|
Rocko Bonaparte posted:How quickly do those Chromebooks boot a contemporary BIOS and OS? I couldn't tell you how long they take to boot UEFI and a stock distro, but I don't think they're appreciably faster than a comparable Intel NUC with an eMMC instead of NVMe. So 30 seconds?
|
# ? Jul 28, 2022 14:43 |
|
|
# ? Jun 8, 2024 15:39 |
|
A couple more things regarding Chrome hardware: I don't know that they make sense if you want to avoid running Chrome OS--buying Chrome hardware, practically speaking, means buying into Google's ecosystem on some level. But I do appreciate that a Linux container installation is one-click away, and it's well integrated enough that it might as well be native, albeit with providing good security/recoverability isolation. The idea that you can purchase a $200 Chromebook for an interested kid and they can go off and learn how to code Python or Ruby or whatever with little effort is a big benefit. Sure you can do that on a Mac, but those are expensive, and I guess you could do the same with Windows and WSL (or Cygwin) but I don't think you can do that on a comparably-priced "Windows S" machine. Or maybe you can--I don't follow that. I was also really attracted to Chrome hardware because I felt that they really thought about how to do secure/attested devices the right way. There's multiple layers of security in place to minimize unknowing compromise and ensure easy recovery should something happen. But at the same time they made openness a requirement, which is why they allow progressively-greater access to the raw hardware for someone with enough inclination, time, (and physical access) to open the machine up. It would've been easier for them to lock the machine down hard and say "forget it".
|
# ? Jul 28, 2022 14:55 |
|
I bought a new chromebook and it errors out when enabling the linux vm/container without any useful message. Pretty drat annoying. Overall it is like using some gnome-like thing where there are even less configuration options. And I complain about that as someone who enjoys gnome3.
|
# ? Jul 28, 2022 15:50 |
|
Rocko Bonaparte posted:How quickly do those Chromebooks boot a contemporary BIOS and OS? For the Dell Chromebook 11-3120 that I mentioned earlier booting XFCE EndeavourOS off its built in eMMC, this is what I see: Power button pressed to bootloader menu: 7 seconds Bootloader menu to login screen: 13 seconds Hitting enter at the login screen to a fully loaded desktop: 10 seconds
|
# ? Jul 28, 2022 18:05 |
|
other people posted:I bought a new chromebook and it errors out when enabling the linux vm/container without any useful message. Pretty drat annoying. Powerwash?
|
# ? Jul 28, 2022 23:30 |
|
buglord posted:scroll speed In my experience this has always been a problem on linux, until the last time I tried Fedora. Fedora Workstation seems to have solved the issue system-wide. I've no idea how
|
# ? Jul 29, 2022 00:42 |
|
ihafarm posted:Powerwash? Yeah, Powerwash is my recommendation. Or if that doesn't work Recovery mode reinstall Chrome OS. Those two things have solve most every problem I've had that wasn't a known issue that started with an OS update.
|
# ? Jul 29, 2022 00:58 |
|
Comatoast posted:In my experience this has always been a problem on linux, until the last time I tried Fedora. Fedora Workstation seems to have solved the issue system-wide. I've no idea how Really? I don't remember it being a thing for a long time, but I've been pretty much 100% OpenSUSE and KDE for several years now. Fake edit: Oh my god, it's been around a decade.
|
# ? Jul 29, 2022 06:42 |
|
CaptainSarcastic posted:Really? I don't remember it being a thing for a long time, but I've been pretty much 100% OpenSUSE and KDE for several years now. I miss opensuse sometimes. Such a good platform, mostly. Shame about the ancient docker tooling tho
|
# ? Jul 29, 2022 17:04 |
|
RFC2324 posted:I miss opensuse sometimes. Such a good platform, mostly. Why wouldn't you pull direct from docker if your gonna use that? Or Podman?
|
# ? Jul 29, 2022 18:03 |
|
Mr. Crow posted:Why wouldn't you pull direct from docker if your gonna use that? Or Podman? Couldn't find a recent version packaged for opensuse that wasn't some rando on the suse version of ppas
|
# ? Jul 29, 2022 18:49 |
|
Does anyone know much about how abrt works? I'm on Centos 7 and I've been chasing the cause behind a couple machine check exceptions that keep popping up during every startup. I think what's happening is that abrt keeps showing me these old problems because the timestamps on the problem directories are always the same. I'm now pretty sure that the original exceptions were caused by me trying to undervolt too much, which I am not anymore. So I go to delete the problems with "abrt-cli rm <dir>" but then the exact same directories come back after a reboot and abrt continues to show them to me. Any idea what's going on? I guess it doesn't really matter because the computer works fine but it would be nice if I could satisfy my OCD by having it stop telling me it encountered a problem several months ago. Edit: Well I figured it out. The errors I got were dumped into /var/crash and abrt parses that directory so that's why they kept coming back. I just needed to delete those too and all is good. Hikaki fucked around with this message at 09:45 on Aug 1, 2022 |
# ? Jul 31, 2022 09:11 |
|
Another BTRFS question. Is it possible to achieve the following configuration? - Two physical drives of different capacity, let's say the small one is 2TB and the big one is 3TB - Two data folders: /important and /unimportant (could also be /data and /data/important if it makes things easier) - The data under /important is replicated in RAID1 on both drives - The data under /unimportant is not replicated (single mode) - Both drives live in a single JBOD-esque filesystem, and I don't have to manually choose how to split the total (A+B) capacity - End goal: I'm free to write either Important or Unimportant data without periodically janitoring the partitions. The Important data will be duplicated, so I can potentially write up to 2TB of Important data, in which case I would have 1TB left for Unimportant data on the big drive. Realistically, the ratio between Important and Unimportant data is gonna be between 1:10 and 1:100, so I won't have a lot of balancing problems. I just want to avoid having to guess partition sizes if possible.
|
# ? Aug 2, 2022 16:22 |
|
When I checked for a similar function it was planned and an experimental buggy mess. It might stop be in that state of even having been abandoned. Or maybe it works now, but certainly was less of a priority then raid5. It would work by setting a file or directory to single mode, similar to the function for compression.
|
# ? Aug 2, 2022 16:44 |
|
NihilCredo posted:Another BTRFS question. Is it possible to achieve the following configuration? No: you within a single btrfs volume you can have subvolumes, but you can't have different mirror/stripe options per subvolume. (Yet, it's a wishlist feature, but who knows when it will happen.) A btrfs raid1 drive pool can use 3 or more unequal size drives to get the maximum capacity that is possible (ie 2TB+3TB=6TB = 10TB capacity), but that's still mirroring on everything. btrfs isn't a terrible choice for your setup because it's relatively fast to resize volumes, much more than ZFS. But you're still gonna be doing everything manually when you want to change the balance of Important and Unimportant.
|
# ? Aug 2, 2022 18:27 |
Wait, btrfs doesn't have an equivalent of ditto blocks? On zfs, you set copies=2|3 which tells zfs to write the same block multiple times on the vdev that the dataset is on*. I believe this function is essentially reusing the exact same code for metadata, which also gets written multiple times to disks. *: It's not an even distribution, hence it's not a replacement for the actual mirroring or striped data with distributed parity - the primary use-case is if you've got a machine which can't fit multiple drives and you want some protection against UREs without having to resort to backups.
|
|
# ? Aug 2, 2022 23:36 |
|
BlankSystemDaemon posted:Wait, btrfs doesn't have an equivalent of ditto blocks? Your red txt is perfect.
|
# ? Aug 3, 2022 00:18 |
|
I would like to automate the downloading of Linux isos on my EndeavourOS install (Arch derivative), something that was trivially simple on Windows, using Sabnzbd and Radarr. I have done my best at googling, and it seems the simplest way for me to do this is to run the radarr.service as my own user, by editing it and adding changing the user and group in the override section. I have <indicated> where I figure I need to make changes. code:
code:
I'm hours away from going back to Windows to avoid all these permissions headaches. I just want to download Linuxeses.
|
# ? Aug 3, 2022 08:34 |
|
Who's the owner of the download folder? Does it have permission for you to write? Does it work if you leave the group as radarr?
|
# ? Aug 3, 2022 09:54 |
|
The folder is owned by my user. If I leave the group as radarr I cannot access the folder. I've read so much on folder permissions and users and groups, my brain has gone blurry.
|
# ? Aug 3, 2022 10:32 |
|
What does systemctl status radarr.service and journalctl -xe say on the matter after you attempt to start the service? As root or via sudo ofc.
|
# ? Aug 3, 2022 10:39 |
|
Systemctl after I try changing the user and groups:code:
code:
|
# ? Aug 3, 2022 11:16 |
|
yeah, we'll need to go up in the journalctl output, the "at NzbDrone.Common.EnvironmentInfo.AppFolderFactory.Register() in D:\a\1\s\src\NzbDrone.Common\EnvironmentInfo\AppFolderFactory.cs:line 55" line at the top is clearly cut-off output of the error. Go up until you you see something like "Unit radarr.service has begun starting up"
|
# ? Aug 3, 2022 11:29 |
|
code:
|
# ? Aug 3, 2022 11:38 |
|
The problem is it can't write to /var/lib/radarr. You either need to make that location and the existing files writable by jt4527 or change the "-data=/var/lib/radarr" part of the service file to write to a location owned by jt4527.
|
# ? Aug 3, 2022 11:44 |
|
So I would chown that directory to jt4527:jt4527?
|
# ? Aug 3, 2022 11:45 |
|
That should work. Put a recursive -R on there. That said; why does the default install run as root and why do you need to run it as not root?
|
# ? Aug 3, 2022 11:48 |
|
I've no idea why it runs as root, but when I try and setup up my folders in radarr I only have access to the top level directories and when I click save, it says the directory is not writable by the user radarr. I just want to run the service as a user I know has access to the files and folders.
|
# ? Aug 3, 2022 11:51 |
You should probably fix it so it doesn't run as root. On FreeBSD, daemon(8) can drop privileges for any arbitrary program or script - I'd be shocked if Linux can't do it somehow.
|
|
# ? Aug 3, 2022 11:52 |
|
I probably should, but I've got it working now and that'll do for me. Thanks Tesseraction and Pablo Bluth for taking a look for me.
|
# ? Aug 3, 2022 11:58 |
|
Sorry, I may have misread radarr as root, confusing the issue. A quick google shows install notes that say: User=radarr Group=media If you want access radarr content as jt4527, then you could also add that account to the media group code:
Or you add the radarr user to a group that gives it permission to access a folder owned by jt4527. But not being familar with radarr, how you machine is exactly setup or exactly which directories you want radarr to write to, it's hard to provide exact solution. In theory you've lost a bit of security benefit of separation between radarr and your normal account (eg radarr has a flaw that allows an attacker a foothold and they can immediately pivot to stealing your browser saved passwords) but the risk assessment is your own.
|
# ? Aug 3, 2022 12:17 |
|
Being fair that guide does later say in a red box quote:Note: This assumes you will run as the user radarr and group media. You may change this to fit your usecase. It's important to choose these correctly to avoid permission issues with your media files. We suggest you keep at least the group name identical between your download client(s) and Radarr. Which does make the warning about permission issues. Ah, permissions, the gift that keeps on stinking.
|
# ? Aug 3, 2022 12:20 |
|
Pablo Bluth posted:Sorry, I may have misread radarr as root, confusing the issue. As it's finally running I shall leave as is for now, but knowing a little more about what I need to be doing I will have a look at doing it properly in the future. Thanks again.
|
# ? Aug 3, 2022 12:38 |
|
UNIX permissions are super simple, try proper ACLs if you want stink. The guy posted how his Radarr service looks in his very first post asking about this:Dead Goon posted:
It should be easy to decipher that the service will run as user radarr and group radarr by default. What the -data option does you would have to look at documentation to be sure, but given that it's pointing to a path under /var/lib, a good guess would be that it's used for storing program state. That means it needs to be writable for the service process. Technically and security wise it would be nicer to set up fitting group and adding yourself and radarr to it, running the service as system user radarr instead of giving the software full access to everything your login user account can do when the only thing you need is to be able to access the data it spits out to disk, but I know most people just don't care enough to do this at home.
|
# ? Aug 3, 2022 12:58 |
|
Keito posted:Technically and security wise it would be nicer to set up fitting group and adding yourself and radarr to it, running the service as system user radarr instead of giving the software full access to everything your login user account can do when the only thing you need is to be able to access the data it spits out to disk, but I know most people just don't care enough to do this at home. But also, the security risk is that Radarr has bugs or is straight-up bad software that will delete your files. This has nothing to do with other potential users on a time-share machine.
|
# ? Aug 3, 2022 15:46 |
|
Keito posted:UNIX permissions are super simple, try proper ACLs if you want stink. Oh for sure, but there's no more frustrating cause of a problem than realising you hosed up a permission/file group.
|
# ? Aug 3, 2022 15:57 |
|
Dead Goon posted:As it's finally running I shall leave as is for now, but knowing a little more about what I need to be doing I will have a look at doing it properly in the future. Keep this in mind because this is how a *lot* of linux services work: they make a new limited user to run themselves as, and rely on group permissions to gain access to files. This can get a bit complicated to keep track of because you now have a bunch of groups. And a folder can only be owned by one group, nor can you have a group in a group. So radarr by default does the radarr:media, and you're supposed to just change the group on home/myuser/Videos or whatever to myuser:media, with r/w given to the group. That way you have ownership and both you and radarr (and anyone else in media) has full access. But radarr can't see anything else in your home folder. But maybe you also have a sambashare group controlling access to folders you're sharing to the network, and you want to share ~/Videos. At that point you can add the network users from samba to media group, or radarr to sambashare. Start taking notes when you have more than 3 or 4 of these types of interactions.
|
# ? Aug 3, 2022 16:46 |
|
When I finally worked out the find command for fixing all the perms in a tree my life was changed In the short term I had to rebuild my box because fixing all the perms in the whole filesystem turns out to be a bad idea lol
|
# ? Aug 3, 2022 16:47 |
|
Klyith posted:This can get a bit complicated to keep track of because you now have a bunch of groups. And a folder can only be owned by one group, nor can you have a group in a group.
|
# ? Aug 3, 2022 17:13 |
|
|
# ? Jun 8, 2024 15:39 |
|
Dead Goon posted:The folder is owned by my user. Leveraging linux ACLs would trivialize this, leave the service running as radarr user/group and let it control everything and just add your user to the ACL, both default (so new folders get it) and current lists. code:
But also you shouldnt need to leverage ACLs. Lotta ways to cut the problem. Mr. Crow fucked around with this message at 17:55 on Aug 3, 2022 |
# ? Aug 3, 2022 17:52 |