|
big black turnout posted:I'm using reiserfs for its killer performance, though it does suffer from the occasional fragmentation issue 
|
#
?
Apr 20, 2024 21:40
|
|
|
|
| # ? May 22, 2024 07:06 |
|
|
yessssssss
|
|
#
?
Apr 20, 2024 21:41
|
|
|
Beeftweeter posted:i know what fs i'm using (ext4 or 5 on linux laptops, ext5? are you from the future?
|
|
#
?
Apr 20, 2024 21:44
|
|
|
spankmeister posted:ext5? are you from the future? i could swear one said ext5 but i guess not lol. see, i really don't care
|
|
#
?
Apr 20, 2024 21:46
|
|
|
spankmeister posted:ext5? are you from the future?
|
|
#
?
Apr 20, 2024 21:53
|
|
|
mystes posted:A lot of chromebooks don't have battery backup for the RTCs so it can be hard to stay in the right decade every time the battery dies, disco's back baby
|
|
#
?
Apr 20, 2024 21:55
|
|
|
Soricidus posted:imagine even knowing what filesystem you’re using, let alone caring, unless someone’s paying you to that's kinda the whole point
|
|
#
?
Apr 20, 2024 22:19
|
|
|
my impression of linux filesystems is that your choices are ext4 which you should use, the one that is slightly faster than ext4 but loses data if the system crashes, slow zfs, slow reimplemented from first principles zfs, and a bunch of filesystems you should avoid
|
|
#
?
Apr 20, 2024 23:31
|
|
|
I wonder if the wikipedia editors would allow it if it said "written by a convicted wife murderer" with a citation.
|
|
#
?
Apr 20, 2024 23:51
|
|
|
lol reiser fs support persisted in the kernel until last year
|
|
#
?
Apr 21, 2024 00:02
|
|
|
Clark Nova posted:lol reiser fs support persisted in the kernel until last year they finally killed it?
|
|
#
?
Apr 21, 2024 05:59
|
|
|
spankmeister posted:they finally killed it?
|
|
#
?
Apr 21, 2024 06:04
|
|
|
spankmeister posted:they finally killed it? Only because nobody wants to volunteer to maintain the code lol.
|
|
#
?
Apr 21, 2024 07:51
|
|
|
Someone would probably get volunteered if that thing had any users. But an fs without users is a great candidate for killing, because nobody will notice it gone. In other news, I had the misfortune of using iscsiadm. I had somehow avoided dealing with it until now. I am not a fan. The CLI is loving garbage. Even with a conceptual model of how iscsi works in my head it's confusing to use. The manpage is not much better.
|
|
#
?
Apr 21, 2024 08:10
|
|
|
Tankakern posted:dont use btrfs on lvm unless you’re also using controller-implemented (“hardware”) raid
|
|
#
?
Apr 21, 2024 10:33
|
|
|
eschaton posted:unless you’re also using controller-implemented (“hardware”) raid that kinda reminds me, does an implementation found in uefi count as hardware or software raid? i know the option is there in some machines because they use separate controllers, but i guess intel also has some weirdo implementation that even shows up on one of my laptops (though being from aliexpress it does apparently have a "debug" bios with a lot of different options that just don't work on the machine). it's not like i could really use it anyway since there's only one msata slot (not even nvme/m.2), i'm just curious if that kind of thing uses a separate controller i know that a lot of them are just hardware integrated into the mobo, e.g. the one on my desktop mobo since it requires drivers, but it's not made by intel (some gigabyte braded GRAID thing that is afaict a rebranded jmicron deal)
|
|
#
?
Apr 21, 2024 16:12
|
|
|
Antigravitas posted:Someone would probably get volunteered if that thing had any users. But an fs without users is a great candidate for killing, because nobody will notice it gone. i don't really understand how iscsi works at all, but i'd like to since even windows supports it and so does my consumer-grade router, lol, and i'd like to disconnect the external nvme ssd i have it using atm since that regularly gets third-degree burn hot despite having a thermal pad and heatsink in the enclosure
|
|
#
?
Apr 21, 2024 16:17
|
|
|
take my wife, please
|
|
#
?
Apr 21, 2024 18:38
|
|
|
Iscsi itself isn't all that complex. You have scsi, and you do it via IP. Nodes have an identifier, and can expose block devices. It gets a bit more complicated because to get redundant paths it's normal to expose the same LUN via different IPs from different network interfaces over different storage switches. And on the client side you would also use several network interfaces. So let's say you have 1 PB of storage using 2 controllers with 2 network cards. They expose a LUN, so if you ask the storage controller it tells you you can connect to the same LUN via 4 addresses. On the client side you may have 2 network interfaces. So if you let autodiscover wire everything up you have 2 * 2 * 2 block devices all pointing at the same LUN via different paths, so you use multipathd to condense it back down to a single device. Imagine you have 8 LUNS, you get 8 * 2 * 2 * 2 block devices on the client. You need good tooling to not go crazy, and iscsiadm is not a good tool.
|
|
#
?
Apr 21, 2024 20:04
|
|
|
oooh so that's the point with multipathd
|
|
#
?
Apr 21, 2024 20:11
|
|
|
you learn something new everyday!
|
|
#
?
Apr 21, 2024 20:11
|
|
|
Beeftweeter posted:that kinda reminds me, does an implementation found in uefi count as hardware or software raid? usually, you'll know immediately when hardware raid is present because you get another screen after boot with probably another hotkey to enter it and a very dumb UI that puts 1980s BIOSes to shame honestly between server cpu cores costing fuckall, and hardware raid being just as if not more flaky but much harder to debug than software raid, i avoid it. the only added value of expensive controllers these days is battery backed buffers, but outside hpc, big write buffers are kinda obsolete with the speeds u.2 ssds get these days
|
|
#
?
Apr 21, 2024 20:27
|
|
|
Truga posted:on a desktop pc, it's certainly a software raid unless you separately bought an expensive raid controller. there's a "raid controller" on the motherboard that can handle your drives, but it uses the cpu to crunch poo poo, there's no dedicated hardware for it that all makes sense. so what's the deal with the GRAID thing i have on my gigabyte mobo? i know what you're talking about re: addon cards having separate loading screens, they've done that forever — but if i enable that option, it does have something similar. it comes after the uefi environment loads though, so maybe it uses some program to load the firmware from the chip or something? idk like i said i think it's some rebranded jmicron thing, but since it's in storage i can't tell you what specific model
|
|
#
?
Apr 21, 2024 22:01
|
|
|
Beeftweeter posted:that all makes sense. so what's the deal with the GRAID thing i have on my gigabyte mobo? i know what you're talking about re: addon cards having separate loading screens, they've done that forever — but if i enable that option, it does have something similar. it comes after the uefi environment loads though, so maybe it uses some program to load the firmware from the chip or something? idk GRAID THESE NUTS
|
|
#
?
Apr 22, 2024 04:17
|
|
|
I use btrfs on LVM that is provisioned over an encrypted block device that itself is split over RAID 10. What’s the big fuss???
|
|
#
?
Apr 22, 2024 04:19
|
|
|
sb hermit posted:I use btrfs on LVM that is provisioned over an encrypted block device that itself is split over RAID 10. I use windows and don’t think about my file system at all
|
|
#
?
Apr 22, 2024 04:41
|
|
|
i think about my filesystem a lot when i find myself using windows because that filesystem is slow as poo poo though technically that's because of windows io rather than any particular filesystem running on it
|
|
#
?
Apr 22, 2024 05:44
|
|
|
sb hermit posted:I use btrfs on LVM that is provisioned over an encrypted block device that itself is split over RAID 10. just remove lvm from that equation and we wont fuzz at all
|
|
#
?
Apr 22, 2024 08:18
|
|
|
Help! USB isn't working on my router. It's an Aruba AP-303H running OpenWrt. It's based on a Qualcomm IPQ4019 SoC.  The SoC two USB controllers for USB 2.0, and USB 3.0. The external USB port is on the USB 2.0 controller, I'm pretty sure. Lsusb can see the two hubs (part of the USB controllers inside the SoC). Here's a bootlog. A keyboard is plugged into the USB port. But I think it fails when trying to talk to the USB 2.0 hub inside the SoC. Pastebin: serial console full boot log Very early in the boot, the kernel flips on a GPIO to power up the USB port. A USB power meter plugged into the port will turn on at exactly this point. code:code:Here's my package list. I just shoveled in anything USB-ish, to see if it was just a missing driver. Didn't help. code:e: People of the Future! This has been fixed: post, PR. ryanrs fucked around with this message at 06:45 on Apr 25, 2024 |
|
#
?
Apr 22, 2024 08:28
|
|
|
it's pretty domain specific this, shouldn't you be asking on the openwrt forums?
|
|
#
?
Apr 22, 2024 08:38
|
|
|
Tankakern posted:oooh so that's the point with multipathd Truga posted:on a desktop pc, it's certainly a software raid unless you separately bought an expensive raid controller. there's a "raid controller" on the motherboard that can handle your drives, but it uses the cpu to crunch poo poo, there's no dedicated hardware for it Given that the XOR instruction is computationally cheap and the erasure codes can be offloaded to QuickAssist/GFni, or done using SSE4 or AVX2, there’s no reason to use anything but real software RAID. Sapozhnik posted:i think about my filesystem a lot when i find myself using windows
|
|
#
?
Apr 22, 2024 08:47
|
|
|
Tankakern posted:it's pretty domain specific this, shouldn't you be asking on the openwrt forums? Yes, but maybe there is a Linux USB expert around here that understands the logs better than I do.
|
|
#
?
Apr 22, 2024 08:52
|
|
|
Sapozhnik posted:though technically that's because of windows io rather than any particular filesystem running on it incorrect use of Windows I/O, usually by pretending its equivalent to UNIX I/O because used correctly the VMS (and Windows NT, and Lisa, and classic Mac, and HP3000, and…) asynchronous-callback I/O model provides much better overall throughput than the UNIX “just block and let something else run hope you weren’t interactive LOL” I/O model there’s a reason people who want to do high-throughput I/O on UNIX use asynchronous I/O (remember on UNIX to always check whether every syscall encountered EINTR or EAGAIN and retry!)
|
|
#
?
Apr 22, 2024 08:57
|
|
|
well the openwrt forums are full of clueless openwrt users so i can't imagine very many openwrt experts hang out there (wild guess) maybe a DTB issue?
|
|
#
?
Apr 22, 2024 08:57
|
|
|
BlankSystemDaemon posted:Given that the XOR instruction is computationally cheap and the erasure codes can be offloaded to QuickAssist/GFni, or done using SSE4 or AVX2, there’s no reason to use anything but real software RAID. no, see, if you use hardware RAID and software RAID, you get multiple layers of protection! you can lose an entire array without losing your Linux ISOs!
|
|
#
?
Apr 22, 2024 09:00
|
|
|
shackleford posted:well the openwrt forums are full of clueless openwrt users so i can't imagine very many openwrt experts hang out there port NetBSD to it
|
|
#
?
Apr 22, 2024 09:02
|
|
|
shackleford posted:well the openwrt forums are full of clueless openwrt users so i can't imagine very many openwrt experts hang out there back then, the strength of it was that if something went wrong, you had the option of fixing it. it might, and in my case did, usually require both blood, sweat and tears, but you could usually hack together _something_ that'd make it work it might've also been free, but the 'free as in beer' was as much a joke about getting free beer after having to do a bunch of volunteer work maybe this was also an issue back then, and i was just too wrapped up in trying to fix poo poo to notice, but there's an ever-expanding group of people who don't seem interested in fixing issues, and just 1) want things to work 2) expect a hobbyist to be responsible for it, because their name is on it i dunno if there's anything that can, or even should, be done about it - but it seems like it's going to have some effect eventually, if it hasn't had one already eschaton posted:no, see, if you use hardware RAID and software RAID, you get multiple layers of protection! you can lose an entire array without losing your Linux ISOs! orz BlankSystemDaemon fucked around with this message at 09:12 on Apr 22, 2024 |
|
#
?
Apr 22, 2024 09:09
|
|
|
ryanrs posted:Yes, but maybe there is a Linux USB expert around here that understands the logs better than I do. error -71 is -EPROTO, returned by get_bMaxPacketSize0 when the device returns something other than 8/16/32/64 bytes per packet
|
|
#
?
Apr 22, 2024 10:46
|
|
|
BlankSystemDaemon posted:maybe this was also an issue back then, and i was just too wrapped up in trying to fix poo poo to notice, but there's an ever-expanding group of people who don't seem interested in fixing issues, and just 1) want things to work 2) expect a hobbyist to be responsible for it, because their name is on it this has never not been the case
|
|
#
?
Apr 22, 2024 12:06
|
|
|
|
| # ? May 22, 2024 07:06 |
|
|
BlankSystemDaemon posted:maybe this was also an issue back then, and i was just too wrapped up in trying to fix poo poo to notice, but there's an ever-expanding group of people who don't seem interested in fixing issues, and just 1) want things to work 2) expect a hobbyist to be responsible for it, because their name is on it also, and i post about it ironically sometimes, but there's an underlying truth: 3) people (not the askers) who are interested in having a very online argument about how actually it *does* work, or when it truly is unavoidable, the details of the structures of blame for it not working. that was always a bit of a thing, but spending some unreasonable effort on getting something working is then quite disliked, as it can be construed as a claim that it didn't previously work.
|
|
#
?
Apr 22, 2024 12:13
|
|
