My apologies if I'm asking this in the wrong place, what would be the right ftp server to find source code tarballs for old, legacy versions of FreeBSD for research purposes?

I've looked through ftp-archive.freebsd.org, but I had serious trouble downloading the source code for the version I'm looking for (5.2)

rwp: thank you! "fsck -f -y" did the trick. it deleted /tmp, i made a new /tmp, and everything was back to normal

but you were right. this happened: bpa.st/UI7A

is this indicitive of hardware failure?

That certainly looks pretty much like a disk failure to me.

Regarding fsck I am guessing the -f was required because the disk was not marked as "dirty" and therefore the fsck was actually skipped in that case unless -f was given. So even though you ran fsck against the disk it didn't actually do anything without the -f.

Failures like this are why if at all possible I always use raid/mirroring on machines. Because then when things fail the system continues to function and the failing device can be replaced and recovery from backup is not needed.

Laptops and Raspbery Pis and mini-PCs are not conducive to having two devices for redundancy but desktops and anything I can put two storage devices in I always put two in and then mirror them.

rwp: i was thinking about getting a replacement and in single user mode, using dd to clone the disk

Hopefully that will read successfully and you will get your current data off the drive.

If there are problems then the GNU ddrescue tool is really about as good of a utility as it gets to do everything possible to extract as much raw data from the drive as possible.

Is it too much to expect that your data is already safely backed up somewhere? Following the Backup Rule of 3-2-1 with at least 3 copies of the data, on at least 2 different types of backup, with at least 1 copy off site?

hi quick question. I just installed Fbsd and I encrypt it my home with zfs. Now when I'm log in into my system. it says I cannot execute or create files in my home dir?

I see a question mark at the end. But that is more of a statement than a question.

ls

Whoops!

I have a couple of ZFS filesystems I don't seem to be able to unmount. I get "device busy". Nothing from the FS shows up in fstab. How can I unmount/diagnose?

I mean in fstat, not fstab. Fingers think they know how to type.

`zfs unmount -f /j/s/web` gets me a "cannot unmount '/j/s/web': pool or dataset is busy"

Do I just reboot?

if gh00p were here, i would suggest checking for nullfs mounts

fstat will tell you what files are still open

cannot mount 'zroot/home/': encryption key not loaded

how do. I load up the key?

Anyone have experience with thin jails on a loopback device, and ipv6?

I'm using Bastille, and there's a cloned lo1 == bastille0 device on which the ipv4 aliases for the jails live. Not sure how to configure ipv6 though.

Hello, all. I have filed an issue about the dangerous behavior of `reboot' that we discussed here on the 23rd of July: <bugs.freebsd.org/bugzilla/show_bug.cgi?id=280941>. Everybody is welcome to suggest ways to fix it. Notify: Grabunhold .

ant-x: Thank you

YW, VK. better late than never.

On a loopback where ipv4 aliases live for jails, could I assign fe80::x/128 addresses to a jail?

ant-x: nice :)

ant-x: i'll add my experience today

Your experience is very educating.

i'd suggest "many users (and especially those coming from Linux)" read the documentation, and be done with it

ant-x: i've tried my best in summing it up

I know someone who refers to every *x as "Linux"

yuripv: i encourage you to read my comment on that bug report for my opinion on why that doesn't fix the problem (comment 2)

I mean, at luser level, if you can use one modern *x, you can use 'em all with the possible exception of Solaris

while reading documentation is certainly always a good thing, it's very hard to read all and every documentation that may or may not be relevant. is there any reason not to add at least a hint to the "reboot" command?

yuripv, What about the good ole principle of least astonishment? Astonishing behavior does not become any less so if it is documented.

Licca, is it good or bad, that so many *x are so similar and have so little indifiduality?

Grabunhold, you wrote it well and in good English. Is it your native language?

ant-x: well, I guess compatibility is a good thing, but i'm also all for users at least being educated to the point of being able to properly name the OS they're using ;)

Things become more confusing when the OSes are more similar than their names :-)

ant-x, A little A, a little B...

ant-x: no, my native is german. maybe i'm at an advantage because english and german are not that different. but i also took a lot more time pondering that sentences than the stuff i blurt out on irc here, haha

Grabunhold, but when will that `reboot' hint be shown? During an artificially added pause before the reboot?

I am not a fan of GNU - feel their software is slow and bloated, and GPL fanboys are annoying

Licca, I like GNU Troff very much, though. And I prever SVN to Git (don't throw bricks at me).

ant-x: when typing in "reboot", i'd think?

once in a while, GNU does have the best version of something

immediately?

though, I didn't think Subversion was GNU?

Grabunhold, you don't mean the hint will come from the shell as the user types the command? It must come from `reboot' itself, whem the user has typed it in and pressed enter.

at least via typing it in via ssh that would probably suffice. i didn't think about typing it in directly on the machine

Licca, GNU Savannah is for SVN. It is their main version-control system, I think.

ant-x: of course, printed directly by the reboot binary

git comes from Linus Torvalds and github is owned by Microsoft lol

ant-x: and you are suggesting breaking the POLA because (ex)linux users are suffering :)

The ugnly and bloated thing called Github is hopefully not endorsed or affiliated with the Git project

what irks me working on a Dell Unix VM is that you have to type "shutdown -y -g0" instead of "shutdown -h now"

though I think it does allow "/usr/ucb/shutdown -h now"

yuripv, no: because the current behavior is illogical and inefficient. The frequent commands should be the shortest and simplest, therefore `reboot' instead of `shutdown -r now'.

Grabunhold, the user that typed `reboot' and pressed Enter is not guarrateen to the notice message, for the reboot may be fast...

I've seen systems where there's a default standby of 30 seconds for reboot and halt

ant-x: as I said, i only thought about the "it's typed in via ssh" usecase where the message would linger on the remote terminal

Licca, if that standby has other uses than displaying a usage hint, then it is probably OK.

generally it's to allow takesy-backsies

Grabunhold, but then reboot has no idea of how it is invoked, and many people use it directly on their PCs (I think).

Licca, rather than "Hold on to your false teeth, genglemen"?

/sbin/reboot would be able to detect if its controlling tty is a console, an xterm or a remote, not?

using the same code that tty(1) uses

Licca, able -- maybe. But should it have use that dependency instead of just doing its one thing?

libc?

I mean a conceptual dependency, awareness of the environment and its types, to modifyt the behavior in a modal manner.

I would argue that it shouldn't matter. some unices have heavy requirements, like for example, you have to run it from a console with pwd='/'

(this is often true on System V, in my experience)

I'm not sure if baseline SVR4 has those requirements (*cough*Icancheck*cough*) but I know Dell's complains if I'm not in / when I type shutdown

ant-x: is there a reason that "reboot" has to know if it's invoked using ssh or not? it should print the message unconditionally, in my opinion

Grabunhold, only to make sure the user has time to see it.

How frequent and standard are situations where one does need `reboot' instead of `shutdown -r now'?

Is it not too good a verb to be reserved for rare a task?

I believe that if a program should try to avoid dependencies on environemnt and higher-level things, if at all possible. It its behavior is determined only by the its command-line parameters, it is easier both to maintain and to use. Of course, they may be well-justified exceptions.

one could argue that rebooting and shutting down are _so serious_ that not only should commands like "reboot" and "halt" not exist, "shutdown" shouldn't be in the $PATH at all, even as root

(I think that's daft, but I have seen that)

Licca, that's possbible. PC users can define aliases.

ant-x: well, i guess adding time for the user to see it as a good thing. but atm we have nothing at all, and I feel this discussion getting into the weeds

a simple print to std(out or err) is very low-barrier and would at least be something

maybe you'd see it flashing by on the real console, and via ssh it will linger in the remote terminal

of course i'm still in favor of actually changing the meaning of "reboot", but i'll take what I can get

Grabunhold, it might help if you made it clean in the issue that you too prefer the breaking change of the `reboot' behavior...

s/clean/clear

ant-x: done

Yeah, I have already read it. Thanks.

(I recevied a notification by e-mail)

ant-x: thanks for staying behind this and writing that bug report

No problem. Things happened in my life that made me stop my FreeBSD-related activities. I still cannot cause `lighdm' to show the "Power off" and "Restart" menues, use a raster font, or a vector font with good hinting as in Windows (font-hinting is one things they did good), &c.

for me, FreeBSD is a pure server OS at the moment. my desktops and laptops all run some flavor of linux. i might have even typed "reboot" again when i meant proper shutdown on one of the FreeBSD machines simply out of muscle memory.

i can't even remember, too many reboot commands

i'm in a little bit of shock, too. i've been doing it wrong for years.

I could prolly use FreeBSD and Debian Linux more or less interchangeably, never mind Debian Linux and Debian kFreeBSD when that was a thing

There are hundreds of Linux, but only one true original *BSD, which is why I hope to find it easier to stick with.

all the BSD freenixes have a common origin but they're more specialized than Linux distros

that said

it's all Jolix, ultimately

NetBSD and FreeBSD are the only OSes I know that are direct forks of Jolix

less churn is actually one of the reasons i'm using FreeBSD, so i totally get the argument not to change the meaning of "reboot". but in this case, i'd argue it's worth the breakage. but that's just me of course.

I think if it weren't for two people in particular, we wouldn't have any freenixes - not even Linux. and it's not Linus or RMS

it's Bill Jolitz and Keith Bostic

all the dockers and podmans and kubernetes and whatnots have way too many moving parts and evolve way too fast, i just want my fileserver and a few other things. i'm hoping FreeBSD + jails + Ansible will give me some peace of mind

+ ZFS of course! zfs ain't no fun on Linux

I'm old-fashioned

Licca: i knew about jolitz, but never heard of bostic. will have to read up on him

I've been wanting to create a "traditional" Unix clone, and I was like, well, I can write some of it... but I'd need to pinch the rest from the BSDs :/

Bostic drove the project to Theseus the AT&T code out of BSD

enabling a fully open-source BSD

ah, of berkley db fame, too

I mean, my bias is toward NetBSD as the more direct descendant, but it's also a fair argument that NetBSD and FreeBSD, since they both built off the same codebase, are ultimately equals

where OpenBSD is more "I'll start my own BSD! With blackjack and hookers!"

that just happened to find its niche

in which case it's more similar to the later Drag*cough*

hi, trying to reconfigure an Intel iwn 5100 in FreeBSD 14, ir worked before but now no longer can do ifconfig wlan0 up scan (iwn0: iwn_read_firmware: ucode rev=0x08530501 iwn0: scan timeout

I redid the steps from the handbook and now it works

i happen to agree with the breakage being worthwhile. adding a flag reboot -fast or something to restore existing behavior would make things much more intuitive going forward IMO

-q, maybe?

uskerine, You were not in the channel to chat at the time but on Sunday I noted that ffmpeg, firefox, and ungoogled-chromium finished building and I was offered an upgrade for them. Meaning that they were once again installable.

I see I missed a reboot discussion but I will comment that when I worked on HP-UX it annoyed me that HP-UX reboot -q was the equiv of Linux reboot -f and that it was yet another point of portability differences.

That said when people read that they should reboot their machine they should typing in "shutdown -r now" rather than reboot. That's an unfortunate counter-intuitive thing.

the main exemplar of SVR4 I've got is Dell Unix

I've tried and failed to set up a cross-compiler for a known SVR3.2 for a project lol

I can make builds of my Strix tools for FreeBSD, OpenBSD, NetBSD, Linux and SVR4 lol

(I mean...why...but still)

Strix is still useful on a "barebones" SVR4, but who's using that in 2024?

for example, it brings some tools, like grep and rm, up to Posix

hm

SVR4's rm only knows -rif, Solaris adds -R, Net/2 _also_ adds -d, 4.4BSD-Lite adds -P

I don't support -P, but I support -vx, both claiming FreeBSD and NetBSD (plus OpenBSD for -v)

and I also support -I from GNU

looks like I'm basically in line with current Posix lol

Licca: both NetBSD and FreeBSD are direct descendants; the major difference is that after the lack of a VCS for 386BSD was identified as a problem, NetBSD started at the first version of 386BSD, while FreeBSD started with the last version.

Both arguments had merit to them

"VCS" ?

I know NetBSD and FreeBSD had different priorities

version control system

(at least I'm assuming there isn't another vcs acronym)

though my understanding was more that NetBSD wanted to continue the path BSD was already on and FreeBSD wanted to focus on commodity hardware (read: x86 PCs)

there's "Video Computer System" (i.e., Atari 2600)

I figured someone might be able to work out the initialism from context :)

Licca: project philosophy was also different, sure - but they both share a common ancestor, is the point.

had x86 turned out to not be the platform du jour, i'm sure freebsd would've pivoted to something else (quicker than ppc in 6.0 according to arch(7), i mean)

the focus was very much on a high-performance unix-like on x86 in the early days, though

well, you could argue that AMD64 is different from x86 if you wanted to really split hairs

also, things like alpha support aren't mentioned in arch(7)

I wouldn't say common *ancestor* so much as common *direct ancestor*

i ran an absolute shedload of jails on an EV7

because DragonflyBSD and OpenBSD also have Jolix as a common ancestor but they're not forks of Jolix

dragonflybsd forked from freebsd and openbsd forked from netbsd, though

right

(assuming that with enough age, "DragonflyBSD" doesn't remain a voldemort around here)

ppc, sparc64, ia64, arm(v4/v5), and alpha have all come and gone

so far as i know, dragonflybsd is happily chugging along doing its own thing

I meant that due to bad blood Dragonfly might be The BSD That Shall Not Be Named®™©

if you know what I mean

that water's so far under the bridge at that point, i'm not even sure you can hear it

hence me choking off the name earlier

lo0l

i can't imagine anyone would have the energy to sustain a grudge like that

oh, I can very well imagine

but then, I'm used to being around people who simply cannot give up a grudge, ever

and I've been attacked by such people

no future in that

these are the type who don't think about the future

no future in that, either

I prefer to just do my issh and be done with it, but I've often been focused on self-defence

in some sections of the Internet, I've become kinda infamous lol

because I had some expertise in audio work and synching anime, a friend got me broadcast audio for Dragon Ball Z. I made encodes with it, killing a golden goose a few people were relying on

you'd think a lot of the comments on a certain BT tracker would be appreciative of me leaking it but instead it was mostly people ticked off at me because they couldn't sell their pirate audio files anymore

maybe conversations about committing copyright crimes aren't the most appropriate for something that putitively exists in and under a copyright framework? :)

:P

my attitude toward copyright is sort-of mixed

"sue me" kek

I'm not gonna step on the little guy or on some open-source project. but when it comes to big companies or old unobtanium, I really don't care

I was a bit more open to messing around with the MS-DOS 6 leak, until Microsoft and IBM released 4 - then I backed off because I want to encourage them to release 6

because honestly, there's not THAT much difference from 4 to 6 - just that the kernel and shell enhancements are important enough

if they were to release MS-DOS 6.21 code, I'd be on it like white on rice and not only that, someone's trying to get 3.3 and if they get that I'd make an open 3.31

somewhat more related, about 25 years ago I thought of making a Unix-like environment for 16-bit DOS - there's a couple already, to be fair - and I started from the BSDs arbitrarily. of course at that time I was a C n00b

kevans: was just thinking this.

and it's going on for days

"#FreeBSD",

Whoopsie.

yuripv: it has? My apologies, I usually rely on people to poke me in #freebsd-ops as necessary

Licca: are you no more C n00b

rwp: yes I saw it, I did not know if it was saturday or sunday but I remember I checked the day after and it was fixed

I haven't looked at the other package status yet. Let's hope for the best for it.

I read again about the GPU accelration status in FreeBSD, I am wondering if there will ever be light at the end of the tunnel. The closest thing to use GPUs for computing is OpenCL with the Mesa runtime in a relatively old Radeon, and even in that case there are people reporting random issues (off topic)

hi...did anyone have success to install Haiku in byhive, please?

on 14.1 apache 2.4.62 core dumps on graceful but works fine when starting manually: pid 67035 (httpd), jid 0, uid 0: exited on signal 11 (core dumped)

is this a known issue ?

last1: If using mod_php8*, yes. It's been a problem for a long time from what I recall.

Solution: don't use mod_php and use proxy_fcgi with php-fpm.

thumbs: That's been the consensus. I migrated away from Apache+mod_php in favor of Nginx+PHP-FPM a long time ago. I believe the graceful core dump was the beginning of me looking for an alternative.

last1: forums.freebsd.org/threads/apache24-graceful-restart.94544

wow, fresh one too

ek: Or httpd + proxy_fcgi + php-fpm

Yes, the mod_php DSO has a tiny speed advantage; it also segfaults a lot.

last1: And, of course, this one: bugs.freebsd.org/bugzilla/show_bug.cgi?id=267435

Nearly two years old and still going strong.

Using mod_php was a bad idea 12 years ago.

anyone knows why FreeBSD dropped usage of pecl-geoip ? freshports.org/net/pecl-geoip/Makefile

the alternative is much worse

I thought MaxMind stopped producing a free version? No?

you can still have a free account

2019-02-15 net/GeoIP: Legacy databases no longer available. Switch to net/libmaxminddb instead

and pecl-geoip seems to be abandonware anyway

i can count the num of files in a dir recursively but how can i get the total line count of all text files recursively?

polyex: well, file(1) says, "The type printed will usually contain one of the words text... executable... or data"

output looks like

/etc/security: directory

/etc/rc: POSIX shell script, ASCII text executable

so it's colon separated

why doesn't find . -type text limit results to only text files then?

oh, "text file" isn't one of the types for -type

"regular file" is, i.e. not a directory or device file or socket

-type f finds those; but it does not search their contents to see if they are texty

find only stats the directory entries, and lets you query the data there

find . -type f -exec wc -l {} +

find . -type f -exec cat {} + | wc -l

omg nice ty!

oh it doesn't filter for only text files