show ethtool on Linux can show TDR info to help diagnose hardware faults, ifconfig is purely configuration, I have scrubbed that man page a few times so I dont think there is anything there, as for netstat, there is a lot of stuff there, but as far as I am seeing right now (unless I am missing something) I dont see anything to debug hardware issues

specifically faulty cables

i know nfsv4 can be configured through an stunnel to encrypt transport and not have to bother with a full blown kerberos setup, but has anyone tried wrapping it with spiped

essentially the same concept as with the stunnel

dkeav: why not just use NFSv4 native TLS support?

will have to investigate if all clients can support it yet

re: cgit.freebsd.org/src/commit/?id=908…4f6af8a95d6c6ae65caed4371f96afb6409 does that mean 14.x runs slower now? or that software compiled on 14 is slower now? or?

looks like it's just build time which is slower?

kerneldove_: it means running clang is slower, nothing else is affected unless it links to libprivatellvm

and note that enabling the new option adds several hundred MB to the installed size of the system, there was another commit recently which reduces the overhead of the shared library build

rebuilding from 15.0 -> 15/stable now

15 is stable now?

warsoul: "stable" is the branch the next minor release is cut from, so stable/15 is what will become 15.1-RELEASE. some people run stable as a compromise between release (slow updates) and main (can be quite unstable)

ok ty

yeah, i usually track stable on systems i can; i am curious if I see vastly slower build times with the 15 series

so far i haven't noticed much really, freebsd build times have increased with each release over the years but when clang/llvm was introduced, that was the most significant I think

i believe the benchmarks that show it being slower but i can't say i actually noticed it until someone pointed it out

because of the chatter on this topic i ran some tests using the same hardware and here's a table I built - I will update it with 15.1 in probably about 1000 seconds :-) pastebin.com/m2gNjQC5

its less a test of speed of building and more of size of what to build per fbsd release though

funny 15.0 build time ;)

oh yeah, wow, bsd devs flexing :-)

flexing how?

I haven't followed how OpenZFS is tracked, but I don't reckon 2.4.0 will be backported to 14/stable, right? (not really a backport, since out of tree openzfs officially supports 14). I accidentally on purpose upgraded a root pool while I was on 15,

amigan: i think stable only gets patch releases of openzfs

it is an interesting topic, though.

in llvm's Cmake it is said that dynamic linking should be used by devs. so, it is you, then to decide how to build llvm. anyway, there is WITH_LLVM_LINK_STATIC_LIBRARIES that can be set in /etc/src.conf

...beg pardon -- bad connection. Trying via an SSH tunnel this time.

I have tested expr '/' : g on several Linuxes, and it completed without an error, not can I see from the documentation why this command should error out.

Also, even the simple ``expr / : a'' fails with a syntax error.

Looks like a bug in the FreeBSD expr.

hm, why would you be allowed to use / without an expression before it? what should that even mean?

nimaje: / is just a character. It constitutes an single-character expression.

expr takes three artuments: 1) string to be tested (/), 2) the operator (:), and 3) the regex to match (a).

expr / : a # a legal invocation as far I can see.

ah, expr does regex too for some reason…

nimaje: it /is/ supposed to do regex, with the : operator, yes.

well, freebsds implementation conforms to posix "After argument processing by the shell, expr is not required to be able to tell the difference between an operator and an operand except by the value." if there is an operator where an operand is expected, then expr is allowed to treat that as a syntax error instead of deciding that that operator should be a string operand instead

is file access slow or why does pkg take like forever for unpacking

nimaje: thanks for doing my homework and showing that the FreeBSD `expr' is actually the correct one.

yamada: re: slow unpacking> I had the same problem yesterday, on FreeBSD 14.3p5.

ant-y: well, both behaviours are ok, expr is just not required to disambiguate there

ant-y, i use freebsd 15

yamada: Anyway, in my case, extraction has taken about 20 times longer than downloading. It looked as if there was a lock on a file, becase the other files (of comparable size) were extracted quickly.

nimaje: re: expr> the FreeBSD version does the least thing required, whereas the Linux version supererogates, and reliance on those extra features causes grief if one is unaware they are not guranteed.

for most things expr does there is a good replacement, but no idea what the replacement for the regex matching would be or even if something exists

/join #freebsd-ops

What is a simple way to have connections to gmane.io:119 forwarded to localhost:6001 ? Direct NNTP access over mobile internet is very poor here, but it work OK over an SSH tunnel.

(Save for IRC, by the way)

I want to forward those connection on the OS level, to avoid adapting all my confuguraiton files with IRC and NNTP servers.

(Same /for/ IRC)

I'm dreaming up nonsense involving ipfw and nat, but it's not specific enough to tell you exactly what

You would also want to create some gender of VPN over your SSH tunnel.

LXGHTNXNG: sounds hard.

No VPN needed so far, to forward plain-text connections, at least.

how does systat -ifstat get the load of a nic? is there a cmd we can run or a sysctl we can probe?

ant-x, simple socket forwarding using ssh -R (or -L depending on who initiates the handshake) should do the trick? There's also ssh -D if you just want a socks5 proxy to your lan or something

sorry if you are not the same person as ant-y

ant-y: expr on bsd and I think illumos/solaris is treated as a division operator. I think you need to escape it; coreutils expr treats it as a string literal

err, sorry, i see you already got your question answered

what is the point of making pi images when they use FAT parameters that are known to work on pi firmware

then need to use FAT params that are known to work

I mean pi is to blame....

hrm? isn't the point just to work on a Pi?

hodapp: yeah, but the pi does its own weird thing and doesn't read it correctly so it doesn't boot until you manually change some things.

you have to change the FAT boot partition metadata

at least on this pi 4B model

Noticed an interesting behavior - when using multiple jails on the same IP address and restarting them (ala `service jail restart`), the IP is unbound from the host interface when the _first_ jail is shut down, which can cause software on subsequent jails using that IP (looking at you Unifi) to take forever to continue to shut down

And since there doesn't seem to be a deterministic order to shut them down you can't just configure it to do the "problematic" jails first

futune, using ssh for that purpose seems strage, because I need to forward connections to localhost (my machine) to a third-party server (e.g. an NNTP server). There is not way to establish an SSH connection between them.

ant-x, oh, I assumed you controlled both parties. My bad, sorry.

I think in particular it causes issues with certain software that expects to send "I'm going away!" messages to client devices when it's shutting down

futune, pro noblem. My situation is that nntp:// irc:// and even svn:// works poorly or not at all from the mobile internet that I have, but they miraculously amend themselves is roiuted through an SSH tunnel (via an intemediate pubnix server). In order, however, to use it I have to modify all of my portable config files, because, instead of connectig to irc.trololo.org:119 I have to direct my IRC client to localhost:6001 (the source end of an SSH tunnel).

silly question, how do I get the freebsd sources using freebsd-update?

I'm tinkering with an ancient arm6 system, so I'm hoping for a tgz

zip: if you just want a tar file, download it from download.freebsd.org/releases/amd64/15.0-RELEASE

appreciated

going so far as using git seems like overkill for this wee gadget

once you have it installed, i think freebsd-update should update it automatically, you might want to check the config file to be sure

it should do

perhaps it's silly but I was thinking of rebuilding the kernel to see if I could persuade it to show boot output on the USB CDC interface. Which probably isn't really possible

allegedly it just works™ if you load the drivers in the loader, but I suspect that's for systems which don't already have a serial port specified

Once again the lesson is that the pi zero is a piece of shit and I should throw it in the bin before I'm tempted to try anything with it

I'm pretty sure it just browns out and that's why it's so glitchy

At least with an arduino you can pull a pin up and it turns itself off and you can use it for its USB serial chip

if anyone was curious about the spiped/nfsv4 thing, it does indeed work

easy peasy really

dkeav: if you have NFSv4.0 clients (e.g., macOS), you may have issues with delegation callbacks not working as they are out-of-band, but i'm not sure macOS even supports that to begin with

(just don't enable issue_delegations in that case, it's disabled by default)

no apple stuff here, mostly just freebsd and linux

well no, only freebsd and linux heh

well i just installed freebsd but im using icewm as my window manaager

idk why i stopped using arch to install freebsd with ice but i did

and now im here

Welcome?