I actually can't get crash dumps to work

First thing I always did was recompile all the things without pulse audio

it just freezes after "dumped 256 of 2048 MB"

Sound was always crackly and crappy so I would just remove support for it and sound was not crackly

ring0_starr: how big is your swap

4096 MB

Are you sure it freezes and not doing other things ?

how can you tell the difference? it's sitting there with no change for like 45 minutes before i power cycled it

Damn I would of power cycles after 5 min heh

this is apparently working on the first try, which is suspicious

hodapp: Kinda sounds like FreeBSD.

Don't be suspicious. It's quite simple.

TIL you can have an inet address on each side of an epair.

via freebsdfoundation.org/wp-content/up…s/2020/03/Jail-vnet-by-Examples.pdf

Although there are a couple typos in it.

satanist: Just as an update, I feel like this isn't quite all together yet, but ditching the bridge altogether and just having a set of epairs on a private network, I'm able to have my desired destination IP in the jail, and then use the private IP inside the jail as a gateway for the floating IP, so essentially doing what you suggested and *just* using routing. I'm still not altogether clear on why I

can't get it to work as it would if all the addresses were in the same block, but I can poke at that.

Hm, and if I stick with this I'm also going to have to throw in some NAT, so I'll keep exploring.

mason: nice to hear that it's working

satanist: Well. Sort of working. I've not yet figured out how to get the floating IP to be able to send traffic through the host directly, so now I'm struggling with NAT. Evidently kernel NAT is broken with TSO, so that was an exciting adventure. Now I'm trying to figure out how to get snat working for just the one internal address with natd.

did I understand you correctly: you get one ip-address from your provider (your floating ip) and a complete /24 network for your own propose?

Not quite a diagram but broken up a bit: bpa.st/JDPQ

No whole /24 to myself, although I had one years ago.

No, wait, it was a /16 years ago.

in the 90s

I'm basically looking for the ipfw or natd equivalent of: iptables -t nat -A POSTROUTING -o $EXT -s $INTNET -d 0/0 -j SNAT --to-source $EXTIP

I mean, ideally I'd find a way to not use a private network as a sort of bridge, but I'm not succeeding in getting traffic to flow if I have unrelated networks on each side of an epair, even with routing explicated.

Anyway, it's past 3AM here and I need to hit the sack, but I'll be happy to glean any ideas that occur to anyone while I sleep, when I check back in the further-AM.

ek: it's more the feeling that somewhere in my 'cp -a'-ing things individually I *must* have fux0red something up

so far the only 'problem' (which isn't really one) I've noticed is that 'zpool status' listed the pool by device name and not disk ID, which looks like just something I can change in the bootloader

and samba complaining about /var/log because I'd not copied that tree over yet

why you want to use zfs hodapp

ndut7: my NAS was already on it for 10+ years, and really I just wanted the benefits of things like BEs and ability to very easily back up a live system with send/recv

all you need is just nfs

and the rest

zfs and the rest is not much use anyway

unless you know what your doing with it

and have a lot of storage

just mount it to your current storage or if your use windows just samba then

zfs or other system is just achoice don't force it

doesn't mean ntfs and the rest is bad too

uh, NFS to... what?

How is the state of audio in freebsd? i mean is it possible to do audio production like you do in linux via jack?

I just want to say thanks to [tj] for introducing me to tshark

which allows me to craft such gems as

tshark -r bridge.pcap -Y "ip.flags.df == 1 || tcp.options.mss_val" -T fields -e frame.number -e ip.src -e ip.dst -e ip.len -e ip.flags.df -e tcp.options.mss_val -E header=y -E occurrence=f | column -t |sort -n -k 4 |grep -v 1500 | tail -10

ape_din: good. jack is available in FreeBSD too. check out m.youtube.com/watch?v=fNALnFIfenM and anything on Goran's blog meka.rs

yeah i know this guy, also foudn this other one: youtube.com/watch?v=MvJD476tNN0

and also more recently there is a lot of work going into audio from christos@

see freebsdfoundation.org/project/freebsd-audio-stack-improvements & a bunch of other status updates he's been making

(i can't see yt on my desktop so these urls are all blank for me)

its practically christos

haha

anyway i just have to understand how it works and do lot of reading i guess. and see also if my latop will handle it

dch: Goran is well known in freebsd circles?

I only met them because I bumped into them on XMPP, never seen them in person though

polarian: his picture's here, just he's only even more enthusiastic in real life :-0

erm :-)

so he is well known then?

dch: yeah, I learned about tshark way too late in life; if I had $1 for every time I'd said "Damn, I wish I could use wireshark on this headless machine" before that, I could at least buy a decent Ampere box for my own usage

kevans: dch's rule#1 of troubleshooting, you always wish you'd reached for *shark earlier than when you did

then I finally for unrelated reasons checked `pkg info -l wireshark` and .oO(Huh, what's tshark? oh you sonofa *bangs head on desk*)

I do think subpackages wireshark-core and wireshark would be a good end-goal, though, if you can pull in tshark without needing to drag in a whole bunch of qt stuff at least

LUA=off: Lua scripting language support

that's depressing, too

whaa?

Hey all. I have a weird case. I have a FreeBSD 13.2 machine which I want to upgrade to 14.2, but, I want to do it with FreeBSD-rustdate, because it's a slow machine and FreeBSD-update takes a long while :-) I tried just cp'ing the binary, and I got the following: "Shared object "libssl.so.30" not found, required by "freebsd-rustdate""" I'm wondering if I can find that SO from a package maybe? or better thoughts?

we <3 lua in these parts, and it'd be surprising if that's really a heavy enough dependency to care enough to turn it off

kevans agreed. I really like wireshark, but I am mostly in a console. somehow that was my main "reason" to learn tcpdump, that I even forgot that wireguard exists at this point.

antranigv: sounds like you're playing with fire

yeah, I'd figure that Lua - especially embedded Lua - brings no other dependencies

kevans or what I like to call as "Monday" :P

antranigv: how did you obtain freebsd-rustdate? presumably didn't build it yourself?

kevans there's pkg install rustdate as an option (from newer version), I can also build myself I guess, but I remember there was a bug that I fixed on 12.X as well. lemme check my journals

antranigv: right, and the official packages will be built against 13.5, so even if you obtain that .so you could find other problems lurking due to the provenance

(or is 13.4 still technically supported?)

yeah, sorry: built against 13.4 at the moment, which is still supported until probably the end of next month or so?

release_4 exists. I am on 13.2. I might even have 13.1

might be better to just build indeed

i think that's a job search platform *rurns away*

lol

haven't logged into indeed since like 2016 and I *still* get spam from them

I'd love advice if anyone has it. I need a minimal NAT for a machine to take packets that come in from an interface and send them out through SNAT? Looks straightforward via kernel NAT and ipfw, but evidently there's a bug and kernel NAT won't work in the presence of TSO. I'm struggling to figure out how to do this with natd but I'm not finding it. I just want the single source of traffic NATted and

nothing else.

Seems like using the Java-based IPMI isn't an option: 0x0.st/8yVF.png

mason: if TSO is the source of your issues--just disable it

I can only assume that's the problem, but it's not sufficiently clear to me. I guess I can try it.

Is what I'm trying to do not possible with natd?

mason: what's that bug?

rtprio: from ipfw(8): "Due to the architecture of libalias(3), ipfw nat is not compatible with the TCP segmentation offloading (TSO). Thus, to reliably nat your network traffic, please disable TSO on your NICs using ifconfig(8)."

Oh, the Java-based IPMI line was for another channel.

so turn off tso

i thought i had mine turned off for lagg(4) but apparently i do not

I hate turning off hardware features because of software shortcomings, but yeah, that's next.

I'm really curious if that's the issue, though. But it can't hurt to test it.

I'm deeply curious why I can't just say "if you match this, route to this device." If I could do that I wouldn't need NAT in the first place.

i, like a lot of folk don't keep up on ipfw since pf

This isn't even an ipfw/pf thing. Initial tests were with neither.

Alright, I'm cheating my way out of this for now.

And with no NAT required, so TSO stays on. \o/

dropped a version of this mock up of a possible new front page a week or so ago. lots of changes since then. would love to hear your feedback (direct message is fine) freebsd.markmcb.com

markmcb: It's way more mobile-friendly, but oddly the bottom quarter of the screen doesn't render on the DuckDuckGo browser on my phone.

mason, What OS? I'm on iOS DDG and it loads fine.

markmcb: I'd also tend to want to see news and security advisories up at the top where they were, relatively. Although the "Why FreeBSD" sections are nice to see on the front page otherwise.

markmcb: CalyxOS, which is essentially de-Googled Android.

Tor browser on Android doesn't have the gap, but the table forces horizontal scrolling.

Chromium on Android is the best experience out of the mobile options I have. Only other criticism is that I'd cheat up the size of Beastie on the desktop version to be at least what it is on the current page.

is there a higher resolution version of that beastie graphic? scaled larger it looks pixelated, much like the current title logo "The Power to Serve" text

looking

markmcb: I'm seeing some higher-rez images but none quite identical, and the legal status isn't altogether clear. I bet Kirk would know of a good one if it exists.

Then again, they'd probably show up here: mckusick.com/beastie/mainpage/images.html

mason: thanks. that was the main reason for shrinking it a bit. i'll see what i can find.

markmcb: Looks good anyway. Every time I hit the page from my phone I groan a little, and your mock-up is a huge improvement for mobile stuff.

thanks :)

And I tend to hit the page from a mobile browser a fair amount, usually from a non-graphical console next to a rack. Although sometimes then I just use elinks.

mason: i'll work on the table horizontal scroll issue. i just added a tweak to make the date column disappear if the device screen is too narrow, but i'll see if i can come up with something better.

markmcb: are you just going to pitch this idea of the hompage?

rtprio: yeah, i've been discussing it a bit in the Discord #documentation channel. right now just mocking it up to get feedback if people like the direction or not.

i don't really use the homepage

markmcb: sleek

TIL "service enable"