Dooshki: "Warning: exporting /test_export exports entire / file system" means that /test_export is not a mountpoint, therefore an NFS client that can mount /test_export can access the entire parent filesystem (i.e., the root filesystem in this case)

ha! pf does not need to have a rule to allow (openBSD) DHCPD since it only listens on raw ip and also does not show on sockstat -l

gustik: i think that's true of most/all DHCP servers since you can't easily receive and send broadcast UDP in the way DHCP needs. you may need a pf rule if you're doing dhcp relay though...

I was wondering how it is possible that my wifi clients got IP addr configured over DHCP despite me not having a rule included yet.

ok

so it is not just me, it's generally the case ... great

the rule helps anyway, I added it later in order to track states

well, i'm not sure about ISC DHCP but Kea also works this way

DHCP works with or without it, but putting a rule for pf helps to see what goes on, if packet numbers rise etc

I had ISC DHCP before and I can not remember if it needed a rule

in my old configuration a rule was included in the pf.conf

but I do not remember if it would have worked without

this is a lot more sensible with DHCPv6 since it uses link local and multicast addresses and doesn't rely on broadcast traffic

what was frightening to me is that the firewall(pf in this case) did just pass trough packets despite having block drop in rule as gernal rule

yes, DHCPv6 needed a rule, of course

that was not working without, I am using DHCPCD for that

btw. FreeBSD should include some DHCPv6 client in the base, since I have reinstalled my router (PCengines) where my old SSD (msata) was having issues with write and random read errors, so I figured the drive is dying

I did not replace it because a PCengines is 10 years old, so I did a BIOS (firmware) update and got SD card slot working and have it now on SD card, which did not cost me anything, because I had one 128 GB spare, which is 2x the size of the old mSATA SSD (64GB)

long story short, it works, but I needed to put in my smartphone over USB tethering, which worked fine, in order to `pkg install dhcpcd` so that I could get my /56 IPv6 prefix over PPPoE and so connect to the internet

gustik: work is ongoing to replace dhclient with dhcpcp which supports dhcpv6

I have a DS_lite connection, so there is no IPv4 per default

yes, great

er, dhcpcd

for the time being, my smartphone saved the day

yes, I am totally dependent on dhcpcd, since my internet works like this: 1.) Get PPPoE connection working with IPv6 linklocal 2.) DHCPv6 getting /56 prefix and AFTR name (along with DNS servers IPv6only) - HERE there is IPv6 connectivity already 3.) create a GIF tunnel over IPv6 to pass IPv4 carrierNAT in - FINALLY there is also IPv4 connectivity

i think you are the first person i've ever met who actually uses DS-lite, it seems like 464XLAT has mostly replaced it before it got any real traction

the last part I am doing with dhcpcd hook, where I take the $new_dhcpv6_aftr_name and create the `ifconfig gif1 inet6 tunnel someaddr aftraddr` etc, then just defaultroute to something like 192.168.xxx and done is it

yes, I am using it since 2014

do you mind if i ask where you live (US, Japan, ...)? only because i'm curious who uses this

in germany I had DOCSIS cable modem with real dualstack, which was more fun, because on DOCSIS you have real ethernet and you can use ISC DHCPv6 client etc. BUT NOT ON VDSL2 ... like here, because there is PPPoE and ISC DHCPv6 did not support that back then (there were patches) I do not know if that changed last 10 years, I did not check

however, dhcpcd was the only thing that could do DHCPv6 PD over PPPoE

I am here on Orange Slovakia VDSL2

Orange being French telecom

ah interesting

so I suppose the french will be using this method too, since here in Slovakia I am the only one who actually has a clue how it works

Orange provides their routers which do this, but they are not configuring them ... LOL

other providers in Slovakia do not have a clue about IPv6

however, I seen that since 2016 Deutsche Telekom in Germany provides IPv6 on VDSL2 too

so they may be having the same kind of configuration - with PPPoE and DHCPv6

when i was using VDSL2 (and ADSL2+, for that matter) i just got a static route from my ISP so i never had to mess with this stuff

what surprised me lately is that I wanted to use my backup router which is a Mikrotik RB750GL with OpenWRT on it, it failed to configure my network here

which was shocking because OpenWRT used to do it well, however, upon inspecting tcpdump I seen that odhcpc6 or how is it called on OpenWRT, it did not even ask for a prefix, despite being configured properly

ivy: where are you from?

gustik: UK

ok

I had in Germany an MNet connection which did IPv6

also ADSL2+

but that was real dualstack ... but I can not remember any more (because that was like 2012) ... how I got the prefix

KabelDeutschland/Vodafone on their DOCSIS gave out only one /64 so I suppose there was the same

2012 was not a year of prefix delegations yet

nowadays that Vodafone/KD /64 would be a great setup to try NPT66 on it

however, speaking of IPv6 I noticed some weird stuff while I kept my connection IPv6only

everything that's IPv6 worked, like google, youtube(of course) etc, bing etc, all the rest ... but not wikipedia and some niche websites

and I was investigating the issue, first I was thinking my local-unbound is messed up, but then I noticed that dig +trace shows the same issue

the reason for misbehaviour of SOME websites was that it really could not resolve their IPv6 since some of the upper level DNS servers are IPv4 only LOL

so even though they were reachable over IPv6, only as long as you had a DNS resolver that had IPv4 connectivity LOL

20 years of me using IPv6 and we have still islands of IPv6 connectivity !

ok

I reconnected using IPv6

still somehow strange that it connected me to virginia and first over ipv4 ... hmm

Hello, I manager to create a ssh tunnel from Freebsd client to Linux server, But I cannot put all traffic via tun0, and when I ran this command `route change default 10.0.0.1` (10.0.0.1 is the ip of server, where I can ping and even open ssh session) the internet was gone

s/manager/managed

Did you try to route the tunnel through itself?

@vkarlsen, yes route change default 10.0.0.2 , but I did not have internet after

@vkarlsen, when I changed to the default gateway, everything internet back and I could ping to server 10.0.0.1

Hi. Is this the right channel to ask novice questions about FreeBSD.

I am new to FreeBSD – sort of, haven't used it since 2004 or so.

I've installed FreeBSD 14.2 from a USB that had the FreeBSD-14.2-RELEASE-amd64-memstick.img and somehow I failed to install 'pkg' and any packages. There is no ethernet or wifi working.

So a question: I assume it's possible to install 'pkg' from the USB pen, mounted again? I can see its contents after mounting it under /media/pen

perhaps I can redirect PACKAGESITE to the usb pen, so it can find binaries there?

has anyone tried qubsd?(github.com/BawdyAnarchist/quBSD). I'm frustrated with many aspects of qubes os and curious about this thing

I daily drove freebsd a couple years in the past and it looks like this just uses freebsd stuff I'm already familiar with, so should be less confusing to interact with than qubes maybe. but it still seems somewhat incomplete so I'm curious if anyone's tried it at all

ivy: Interesting, that's really good to know, and yeah, I confirmed that's what it does (thankfully I didn't share my root, other than for that short test, but a directory meant for sharing)

jmnbtslsQE: And thanks a lot again! With the following two lines in exports, it now works perfectly for me, even with NFSv4-only mode enabled:

V4: /shared -sec=krb5p -network=192.168.100.0/24

whoops, sorry about potentially duplicate messages, my client might be glitching out

Oh, it interpreted it as an IRC command

/shared -sec=krb5p -network=192.168.100.0/24

No longer getting any warnings from mountd either, everything's finally working 100% perfectly

hey guys how do i start contributing to FreeBSD? , i have used it but i found problems from Desktop point of view like my screen being laggy on a DE and sound not working after serveral configurations , because i ran it on my laptop which is from 2021 and since FreeBSD is not made for desktops how do i get started?

rudynotfound: write code, share it publicly, advocate - that's how you can contribute

mzar which type of code?

is there anything that can be improved?

hmm.. any type you can

of course

i dont really have an idea

OK, so look at the mailing lists, maybe you will get some impressions

not made for desktops?

perhaps not designed with desktop in mind, but works perfectly on desktop

rubin55: please don't hesitate to use FreeBSD on desktop

ha... rudynotfound left

bummer. i just dug up this link for them freebsd.org/projects

that's ... 5 seconds of my life that I will never get back

it looks like some people want to contribute not for the contribution but for calling themselves contractors

s/contractors/contributors ;p

Might also have to do with people being stuck in tutorial hell and not knowing how to get out of it, so they "try to contribute"

what's better program design. to accept multiple options (cmd -f foo -f bar) or to accept multiple values per option? (cmd -f foo,bar)

demido: imo the former is a clear win in most cases; less fragile parsing, more obvious what's happening

ya i agree ok great ty!

demido: i think it depends on what the option does, but i might lean towards the former as well since it makes things easier to script

one example of where a single option makes sense is pw usermod -G, since the value is a single value that happens to consist of multiple elements, using multiple -G options there would reduce functionality, or require a second option

hey everyone, I'm relatively new to FreeBSD (haven't used it in more than a decade). I am currently tracking 2025Q1 of the ports tree and noticed that lang/dotnet8 is broken there but not in main (due the LLVM default version changing). I've manually replicated the fix in 2025Q1 and it's fine now. What's the process to get the port maintainer to apply the fix to 2025Q1 or can I cherrypick their

commit and submit a PR? This manual fix won't hold everytime I update and rebuild the packages in poudriere. Thanks in advance!

jnb: i prefer to keep my own local ports branchs to let me easily fix things like this myself. as far as fixing it upstream, i would be inclined to cherry pick the fix and submit it as a git am in a PR

ivy: never thought about having my own local branches before, great suggestion! I'll also cherrypick and submit a PR upstream. thx for your help!

Are you people able to see emojis rendered correctly in firefox ?

mine sometimes display as that square with hex numbers. im not sure what is missing so firefox can render them correctly

Sounds like a missing font to me