore, xfs defaults should be fine now. | 22:06 |
| gnarface | sfox: eh, no, i would suggest you migrate one then see how you like it. don't get crazy. | 22:07 |
| sfox | gnarface, If XFS is really so much better, why doesn't Devuan ship it by default instead of ext4? | 22:08 |
| gnarface | (er, more than 1 core i mean) | 22:08 |
| gnarface | sfox: NIH syndrome | 22:08 |
| gnarface | well, it's literally not faster in every respect. deletes are still slower, and streaming of large single files is slower (a bit) so in certain performance benchmarks it doesn't look as appealing as ext4 for "what the kids are doing these days" (mostly producing HD content at home for youtube, apparently) | 22:09 |
| gnarface | but for the situation YOU describe - a giant filesystem of billions of little files gathered over decades.... that's what xfs is for | 22:09 |
| sfox | Thankyou for the suggestion. | 22:10 |
| gnarface | no problem. just try it, you might like it now. | 22:11 |
| sfox | Performance is important, but correctness and now tripping up and having silent corruption every couple of years due to carelessness of coding is more important to me and I can give up a few IOPs for. | 22:11 |
| sfox | *not tripping | 22:11 |
| sfox | I wasn't aware that this was a recurring pattern for ext4. Clearly something I need to look into. I started using it because it was the default and i figured some engineer before me probably chose it for a better reason and sticking with the defaults would be better supported. | 22:12 |
| gnarface | nah, it's chosen because it's the community darling, that's all. | 22:13 |
| gnarface | with open source, merit is often determined just by how many volunteers you can wrangle | 22:13 |
| gnarface | (wayland proves that) | 22:14 |
| sfox | gnarface, can you shrink a XFS filesystem yet? | 22:14 |
| gnarface | heh. no :( | 22:14 |
| sfox | pain | 22:14 |
| gnarface | but you can both grow and fsck them online! | 22:14 |
| sfox | gnarface, one of the reasons I switched my servers to FreeBSD is because I was tired of conflicting interests and constantly changing directions the Linux ecosystem has. | 22:16 |
| gnarface | i don't blame you | 22:16 |
| gnarface | we should move this conversation over to #devuan-offtopic | 22:16 |
| rwp | I also say +1 for xfs. Back at my previous $EMPLOYER we used it on file servers with heavy use and didn't have performance problems then. I can't replicate that load now. | 22:46 |
| rwp | Now I use it for the builtin fast fsck behavior. Recovery after a power loss is always fast. | 22:46 |
| rwp | XFS cannot be grown incrementally as efficiently as EXT4. With XFS if one must grow it then double it in size. Or at least use significantly large size increases. | 22:46 |
| rwp | XFS divides each chunk into quads for block allocation. Small growth chunks mean small quads which works against efficient block allocation. | 22:47 |
| golinux | I thought that almost an hour ago | 22:47 |
| golinux | to #devuan-offtopic puleeze | 22:48 |
| rwp | It's only been 30 minutes. Even I am not ALWAYS at the keyboard. :-) | 22:50 |
| _ds_ | I have the impression that this ext4 corruption bug is entirely down to incomplete or inappropriate backporting of patches. | 23:31 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!