|
eschaton posted:I think pics were requested… NICE! what was the stock os?
|
# ? Sep 26, 2021 22:21 |
|
|
# ? Apr 25, 2024 05:55 |
|
the stock “OS” is the HP BASIC/Pascal Workstation system which is extremely primitive, dating from the early 1980s and adapted from HP’s earlier instrumentation systems one reason I’m so eager to get NetBSD booted is so I can dd the hard disk to my server, both to preserve it and see just how it’s all implemented it includes the software modules for setting up and using TCP/IP from BASIC and Pascal, which are hard to find otherwise, so I’d love to get those imaged and maybe use them with some other HP gear (like my adorable little 9816 aka 9000-216, for which I have a LAN card) (oops, wrong pic, edited the right one in)
|
# ? Sep 26, 2021 23:23 |
|
neat. i think the weirdest i've ever had to work on was an early 90s embedded os written for an industrial control system, unix like but not unix, and entirely in german. i think a lot of the knowledge of that stuff is just lost to time at this point.
|
# ? Sep 27, 2021 00:05 |
|
eschaton posted:the stock “OS” is the HP BASIC/Pascal Workstation system which is extremely primitive, dating from the early 1980s and adapted from HP’s earlier instrumentation systems owns amberpos ftw
|
# ? Sep 27, 2021 23:38 |
|
eschaton posted:the stock “OS” is the HP BASIC/Pascal Workstation system which is extremely primitive, dating from the early 1980s and adapted from HP’s earlier instrumentation systems just put ubuntu server on it
|
# ? Sep 27, 2021 23:48 |
|
Olivil posted:owns
|
# ? Sep 30, 2021 11:11 |
|
pretty sure btrfs is now cool and good now that arstechnicas zfs correspondent has written 5000 words about how it is bad and dumb https://arstechnica.com/gadgets/2021/09/examining-btrfs-linuxs-perpetually-half-finished-filesystem/
|
# ? Sep 30, 2021 11:39 |
|
It's cool and good if you only ever use it as a single disk file system, just lmao at everything else. I knew about a lot of its deficiencies, but the article showed me a few additional horrors, so that's certainly something…
|
# ? Sep 30, 2021 11:52 |
|
Oh no a filesystem is not perfect if you use it on random scrounged disks to cobble together an array and expect enterprise-grade reliability who would have thought it?
|
# ? Oct 1, 2021 16:48 |
|
yeah you need ZFS for that
|
# ? Oct 1, 2021 17:54 |
|
Cybernetic Vermin posted:pretty sure btrfs is now cool and good now that arstechnicas zfs correspondent has written 5000 words about how it is bad and dumb https://arstechnica.com/gadgets/2021/09/examining-btrfs-linuxs-perpetually-half-finished-filesystem/ the whole article is a diatribe against raid56, but it's framed like btrfs is in shambles the other issue which is mentioned is already fixed if i recall correctly
|
# ? Oct 1, 2021 18:16 |
|
ars technica is weirdly (but also, not surprisingly) wrong about a lot of very technical stuff at any rate, I'm still of the mind that a dedicated raid layer is better than delegating that stuff to a file system
|
# ? Oct 1, 2021 18:20 |
|
yeah, as the article states, btrfs is a perfectly cromulent single-disk filesystem to replace xfs or ext4, and the additional features like atomic snapshots or reflink file copies can be really nice in that situation
|
# ? Oct 1, 2021 18:20 |
|
I would go so far as to say that snapshots are an absolutely mandatory feature of a modern filesystem unless for a USB drive or something, so that basically means btrfs or ZFS. I use btrfs RAID1 for an array of various size disks, and it works great. I know the possible risks and do regular backups (which you should do no matter which setup you have). Neslepaks posted:yeah you need ZFS for that ZFS doesn't let you cobble together random disks into an array, which is a downside for people on a budget or with hand-me-down hardware, but it also means your disks are going to be new to be of the same size, and hence less likely to fail. KozmoNaut fucked around with this message at 18:38 on Oct 1, 2021 |
# ? Oct 1, 2021 18:35 |
|
KozmoNaut posted:ZFS doesn't let you cobble together random disks into an array, which is a downside for people on a budget or with hand-me-down hardware, but it also means your disks are going to be new to be of the same size, and hence less likely to fail. you also can't expand vdevs yet, which makes trying to add a bit of storage to an array a real pain. there is finally a pull request to add support for it, but realistically it's still a year or more out from being in a production release if zfs were able to overcome the licensing bullshit and be merged into the kernel, btrfs would have no reason to exist
|
# ? Oct 1, 2021 18:47 |
|
sb hermit posted:ars technica is weirdly (but also, not surprisingly) wrong about a lot of very technical stuff The article is completely correct though. The whole point of having a simple mirror of two devices is to protect against one of those devices dying. Having to go through a rescue shell just because one drive dropped off is insane and it was the deciding factor why none of our system drive pairs are btrfs. I didn't know btrfs would even gently caress up when the lost drive comes back. The whole bloody point of that entire exercise is that I don't have to janitor this poo poo.
|
# ? Oct 1, 2021 18:48 |
|
KozmoNaut posted:I would go so far as to say that snapshots are an absolutely mandatory feature of a modern filesystem unless for a USB drive or something, so that basically means btrfs or ZFS. this is a completely deranged opinion
|
# ? Oct 1, 2021 19:05 |
|
Is btrfs still pig slow with mounted container volumes?
|
# ? Oct 1, 2021 19:09 |
|
DoomTrainPhD posted:Is btrfs still pig slow yes
|
# ? Oct 1, 2021 19:12 |
|
Nomnom Cookie posted:this is a completely deranged opinion Checksumming, too. Don't use an FS that doesn't have it.
|
# ? Oct 1, 2021 19:12 |
|
Excellent to know!
|
# ? Oct 1, 2021 19:15 |
|
The_Franz posted:if zfs were able to overcome the licensing bullshit and be merged into the kernel, btrfs would have no reason to exist why should it ever be merged into the kernel? why does anything need to be? as long as the kernel exposes a stable set interfaces, projects like ZFS can and should be able to exist outside it, be built independently, and still just work
|
# ? Oct 1, 2021 19:50 |
|
the zen kernel on your laptop — yes? no?
|
# ? Oct 1, 2021 19:55 |
|
DoomTrainPhD posted:Is btrfs still pig slow with mounted container volumes? not if you remember to chattr +C it btrfs needs janitoring for db and containers, but it's not slow
|
# ? Oct 1, 2021 20:26 |
|
eschaton posted:why should it ever be merged into the kernel? why does anything need to be? because it will hold it to a higher standard of scrutiny and compatibility
|
# ? Oct 1, 2021 21:24 |
|
Turning off COW on a COW fs is some insane hacky poo poo. Having COW in the fs means you can guarantee atomicity, which means you can jettison a ton of extra work an rdbms is doing to keep a consistent state despite other file systems being unreliable. Turning off COW turns off checksumming for fucks sake. Btrfs really is a joke file system for any serious usage beyond a single disk.
|
# ? Oct 1, 2021 21:42 |
|
i lost all respect for btrfs and its advocates with that chattr +C poo poo. admit that your approach has tradeoffs and own it. if you can't do that then you've got no business claiming that you're making software for grownups
|
# ? Oct 1, 2021 22:10 |
|
eschaton posted:as long as the kernel exposes a stable set interfaces this will never happen and you know it and you know why lmao
|
# ? Oct 1, 2021 22:11 |
|
btrfs chat is timely and has reminded me that i will not try to use it when i build a nas
|
# ? Oct 1, 2021 23:20 |
|
btrfs is fine and of course you should turn off CoW when using it with a db or large vms
|
# ? Oct 1, 2021 23:30 |
|
Nomnom Cookie posted:i lost all respect for btrfs and its advocates with that chattr +C poo poo. admit that your approach has tradeoffs and own it. if you can't do that then you've got no business claiming that you're making software for grownups wow
|
# ? Oct 1, 2021 23:31 |
|
it's illegal to tweak your system for your workload
|
# ? Oct 1, 2021 23:31 |
|
Antigravitas posted:Btrfs really is a joke file system for any serious usage beyond a single disk. if you have an nvme drive, you are leaving performance on the table by not using xfs, especially as more software takes advantage of io_uring
|
# ? Oct 1, 2021 23:32 |
|
also don't forget to set the "make it go fast" options on dm-crypt in case your distro's setup program doesn't set them for you automatically (no-read-workqueue, no-write-workqueue)
|
# ? Oct 2, 2021 00:23 |
|
Tankakern posted:it's illegal to tweak your system for your workload it explains a lot about your posting if this is sincere
|
# ? Oct 2, 2021 00:34 |
|
never change, linux thread
|
# ? Oct 2, 2021 01:09 |
|
Tankakern posted:btrfs is fine and of course you should turn off CoW when using it with a db or large vms to be honest, I'll probably switch back to btrfs once rhel makes it standard in its out-of-the-box kernels, at least for my file servers which may come soon, if desktop btrfs continues to gain traction
|
# ? Oct 2, 2021 01:13 |
|
In a non-trolly fashion I don't understand how with the backing of Facebook and RHEL the RAID5/6 issues are still a thing. Is it just fundamentally broken?
|
# ? Oct 2, 2021 04:51 |
|
Gentle Autist posted:because it will hold it to a higher standard of scrutiny and compatibility I’d think having stable binary interfaces would provide a much higher standard of compatibility since their behavior can’t really change
|
# ? Oct 2, 2021 04:56 |
|
|
# ? Apr 25, 2024 05:55 |
|
eschaton posted:I’d think having stable binary interfaces would provide a much higher standard of compatibility since their behavior can’t really change never change
|
# ? Oct 2, 2021 05:21 |