* kreyren just came to complain with his script https://i.imgur.com/LkfYUH1.png blaming devuan for making lsb_release -cs to return 'n/a' and not providing a docker image | 02:23 | |
kreyren | :C | 02:23 |
---|---|---|
buZz | :) | 02:24 |
buZz | buzz@bz-x230:~$ lsb_release -cs | 02:24 |
buZz | beowulf | 02:24 |
kreyren | this is on chimaera | 02:25 |
buZz | i'm sure that'll be fixed once its somewhat targetting users ;) | 02:25 |
kreyren | kreyren@dreamon:~$ cat /etc/os-release | grep ^ID\= | 02:25 |
kreyren | ID=devuan | 02:25 |
kreyren | kreyren@dreamon:~$ lsb_release -cs | 02:25 |
kreyren | n/a | 02:25 |
kreyren | buZz, ik but that ain't preventing me from complaining! :p | 02:26 |
buZz | hehe | 02:27 |
buZz | well you could improve it ;) | 02:27 |
kreyren | and like i can make the dockerimage it's just running that debian thing that generates the userland and adding it as `ADD path/to/archive` >.> | 02:27 |
kreyren | but that won't make it to work as `FROM devuan/chimaera` unless devuan publishes it~ | 02:27 |
kreyren | Or like can i join devuan team to be able to fix these things | 02:31 |
buZz | you totally can | 02:32 |
kreyren | how | 02:32 |
buZz | there's #devuan-dev | 02:32 |
buZz | join there and ask | 02:32 |
kreyren | k will do tomorrow bcs i can barelly see now from exhaustion~ | 02:32 |
buZz | np :) welcome | 02:32 |
* kreyren goes to sleep | 02:32 | |
buZz | nn | 02:33 |
conifer | hi! what is the recommended way of setting up /opt partition? i always had it sitting unused on a tiny root, but it turned out electron apps tend to get installed there | 02:35 |
buZz | i dont know, i dont use a /opt or electron apps | 02:36 |
conifer | and each is basically an extended chromium, so they take 200-300MB each and i ended up mounting /opt on /usr | 02:36 |
conifer | but it feels like a makeshift solution and i don't know what is the proper way to handle it when partitioning the drive | 02:40 |
golinux | You can't just install in /home? | 02:40 |
golinux | I have browers there that are quite happy | 02:41 |
conifer | can i pass some parameter to make them install where i want them instead of their default location? | 02:43 |
conifer | ok, i read the dpkg man so i learned how to do it with standalone .deb file, but what if i add a repository to sources.list and install with apt? | 02:53 |
conifer | btw, is installing in /home better than in /usr? | 02:54 |
djph | FSVO "better" | 02:56 |
buZz | imho electron is the makeshift solution | 02:58 |
buZz | 'one day i'll learn to code but today i'll just wrap this website up with a whole browser!' | 02:58 |
buZz | 'LOOK MA, I'M MAKIN APPS!' | 02:58 |
conifer | yeah, i know | 02:58 |
conifer | but i wanted to try out some new stuff, and when their only destkop clients are electron then i have no other choice | 03:00 |
conifer | anyway, what do people usually do with /opt on desktop systems? leave it on root and not use it? put it on separate partition? put in on /usr? | 03:25 |
conifer | before i started using linux i read some articles about partitioning the drive and if they mentioned opt they claimed it probably won't be used at all, so i never cared about it and had no trouble with it | 03:27 |
fsmithred | separate /home and everything else in the root partition is very common | 03:30 |
fsmithred | separate /boot if you use encryption | 03:31 |
fsmithred | sometime separate /var | 03:32 |
conifer | i usually do separate boot, home, usr, var, tmp, so i was really surprised when all i needed to clutter up root anyway was installing an app... | 03:42 |
yeti | with many partitions the free space always is where I dont need it... | 03:50 |
roo^y | 48hrs after updating a new install of Beowulf with it's packages, I move mouse to wake screen from turning it off with xset dpms force off. System appears to have crashed, as screen is black & only showing a underscore '_' I'd quit firefox beforehand, to ease resources while at idle ..maybe I should find out which GPU driver it's using, & try the other one | 11:22 |
roo^y | Keyboard couldn't even light up Caps lock & Num lock lights again, like it was unplugged ..unless it's a CPU crash | 11:29 |
gnarface | roo^y: that would be a good point to attempt to connect to it over the network by ssh, or at least ping it | 12:07 |
roo^y | yep, I happen to have a identical spare one, I should learn how to do that! | 12:25 |
gnarface | roo^y: it's actually really easy | 13:45 |
gnarface | roo^y: "apt-get install openssh-server" ahead of time | 13:46 |
roo^y | sweet :) will do | 13:46 |
gnarface | roo^y: then from the other machine you can connect with "ssh user@hostname" (it will prompt you for the remote user's password by default, you can increase security later) | 13:47 |
roo^y | i understand | 13:47 |
gnarface | if the user name is the same, you don't have to specify the user@ part | 13:50 |
gnarface | and obviously if you don't have hostnames you can use ip addresses instead | 13:50 |
roo^y | install openssh-server on other machine too? ..or no need, just use terminal? | 13:50 |
gnarface | the client server only needs openssh-client, which will not include openssh-server | 13:51 |
gnarface | that was confusing, sorry, the client machine i mean | 13:51 |
roo^y | got it | 13:51 |
gnarface | any machine can have both of course safely, if you know what you're doing | 13:52 |
ham5urg | Is there any need to control the chain of blocksizes of HD->Partitions->RAID->LVM? | 14:08 |
ham5urg | HD->Partitions->RAID->LVM->FS | 14:09 |
gnarface | it might matter | 14:10 |
ham5urg | How do I show/check all these different blocksizes? | 14:11 |
ham5urg | As far as HDs and partitions I guess I can't configure anything as there are no blocks. | 14:14 |
ham5urg | Only sectors of 512byte. | 14:16 |
ham5urg | but fdisks "I/O size (minimum/optimal):" is sometimes 4096, sometimes 512 (HD, SSD). So I guess this should be the blocksize. | 14:19 |
gnarface | check with hdparm -I? | 14:19 |
ham5urg | lsblk -o NAME,PHY-SeC | 14:22 |
gnarface | you're supposed to set the stripe sizes to some multiple of the physical block sizes and the filesystem block sizes but i don't know the exact math to get an optimal size | 14:36 |
gnarface | i'm sure you could look it up somewhere... | 14:37 |
gnarface | also note that some drives have different logical and physical block sizes | 14:39 |
gnarface | sometimes it matters | 14:39 |
gnarface | sometimes it might be raid controller specific | 14:56 |
ham5urg | trying to use mdraid. just ripped off some HP-raid-controller. | 15:00 |
ham5urg | too unflexible and unreliable these controllers. I'll go with a devuan-raid-box | 15:01 |
ham5urg | devuan-mdraid-box | 15:01 |
ham5urg | hd->partition->mdraid->logical-volume are all 512 in my case. ext4 via tune2fs shows 4096. I assume that's fair as it is a multiple of 8. | 15:04 |
gnarface | yea maybe | 15:29 |
gnarface | i was thinking more like 64k instead of 4k but maybe that would be for hardware raid | 15:29 |
gnarface | i really don't know though | 15:29 |
ham5urg | you think a ext4 blocksize of 64k instead of 4k will boost the performance (even if some space is lost due to files <64k)? | 15:46 |
ham5urg | worth a try | 15:46 |
gnarface | ham5urg: no, not of the filesystem... of the raid stripes | 15:48 |
gnarface | ham5urg: leave the filesystem block size to match the physical block size of the disk | 15:49 |
ham5urg | so down to 512byte instead of 4k (in my case). Here's a screenshot of a VM where I test things. https://pasteboard.co/Jjici4f.png | 15:56 |
gnarface | that won't do me any good and i'm not clicking on it anyway | 15:58 |
gnarface | i know you can get the physical and logical size of the disk itself if you run hdparm -I on it | 15:59 |
gnarface | but that would have to be in the host system | 15:59 |
gnarface | in a VM all bets are off | 15:59 |
ham5urg | I understand | 15:59 |
ham5urg | hdparm -I /dev/nvme0n1 | 16:01 |
ham5urg | HDIO_DRIVE_CMD(identify) failed: Inappropriate ioctl for device | 16:01 |
ham5urg | I have to look why is that | 16:01 |
gnarface | try it just on /dev/nvme0 | 16:01 |
gnarface | also you have to run it as root | 16:01 |
gnarface | everything used to be 512 bytes physical and logical | 16:02 |
ham5urg | the same problem occurs | 16:02 |
gnarface | lately 1k and 4k are becoming more common | 16:02 |
ham5urg | makes sense | 16:02 |
gnarface | nvme is something new so maybe hdparm doesn't work on it, i guess i was expecting a sata drive | 16:02 |
ham5urg | lsblk -o NAME,PHY-SeC | 16:03 |
ham5urg | looks good | 16:03 |
gnarface | i've seen some drives lately with 4k physical sectors but 512b logical sectors | 16:03 |
gnarface | which is super annoying because then you have to make a choice between optimal space usage and optimal speed | 16:04 |
gnarface | ham5urg: after checking the man page for lsblk, i see that you actually want to look at: "lsblk -o NAME,PHY-SEC,LOG-SEC" | 16:06 |
gnarface | if LOG-SEC is smaller then you can get a bit more space from using that for the filesystem size instead, but there will be some extra overhead | 16:07 |
ham5urg | I understand | 16:45 |
ham5urg | nvme0n1 512 512 | 16:46 |
ham5urg | Why uses mkfs.ext4 4kb blocksize and not the blocksize of the device? E.g. 512b. Maybe not much space-loss but increase of speed. | 16:50 |
gnarface | i didn't advise using 4k block sizes with that disk, i don't think that's right, not sure though... | 16:51 |
gnarface | but i was just trying to warn you about stuff like this, look: https://paste.debian.net/1157768/ | 16:53 |
gnarface | for the raid stripes though i believe it would be different, something higher, mabye between 8k and 64k | 16:54 |
ham5urg | Uhh, I will avoid such disks if possible. | 16:54 |
ham5urg | That's more complex stuff than i knew. Thanks for your infos | 16:55 |
ham5urg | Found this, now I understand: https://www.linuxpromagazine.com/Issues/2009/108/RAID-Performance | 17:23 |
ham5urg | Why so many examples do use a chunk size of 64? | 17:25 |
gnarface | ham5urg: i think it's a common default for a lot of popular hardware-accelerated raid controllers | 17:31 |
gnarface | ham5urg: mdraid will probably be less picky but i would expect also slower, so i'd guess you'd want smaller stripe sizes but i'm not sure exactly what | 17:32 |
gnarface | ham5urg: i'm guessing there's still some tradeoff but it won't be as much of a stair-step graph | 17:33 |
gnarface | ham5urg: and you might also want to consider what type of data you'll be storing | 17:34 |
Wafficus | Hi there, I'm running Virtualbox in a Windows 10 host, and Devuan as the guest | 22:00 |
Wafficus | I tried enabling Dvice > Insert Guest Additions CD image | 22:00 |
Wafficus | but how do I enable the actual resolution options? | 22:00 |
Wafficus | I ask because I'm able to to select multiple resolution options under View > Virtual Screen 1 > "Resize to" etc choices | 22:01 |
Wafficus | but when I do so, nothing happens | 22:01 |
gnarface | Wafficus: not sure, maybe try restarting the login manager or xorg whichever is running at that moment | 22:17 |
gnarface | Wafficus: alternately, try setting it before the guest starts | 22:18 |
golinux | I am about to dump VB for qemu | 22:18 |
golinux | Not really relevant except that I've had it with VB "quirks" | 22:18 |
gnarface | i hear a lot about them | 22:19 |
gnarface | the only problem with qemu as i see it is that it's not as easy to learn | 22:19 |
gnarface | it is very flexible though | 22:19 |
Wafficus | got it, had to follow this guide | 22:22 |
Wafficus | https://linuxize.com/post/how-to-install-virtualbox-guest-additions-on-debian-10/ | 22:22 |
Wafficus | I had to mount the CD and then run the related .run file to install it | 22:22 |
Wafficus | then I had to use the Settings > Display > Graphics Controller: 'VMSVGA' option | 22:23 |
Wafficus | now I'm able to resize it no probelm | 22:23 |
gnarface | ah, cool | 22:23 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!