|
I use Gentoo btw.
|
# ¿ Jan 31, 2022 09:18 |
|
|
# ¿ Apr 27, 2024 00:21 |
|
We'll call it /etc2
|
# ¿ Feb 2, 2022 18:33 |
|
I think I remember Earth 2150 could use glide or opengl. Man, I want to play Earth 2150 now. And Warzone 2100.
|
# ¿ Feb 4, 2022 14:25 |
|
I own physical copies of Earth 2150 and its expansions. It's a good game.
|
# ¿ Feb 4, 2022 14:55 |
|
NihilCredo posted:it isn't, the guy openly admitted that he did it for fun and to learn systems-level programming with rust It's not even a kernel driver. So really, ultra-low stakes.
|
# ¿ Feb 4, 2022 20:53 |
|
I was helping someone find out which USB cable was having intermittent issues and was corrupting poo poo on their e-reader. I then told them to open whatever disk utility their distro ships and do an fsck of the file system of that reader, because it was shot to bits. A few extremely confusing exchanges later, I had to check something… quote:Disks 3.3+ is a complete rewrite of the old gnome-disk-utility (aka Palimpsest). It no longer appears to include an option to check filesystems gparted did the trick…
|
# ¿ Feb 20, 2022 21:22 |
|
Sapozhnik posted:Image failed to attach What the gently caress.
|
# ¿ Feb 21, 2022 08:45 |
|
nudgenudgetilt posted:sadly I don't think it'll happen because there is no money to be made in locking down the supply chain Web devs are basically reimplementing Windows, aren't they.
|
# ¿ Feb 22, 2022 09:20 |
|
I doubt they'll fix btrfs' myriad deficiencies. I'd assume they use it as a single disk fs with transparent compression and snapshots, and in that use case btrfs is fine. It just can't do anything more than that.
|
# ¿ Mar 3, 2022 18:15 |
|
Tankakern posted:pretty weird to conflate "anything" with "raid56" I am not. Literally every facet of btrfs is broken by design, from admin tools to behaviour in degraded mode to data loss issues trying to recover from them. Single disk use is the only use case for btrfs. If I needed redundancy with btrfs, I'd use mdraid with btrfs on top. (I don't, because I'm responsible for hundreds of TB of research data, and I'm not going to entrust that to btrfs)
|
# ¿ Mar 3, 2022 19:46 |
|
Farmer Crack-rear end posted:assuming btrfs wasn't buggy, would there be particular benefits to using btrfs raid instead of mdraid? Yes, several reasons in fact. As the fs knows exactly which blocks are used and which aren't, it can reduce the amount of data it has to transfer to a new disk when a drive fails. That's important since you want to return to normal service asap when a drive fails. Since the fs checksums every block it can detect when the platter gives it bad data and instead read from redundancy. That's something mdraid can't do. There are also issues where a drive loses connection to the array for a bit and is out of sync. If the fs directly controls the array, and if the fs is properly designed, it can catch up the disk very quickly. All of these things have massive asterisks and foot guns with btrfs, but they work exceptionally well in zfs.
|
# ¿ Mar 3, 2022 21:18 |
|
nudgenudgetilt posted:I love how every post or high level "how-to" about zfs on linux (or any os, really) related completely ignores how unsuitable a default config zfs is anywhere but on a file server. The typical new zfs user spends about 6 months before they go "where the gently caress did my ram go?" and learn about configuring arc. Uh, on. ZFS relinquishes RAM just fine (though you want it as big as possible because it's better than Linux's page cache), and its default config works ootb (though you really want to enable compression). I wouldn't use it outside of file servers for other reasons, mostly because root on zfs is a trash fire on Linux.
|
# ¿ Mar 5, 2022 12:11 |
|
I don't care about layering violations when it makes my data safer. I'd get lynched if research data gets silently corrupted. Do note that I actually have had several instances of just that happening, from sporadic garbage data returned from disks to just absolutely spewing the bus with hundreds of garbage blocks that the entire Linux storage stack did not even alert on. Disks absolutely lie.
|
# ¿ Mar 5, 2022 15:23 |
|
That seems like a distinction without a difference to me.
|
# ¿ Mar 23, 2022 13:45 |
|
Have you tried not using a toy distro?
|
# ¿ Mar 23, 2022 18:58 |
|
I, too, like the init system thing part. I'm less happy with the log handling thing. The rest is a dumpster fire. Having an easy way to put FlexLM daemons into lowest-privilege container-like isolation with only a few switches is pretty neat. They even get ephemeral users assigned to them.
|
# ¿ Apr 8, 2022 12:46 |
|
Use podman instead.
|
# ¿ Apr 8, 2022 14:13 |
|
Gnome is never okay, actually.
|
# ¿ Apr 22, 2022 11:38 |
|
Probably not a hard dependency but a "recommends" dependency. That system usually works to pull in packages that aren't strictly necessary but that you'd want to install anyway, but it can lead to some weirdness if you're pulling a desktop package into a non-desktop system.
|
# ¿ Apr 25, 2022 21:27 |
|
Earning money with Linux sure beats earning money with Windows. I've done both
|
# ¿ Apr 26, 2022 10:36 |
|
Most of my work isn't with tickets, I just need to keep things running and available and I have full discretion over how the things run. It's so much more pleasant to work with something that isn't actively antagonistic.
|
# ¿ Apr 26, 2022 19:45 |
|
My Selinux story: I found out that setting it to enforcing didn't log all denials, and setting it to permissive didn't either. It logged most of them, but a certain subset (I can't recall which) wasn't logged. In effect, things didn't work and it was impossible to tell why. I was not amused. Not how I'd design a MAC system…
|
# ¿ Apr 30, 2022 23:24 |
|
The Mesa project is pretty impressive. And I think they have some fun names: https://gitlab.freedesktop.org/marge-bot
|
# ¿ May 6, 2022 09:24 |
|
Sapozhnik posted:i hear people have personal computers these days that let them carry their "home directories" around with them Our users have been using their workstations remotely, but their $HOME is still NFS-mounted, Kerberised and all. That setup has a number of advantages for people who do things that don't typically fit on a laptop.
|
# ¿ May 11, 2022 15:18 |
|
The Nouveau developers don't need more information. Nvidia locked down their hardware further after Nouveau started becoming useful and now most features require special signed firmware that is part of the proprietary driver. I honestly don't know why anyone bothers developing Nouveau considering the open contempt Nvidia has shown towards it tbh. gently caress Nvidia, especially for all the wasted time I've had janitoring their garbage proprietary driver across our fleet. e: Oops, an update while the system is running? Better SIGSEGV the X driver and all programs using CUDA, wouldn't want to have a smooth experience on our workstations for minor updates, would we? Nothing else has this problem, but we, Nvidia, we're special, we need special janitoring. gently caress Nvidia, seriously. Antigravitas fucked around with this message at 09:25 on May 15, 2022 |
# ¿ May 15, 2022 09:22 |
|
Or unless you have an NFS mount. I had really some old system that some dolt had configured hard NFS mounts on (no intr either) and when the NFS server became unavailable entire trees on the file system became tar pits that I had to carefully avoid because even an accidental tab complete would lock the shell forever. The infrastructure I encountered when I started that job still infuriates me. One part of the infra also used LDAP, a single Windows DC, Samba localdb, and motherfucking NIS, all in the same lab.
|
# ¿ May 18, 2022 19:55 |
|
Routing mail from nullmailer to a real MTA is a problem that is embarassingly trivial.
|
# ¿ May 19, 2022 20:16 |
|
euroshopper posted:how compatible is opensuse with proprietary printer drivers? [whispering]oh nooooooooooooooo[/whispering]
|
# ¿ May 19, 2022 20:46 |
|
Clearly, they mean Perforce.
|
# ¿ May 21, 2022 20:18 |
|
I'm fairly certain the intent behind DynamicUser is to use it for services you'd run as nobody, but a bit neater. If you need to wire secrets to them you probably want to allocate a real user instead of an ephemeral one. Though, can't you use an env file with such a service? That env file would be sourced by init itself and passed to the service process, iirc.
|
# ¿ May 31, 2022 14:15 |
|
Using a spool directory is a super common pattern and so is recovering from restart/sigterm/crash by reading from it.
|
# ¿ Jul 11, 2022 21:41 |
|
You can use a bleeding edge SDL2 to upgrade even binaries that statically link against SDL2 via SDL_DYNAMIC_API Doc: https://github.com/libsdl-org/SDL/blob/main/docs/README-dynapi.md Source (warning, hax): https://github.com/libsdl-org/SDL/blob/main/src/dynapi/SDL_dynapi.c Sadly, not everything uses SDL.
|
# ¿ Jul 14, 2022 19:51 |
|
quote:Using the ssd mount option with older kernels than 4.14 has a negative impact on usability and lifetime of modern SSDs
|
# ¿ Jul 16, 2022 10:43 |
|
I checked. It's dumber. https://github.com/torvalds/linux/commit/583b723151794e2ff1691f1510b4e43710293875
|
# ¿ Jul 16, 2022 11:37 |
|
Way ahead of you on any system I care about, but I wanted to have transparent compression without tainting my kernel…
|
# ¿ Jul 16, 2022 11:45 |
|
I have no particular allegiance to Linux, but KDE Plasma? Actually good.
|
# ¿ Jul 16, 2022 18:48 |
|
We built a system for archival of large files. It is extremely slow "all of a sudden" and I was called in to investigate. It is completely swamped with random iops, and just an utter deluge of stat() and whatnot calls, just absolutely murdered. Here's a file size histogram of just one directory. pre:1k: 669933 2k: 119527 4k: 125188 8k: 115541 16k: 97343 32k: 75153 64k: 52402 128k: 29671 256k: 11723 512k: 3304 1M: 755 2M: 42
|
# ¿ Jul 25, 2022 15:49 |
|
That's a conclusion I have reluctantly come to as well. It was built for nice, clean, large streaming writes and reads to saturate 10Gbps. Those were the requirements. Instead, someone is using it in literally the absolute worst access pattern possible. ngl, I feel a bit betrayed. We were assured it wasn't going to be used in this way. MATLAB
|
# ¿ Jul 25, 2022 16:20 |
|
The stakes aren't that high, but yes, I have the documentation that they specifically asked for the opposite of this. But regardless, no file system will perform well when used like this. Lmao, not all file system metadata fits in cache and any directory walk generates a storm of random I/O. I'll have to actually help them get their code to not do the thing they are doing…
|
# ¿ Jul 25, 2022 16:58 |
|
|
# ¿ Apr 27, 2024 00:21 |
|
Perplx posted:just throw some optane drives at I do not want to enter the university finance labyrinth. Not my budget . Sure, special vdev, l2arc, more RAM. Is there still money left in the grant? Who knows, not me. Putting lipstick on a pig anyway, the system is designed for the opposite use pattern and it's working as designed. It's either educating users to use it as designed, or design another system. That's what I've communicated to the work group.
|
# ¿ Jul 25, 2022 17:40 |