|
Just sort by modification time; it'll give you the time the file was last written, which will be when the download completed.
|
# ? Nov 10, 2019 18:25 |
|
|
# ? Jun 6, 2024 04:09 |
|
If you mount filesystems noatime, as one should, atime will sort of work like NTFS creation time. Preserving the atime across filesystems is harder than NTFS creation time, GNU cp can preserve atime with the -a option but tar and rsync won't.
|
# ? Nov 10, 2019 18:31 |
|
ext4 stores crtime at the inode level but there really aren't any Linux syscalls that expose it. So you're left with cludgy stuff like this that require sudo or root access: http://moiseevigor.github.io/software/2015/01/30/get-file-creation-time-on-linux-with-ext4/ Edit: I take that back and statx exists now and glibc supports it. http://man7.org/linux/man-pages/man2/statx.2.html So it's really just a matter of using a file system that records the creation time and a file manager that uses statx or similar to get file info. Edit2: Here is a stat replacement in go that gives you crtime (birth time) without root. https://github.com/tklauser/statx waffle iron fucked around with this message at 20:13 on Nov 10, 2019 |
# ? Nov 10, 2019 20:06 |
Edit: Nevermind.
|
|
# ? Nov 11, 2019 02:43 |
|
Had a bad experience with installing the standard debian/gnome desktop on a new machine. Fresh install of stable, not a single thing done yet and gnome-shell was just really going to town on every core. Like, could barely operate the machine it was so unresponsive. Not sure what's really to blame for it or where the bug is, but installing the proprietary nvidia-driver resolved the issue. Yeah, yeah, when I upgrade my gpu next I'll get an amd since they seem to not hate the floss world. On the other end of the spectrum, I had a super quick and easy experience upgrading my debian server from old stable to current stable. Everything Just Worked™.
|
# ? Nov 20, 2019 07:18 |
|
I'm thinking about throwing a modern distro of linux on a dual Pentium III-S 1.4ghz with 4GB of PC133 ram and some flavor of AGP video card. Any guesses if I can watch 720p/1080p youtube and have an overall useable experience with modern websites?
|
# ? Nov 21, 2019 21:45 |
|
I think it will choke on modern video codecs at modern resolutions but hey who knows. e: from: http://forum.doom9.net/showthread.php?s=a9cc2e295c57dfb80c20b828319a8d90&t=143455&page=42 quote:Works on the P-III (850 MHz) as well. (no clue at all how this really stacks up with something like ffmpeg) taqueso fucked around with this message at 21:59 on Nov 21, 2019 |
# ? Nov 21, 2019 21:55 |
|
1080p will be a problem, if you do a bit of googling that era could barely handle 720p.
|
# ? Nov 21, 2019 21:58 |
|
xzzy posted:1080p will be a problem, if you do a bit of googling that era could barely handle 720p. Really? My modded xbox1 with its gimped PIII 700mhz could decode 720p24 h.264 from that era no problem.
|
# ? Nov 21, 2019 22:03 |
|
I'm guessing it has some hardware acceleration for video decoding, but that's just a guess. If you already have this hardware, I say install it and see what it can do.
|
# ? Nov 21, 2019 22:40 |
|
taqueso posted:I'm guessing it has some hardware acceleration for video decoding, but that's just a guess. If you already have this hardware, I say install it and see what it can do. That's the kicker I don't have most of it yet. Either way I'll build it but just getting ahead of myself.
|
# ? Nov 21, 2019 22:57 |
|
Shaocaholica posted:That's the kicker I don't have most of it yet. Either way I'll build it but just getting ahead of myself. It's worth noting too as discussed a few pages ago that chrome doesn't attempt to use hardware acceleration if present on linux for twitch and youtube (It DOES use it for chrome rendering but that barely matters). VNC does, but it's something to keep in mind.
|
# ? Nov 21, 2019 23:50 |
|
SoftNum posted:It's worth noting too as discussed a few pages ago that chrome doesn't attempt to use hardware acceleration if present on linux for twitch and youtube (It DOES use it for chrome rendering but that barely matters). VNC does, but it's something to keep in mind. Good to know thanks. I know there are going to be software limitations as the P6 architecture doesn't support a lot of modern instructions either. So does that also include chromium?
|
# ? Nov 21, 2019 23:54 |
|
Shaocaholica posted:Really? My modded xbox1 with its gimped PIII 700mhz could decode 720p24 h.264 from that era no problem. It took years before the video decoding got efficient enough to do realtime SD region H.264 on Xbox, it certainly wasn't effortless.
|
# ? Nov 22, 2019 00:16 |
|
Shaocaholica posted:Good to know thanks. I know there are going to be software limitations as the P6 architecture doesn't support a lot of modern instructions either. If I remember right there's patches you can put on chromium that will do it. There's nothing technically preventing it iirc. I didn't pursue it much cause I have a ryzen I don't use much so if twitch needs a core I've got plenty.
|
# ? Nov 22, 2019 00:25 |
|
Shaocaholica posted:Good to know thanks. I know there are going to be software limitations as the P6 architecture doesn't support a lot of modern instructions either. This might potentially be a situation where you could compile some packages targeting your specific CPU and actually get some speedup. comedy option full gentoo (or whatever is like that now) install. It's not actually too bad, even on that kind of hardware it takes about a day to do X11 + desktop environment from scratch IIRC, not weeks like people would have you believe I have no idea what the modern gentoo is, its been so long since i played with that
|
# ? Nov 22, 2019 01:00 |
|
Shaocaholica posted:I'm thinking about throwing a modern distro of linux on a dual Pentium III-S 1.4ghz with 4GB of PC133 ram and some flavor of AGP video card. Any guesses if I can watch 720p/1080p youtube and have an overall useable experience with modern websites? I was running Gentoo those days, and I don't think it has changed much.
|
# ? Nov 22, 2019 01:09 |
|
Don’t most distros prebuild i686 still? That’s literally P6/Pentium-III arch unless I’m misunderstanding what i686 means.
|
# ? Nov 22, 2019 04:52 |
|
Shaocaholica posted:Don’t most distros prebuild i686 still? That’s literally P6/Pentium-III arch unless I’m misunderstanding what i686 means. actually it's harder to get a i686 build of distros these days You have to go back a major release to find it for Ubuntu. I guess debian releases one for 10. arch32 is an unofficial thing.
|
# ? Nov 22, 2019 05:22 |
|
I think most distros stopped shipping packages that will run on very old CPUs in the last couple if years. I also think the kernel has been pruning some old code which supported these older architectures for the last few release cycles as well.
|
# ? Nov 22, 2019 07:48 |
|
This article lists some Linux and BSD distros that still support 32-bit processors of that vintage: https://itsfoss.com/32-bit-os-list/ I was curious because I have a dual P-III server sitting around in my garage I haven't fired up in ages that I may want to do something with. It has a PCI videocard in it, and was able to run Win XP okay and whatever Linux distro I had installed on it. It's been years and I can't remember what I had installed on it. I seem to remember it struggling with video in browsers, but able to play from file okay, but I couldn't swear to that.
|
# ? Nov 22, 2019 07:59 |
|
So it looks like Fedora adopted wayland since release 25. And they dropped i686 kernel builds with the last being release 30. So if I wanted to run wayland on i686 then I would need a x86_32 build of release 30...but I can't seem to find that download anywhere
|
# ? Nov 25, 2019 02:14 |
|
That's the thing with running Fedora. You've gotta be prepared to keep upgrading every ~6months. Fedora 30 will probably be unsupported in the next 3 or 4 months. It's all in the name of progress.
|
# ? Nov 25, 2019 02:17 |
|
That's fine this isn't for a daily driver. e: found it on the Redhat mirror but doesn't seem to be hosted at all on any other mirrors. Probably like 3 ppl using the x86_32 build in the world. Shaocaholica fucked around with this message at 08:04 on Nov 25, 2019 |
# ? Nov 25, 2019 02:21 |
Shaocaholica posted:So it looks like Fedora adopted wayland since release 25. And they dropped i686 kernel builds with the last being release 30. So if I wanted to run wayland on i686 then I would need a x86_32 build of release 30...but I can't seem to find that download anywhere
|
|
# ? Nov 25, 2019 14:05 |
|
D. Ebdrup posted:You could also run FreeBSD. Is there a fork of FreeBSD that supports i686 and wayland? QUick googling suggests mainline FreeBSD doesn't support wayland yet.
|
# ? Nov 25, 2019 15:57 |
SoftNum posted:Is there a fork of FreeBSD that supports i686 and wayland? QUick googling suggests mainline FreeBSD doesn't support wayland yet. *: I'm running Kodi with Wayland on 12-STABLE from before 12.1-RELEASE came out, and there it works well. At some point I plan to go to -RELEASE on that box, so I'll find out whether it works for sure. Additionally, after posting this, I also found someone else who reports that it works. EDIT: Also note that 80386 isn't supported because FPU can't be guaranteed, and it's simply too slow without it. BlankSystemDaemon fucked around with this message at 16:31 on Nov 25, 2019 |
|
# ? Nov 25, 2019 16:16 |
|
NetBSD will run on anything
|
# ? Nov 25, 2019 18:23 |
taqueso posted:NetBSD will run on anything
|
|
# ? Nov 25, 2019 18:54 |
|
How are the drivers for the NVidia GTX 1650 4 GB? I'm getting a Thinkpad X1 Extreme(Gen 2) with one of those babies in it, and although I'm aware it might be suboptimal to using Windows.. I imagine I can still play a number of games pretty well with proton from Stream provided the NVidia drivers are pretty okay. I'm fine with the drivers being proprietary.
|
# ? Nov 25, 2019 18:55 |
|
Nvidia's site seems to say it's supported in the latest version. Gaming on Linux with Wine can still be pretty hit and miss (depends on the game, really), but Steam Play and Lutris with DXVK and D9VK are pretty good. Quite a few games have Linux versions, too.
|
# ? Nov 25, 2019 19:22 |
|
I have a 1080 and use the nvidia native drivers and I can play linux native (Factorio) and Lutris (Blizzard games) basically flawlessly. I've played some stuff through proton (Steam) as well, though it is hit and miss.
|
# ? Nov 25, 2019 21:35 |
|
Little homelab exercise I've been toying around with. Could someone talk me through some logistical questions on setting up a cluster with compute nodes that boot from PXE/iSCSI? Perhaps not physical nodes but let's say VMs for now. If I did physical hardware it would probably be boards with IPMI. Presumably I would want some immutable system image to contain the important parts of the filesystem - binaries and libraries and such. Then I would have a PXE image that contains the /boot and grub and such that tells the node where to go to find the iscsi image? And the home directories and stuff are just ephemeral and disappear after shutdown. How do I logistically go about assigning IPs and bringing the nodes onto SLURM as automatically as possible? It seems like the nodes would ideally find themselves the lowest node number that hasn't been allocated yet and use that as their IP. So let's say it finds node 0 is the first number available, that node becomes 192.168.50.0 (where 192.168.50.x is some range I'm not using). Or is this something that PXE could automatically tweak on the boot image to tell it what node it is? Or do I use a GUID as the node number and just let DHCP assign IPs? If I end up having different PXE images (let's say one for GPU machines and one for CPU machines) how do you control which machine gets which PXE image? Is there something on the tftp server that can auto handle that based on MAC or something like that? Or handle it at a network level, move the different nodes onto different VLANs or allocated DHCP ranges so that they access a different "instance" of the tftp server?
|
# ? Nov 27, 2019 00:23 |
|
Paul MaudDib posted:Little homelab exercise I've been toying around with. This is a big question. When a machine sends out its DHCP discover the DHCP server will refer it to a TFTP server which will hand out a kernel and some parameters for it and possibly an initrd if necessary and where to find its root filesystem which can either be an iSCSI IQN or NFS path. IPs should be assigned through DHCP as well, you can just let it select an IP from the pool, or if you have a big list of all the MACs of the VMs from scraping ec2 or something you can generate dhcpd configs specific for every mac. If you're doing hardware and there isn't a real API to scrape, then scraping switch arp tables are awesome for getting a list of everything. You'll want to drop a script into /etc/rc.local or the systemd equivalent that runs at boot to do any necessary set up or registration. Personally I like using IPs as identifiers, I don't know anything about slurm but I'd set my ID to whatever the IP I was assigned if possible, MAC addresses are also cool. If you have multiple PXE images you want to hand out, again you could probably have a master list of all the MACs of what will be PXE booting and generate a dhcp config specific to each, where the GPUs get pointed to a different TFTP config than the non-GPU. Or if that's not possible you can chainload into iPXE where you can do some more elaborate client-side logic of determining how to proceed with booting. It looks like iPXE can read PCI devices attached so you could do a conditional based on that. http://ipxe.org/cmd/pciscan. An alternative to iPXE for determining if you're a GPU instance or not is exactly what you suggested and have the machines be on different subnets/vlans and have a more generalized DHCP configuration that directs subnet A/eth1 requestors to tftp config A while subnet B/eth2 requestors get tftpconfig B. Methanar fucked around with this message at 00:49 on Nov 27, 2019 |
# ? Nov 27, 2019 00:45 |
|
Yes, if you dig into the specifics of pxelinux, machines will ask for a config file based on the install interface's mac address. If that file doesn't exist it chops off an octet and tries again until it runs out of octets. If that still fails it will request pxelinux.cfg. So you would author a config file specific to every node. As for assigning addresses, that's all basic dhcp server config. You can bind ips to mac addresses if you like or just create a pool of addresses the server will hand out to anyone that asks for one. The dhcp server will also specify what boot image to use. Normally for a linux cluster it's going to be pxelinux.0 but if you have alternate boot images (ipxe, whatever) that's another place to customize how a system boots. Another more modern consideration is to look at puppet razor. It's a young product and I haven't had a chance to try it in production yet so no idea how stable it is, but my tests have been promising. It lets you specify classes of servers based on hardware facts (memory, core count, hard drives, whatever) and assign them an OS install based on the results and whether a server role is unfullfilled.
|
# ? Nov 27, 2019 00:54 |
|
Paul MaudDib posted:Little homelab exercise I've been toying around with. I've done this for my Kubernetes cluster. You setup a TFTP server which has the kernel and some other bits. You create a directory layout on the TFP machine which makes it possible to have distinct folders per machine, the Raspberry looks for its ID, some other way is using MAC addresses. After you've done this you export some NFS filesystems based on the IP address of the node ( which you reserved using DHCP ) decompress an Linux image on it and it is off to the races. Took some work, but it works really well and no more SD card problems. Another handy feature is snapshotting the cluster using ZFS before upgrades.
|
# ? Nov 27, 2019 07:23 |
|
https://www.youtube.com/watch?v=4qSziR6sD8Q&t=700s Lol the first 11:40 is just the bootup.
|
# ? Nov 27, 2019 08:31 |
|
Shaocaholica posted:https://www.youtube.com/watch?v=4qSziR6sD8Q&t=700s To be fair, it only takes a minute to get to user mode.
|
# ? Nov 28, 2019 00:21 |
|
I recently took the plunge on using whole-disk encryption on my Fedora install. Everything works, but is there some way to request a reboot (or something close to a reboot) that doesn't require that I trot my happy rear end to wherever the computer is to type the encryption key before the machine will boot up?
|
# ? Nov 28, 2019 06:25 |
|
|
# ? Jun 6, 2024 04:09 |
PBCrunch posted:I recently took the plunge on using whole-disk encryption on my Fedora install. Everything works, but is there some way to request a reboot (or something close to a reboot) that doesn't require that I trot my happy rear end to wherever the computer is to type the encryption key before the machine will boot up? You have to use IPMI, SoL, or some sort of other out-of-band management (VNC server via UEFI-GOP?).
|
|
# ? Nov 28, 2019 09:16 |