46754 andrew 20 0 15371M 3305M swread 1 1:12 0.23% grep

what the fork?

why is this grep taking 15GB of ram?

ah. because it hit a 10GB sparse binary file

grep loads the entire file into memory at once?

i've recently realized this could be a bad thing.

I don't know why it didn't just skip the file for not being text

otoh this is gnugrep, so breakage is to be expected

is that why with gnu grep it stops in the middle of a file if there's a nul byte?

i've always wondered why it does that.

it just stops and says "Binary file matches"

if there's a nul byte then the file isn't text

is that the definition. or is it system dependent on what the line ending is

i recall trying to read in a text file that was just the bytes 0 to 127, and ruby didn't have any problem with it.

so maybe the default file open thing is not text but binary somehow

all files are binary

a file is just a sequence of bytes. the thing is that some programs want to guess whether that sequence of bytes represents something that can meaningfully be shown to the user by outputting (part of) it to the terminal

well. that makes sense.

why should switching from ttyv0 to v1 trigger some kind of video output change with vt?v

i should somehow unfuck if, as it's not an option to have screen going blank for a second and monitor recognizig new signal on every single terminal switch

i wonder if that's somehow machine specific, or maybe people don't ever use actual console anymore, because it's annoying if done more than once

man sun made some absolutely beautiful computer cases

indeed

i miss them

all the things

dkeav: Over COVID, I took one of the Ultra40 cases and converted it to house an AMD5950 running FreeBSD. It's really sweet.

Hi! I just installed FreeBSD 13.2 on my HiFive Unmatched using the documentation but when I try to bootstrap pkg I get an error message:

pkg: Error fetching pkg.FreeBSD.org/FreeBSD:13:riscv64/quarterly/Latest/pkg.txz: Not Found

The site returns a 404, so that part makes sense, but something seems to have gone wrong with the URI generation? Any help would be appreciated!

Do the packages exist for RISC V?

Unless I'm misreading this I'd imagine so: FreeBSD's support for the RISC-V architecture is currently classified as Tier-2, beginning with FreeBSD 13.0

Is there a website where I could check which packages are available for which ISA?

... per wiki.freebsd.org/riscv#Quick_Start they should

Title: riscv - FreeBSD Wiki

... never mind that URL which was for "QEMU"

RISC V is currently has "Tier 2" support (freebsd.org/platforms ; docs.freebsd.org/en/articles/committers-guide/#archs); so no packages exist

Title: Platforms | The FreeBSD Project

"Tier 1"-supported platforms ought to have packages

Hm, that's sad to hear. I suppose that I can still build them from source?

Do read the second URL on that

I suppose you mean this one: docs.freebsd.org/en/articles/committers-guide/#archs ?

Admittedly I'm quite new to FreeBSD, only used GNU/Linux before. So does that mean I can get the base packages like a compiler from the repository and will have to compile the rest myself?

Title: Committer's Guide | FreeBSD Documentation Portal

Re: URL, Yes (section 21.4). Re: base packages: Does not look like that; may need to compile compiler (GCC and/or LLVM) with compiler from the base system.

Perhaps someone may have compiled the packages to offer publicly(?)

Last I read, LLVM+Clang in base did not come with all the parts that would be installed from The Ports or packages

Sounds like an adventure then, thanks!

Happy Computering!

JohnFranklin, See also wiki.freebsd.org/riscv/ports

Title: riscv/ports - FreeBSD Wiki

hey, no freebsd community on lemmy

c/netbsd and c/openbsd exist

eoli3n: does Lemmy run on FreeBSD? ;)

meena0: i will be able to answer this soon

because i'll try to run my instance on freebsd

eoli3n: in my experience, rust has a better track record than go, so chances are it'll just compile, and if it does, it'll run, too

is there a way to sort a list of IP addresses in the 'expected' order using the sort command ?

so that 192.168.1.130 comes after 192.168.1.50 for example

sort -t . -k 1,1n -k 2,2n -k 3,3n -k 4,4n ipaddr.list

credit: oreilly.com/library/view/bash-cookbook/0596526784/ch08s03.html

Title: 8.3. Sorting IP Addresses - bash Cookbook [Book]

woah

complicated for something apparently so simple, but I can confirm it works

spork_css: now try that for IPv6

meena0: if I not miss anything only replace the '-t .' with '-t :' and use -k 1,1f ... (a few more are needed)

asume full v6 addresses so no truncation

The problem with IPv6 addresses are that it uses a run-length-encoding method of collapsing zeros into a double colon.

::1 is 0000:0000:0000:0000:0000:0000:0000:0001 for example. And so addresses don't sort without expanding first.

with -k 1,1fk ... you only need to expand the '::'

*h

^ thats wrong

newish to freebsd. any recommendations best tool to use for managing regular zfs snapshots, pruning per policy, etc.? zrepl looks like it does the trick, but just wanted to check that there's not a defacto standard i'm missing

markmcb, I am sure there are many different ways but I am using the binary pkg zfstools (aka github.com/bdrewery/zfstools) with cron for automatic snapshot creating and expiration.

I am only doing the auto snaps and not any of the SQL stuff that can also be used. Just set the property and let cron run the script and everything is fully automatic.

rsycn get you anything?

rwp, i like that. nice and simple.

I started with it because M. Lucas recommended it in the Absolute FreeBSD book.

If you "pkg install zfstools" the package description emitted will contain instructions. If you miss that then "pkg info -D zfstools" will print the pkg description again at any time you want to read it again.

thanks. i'm coming from btrfs on linux. my mind is blown by how much more polished the zfs toolset and ecosystem is.

I haven't used zrepl.github.io which you mentioned myself so I am a poor one to talk about it but I think that is designed for replicating datasets from here locally to there remotely. Which I am sure was why timothias mentioned rsync. And I might through syncthing out there too. But you were asking about snapshots and pruning so I think zfs-auto-snapshot is the tool.

Title: zrepl - ZFS replication — zrepl documentation

I sysadmin a variety of systems and btrfs is on some of them (because the people who created them like it) but as I work with btrfs more and more I find just how odd and bizarre it is in many ways. And comparing it to zfs just makes btrfs look sad.

zrepl has some advanced features I wouldn't use, but handles basic snapshots and pruning quite well. (see first half of this config: zrepl.github.io/quickstart/backup_to_external_disk.html )

Title: Local Snapshots + Offline Backup to an External Disk — zrepl documentation

btrfs is nice for it's flexibility. especially in a homelab, it lets you use the disks you have and doesn't complain much.

but definitely the feature set, tool set, and performance is nothing compared to zfs

Oh! Well there you go then. Certainly if one were already using zrepl or also needing the other features then it would make sense to use it for automatic snapshots too.

btrfs has some pitfalls for the unwary though. Especially if actually using the multiple disk raid feature! That's really where all the snags exist.

A single drive btrfs such as on a laptop seems robust enough. But if you use multiple disks in a raid config then there are some traps.

yeah, raid56 is not to be touched. lol. i've never had issues with raid1.

Such as if you actually have a drive failure DO NOT REBOOT or it won't reboot. It won't boot a degraded raid without telling it to boot degraded. Which if that is in a datacenter then you are screwed without a remote console or remote hands.

And do not use the boot degraded mode option all of the time thinking that would solve it either as that turns off a lot of normal safety checks.

If the machine is running, keep it running, then replace the failed device hot, and then "rebalance". If you scrub before rebalancing then AFAIK that also breaks things.

It's not that it can't work but that it has some traps and pitfalls to jump over and if you don't know them already then the experience can be harsh.

yeah, you definitely need to understand its quirks

Meanwhile... I have been fighting with zfs with some flaky hardware and have yet to have zfs treat me badly.

For example I have a system that I think the motherboard is having problems. I am still not sure. So I am still flailing around with it a little bit.

The ZFS array had a serious I/O problem and kicked out half the drives and put the array state to SUSPENDED. Sounds grimm, right?

I popped out and back in three of the drives to give them a hard power cycle reset. Said "zpool clear" to clear the error.

Hardware failures are always a delight. :)

That got the array back online clearing the SUSPENDED state. Did not reboot. Everything came back online.

Yeah, I was experimenting with ZFS failures last week. Pulling hot disks, breaking stripes, dd'ing random stuff offline. It impressively survived all my abuse.

If that were mdadm raid (or btrfs!) I would have been really nervous. But having worked with zfs enough I am now pretty confident that the authors took the hypocratic oath very seriously. "First do no harm."

I'm impressed with FreeBSD in general. I'm finding all sorts of things that are just more thought out.

iovctl and sesutil are two small things i was happy to discover

I think I must swap my NAS to a different motherboard and then see if the flaky behavior continues or stops. I am convinced the drives are okay. I already swapped the disk controller card for a different one. And this motherboard recently had a ram socket failure so I am already running at half ram and thinking that failure is more than just the socket now.

All I need now is another system or another motherboard. But life and time is what keeps everything from happening all at once.

yeah, sounds like time to abandon ship

unrelated, is the quarterly pkg set simply more stable, i.e., less churn, or does it get additional scrutiny (e.g., security, testing, etc.) versus latest?

Definitely less churn with quarterly. If there is a security issue then all of them should be getting an update.

If it is a minor issue then it might be skipped on the quarterly. Which is just fine with me.

Very volatile things like Firefox and Chromium seem to be on their own schedule and updating as upstream updates.

For my servers I run the quarterly. I am also running quarterly on *my* desktop. For your desktop, run whatever you feel like running, it's your desktop! :-)

coming from Arch ... so a week without updates is giving me anxiety! lol, just kidding. i'll probably leave it on quarterly. seems like everything i've installed is fairly recent, i.e., not like debian stable where things tend to be ancient

Debian Stable releases about every 2 years. That would have 8 quarterly releases in that time period.

But for example Firefox on quarterly seems to keep up with Firefox on Debian Unstable on my other system just fine. Often ahead of it.

Also remember that FreeBSD has a stable base core that is cohesive. It's released as a single base core release. All else is in the ports. (Ports become compiled into the binary pkgs.)

This allows ports to move asynchronous to the base core. But the system stability is really in the base core.

is there a way to search for all symlinks that point to a certain starting directory ?

aka, I want to find all symlinks in a directory that point to /dir1/blah

searching one dir, or a whole tree?

a whole tree

with find and sh

find ... -lname '/dir1/blah/*' might be what you want then

lname is not the name of the link but what it points to, right ?

right

let me try that - thank you

it does work :) thanks!

RhodiumToad: I think like I owe you a few hundred dollars (or more!) for all the advice you've given me lately