eekay I guess it's just highlighting my nick when it is called out.

hey

But yeah, I am running multipe boxes, so I choose to run a poudriere repo and build my package sets in house.

Howdy luna__

argh. what in the world. what am i doing wrong with these gmirrors where they don't come back after reboot

gmirror supports gpt partition labeling doesn't it? i don't understand why the mirror always disappears after reboot even though i have it loading on boot

Macer, Please say a little more description around the mirror disappearing?

You boot and "gmirror list" doesn't show it?

"gmirror status" says what?

it says nothing

geom_mirror_load="YES" is in loader.conf

Hmm... Maybe the mirror needs to be created/recreated?

Example from my system: gmirror label -h swap /dev/gpt/swap0 /dev/gpt/swap1

yeah that's what i've been doing. and the expectation is that once that is done the mirrors will show again.... hold on i'll do it again and pastebin it somewhere

I am assuming that the partitions exist. gpart show and gpart show -l

And my /etc/fstab entry: /dev/mirror/swap.eli none swap sw 0 0

The .eli configures it as an ephemeral keyed encrypted swap partition. Which is awesome because swap does not need to be persistent.

and on reboot

it's gone

am i missing something in loader.conf or something? i thought it was just geom_mirror that was needed

Looked at your pastebin.com/R1QTsqwk and see that it is working! Yay! What did you do to solve it?

that's not the problem

i can create the swap just fine and use it. my problem is it doesn't persist after reboot

Reading... It disappears at boot? Look at the boot log, look at dmesg, hopefully some error clue.

I don't know. Works for me. And the coffee shop is closing now so I must relocate. Carpooled with a friend. Good luck!

yeah. i'm looking at it now

but i don't see anything geom related

Macer: kldstat | grep mirror, do you see the module loaded on a fresh boot?

i'd have to reboot again to see and rebooting my nas over and over isn't too cool of a thing to do heh

but i'll check next time. i just figured maybe i was missing a fundamental step here .. i thought creating labeled partitions on the drives then creating a mirror using the gpt labeling should do the trick

but i think i thought about that a few reboots ago when it didn't work and geom_mirror was indeed getting loaded on boot

it's just a little irritating having to recreate the mirror/swap on every reboot

(manually) and i'd like to avoid doing some sort of cron tomfoolery with a @reboot script that does it

i didn't think there needed to be tomfoolery

considering that encrypted swap is an option in the installer

i can't remember how many versions ago i install fbsd on this.

i vaguely remember it being around the time freenas turned into truenas.. maybe a version or two after that lol

That'd be around 12 or 13

was it? thought it was around 11 for freenas but i could be wrong.

i felt like it was a good decision because once they came out with truenas scale you just knew they were going to essentially abandon using fbsd even though they said they wouldn't but then did

Yeah, I was sad to see the change. I am currently using truenas to house my VM's.

I was excided to see this in the Q42025 repoort though.

Well, there is still the FBSD TrueNAS fork of zVaultIO that may come to fruition. It was looking really good in the beginning, but the project seems to have halted for unknown reasons.

The 13.3 replacement of TrueNAS works perfectly fine, but everyone was hoping for a 14.x release before 13 reached EOL. Sadly, it's been quiet on that front.

the main reason people use TrueNAS is the web UI, right? i wonder if it would make more sense to put that in ports, if that's feasible, rather than building a whole new OS for it

perhaps

just like if there was a couple of graphana dashboards for unbound and a big red button, i could replace pihole

ivy: That was actually the future intention of zVaultIO. They were going to continue to support the appliance until they could figure out how to make it a port (I'm sure no easy feat).

But, alas, the project seems to have died. Very unfortunate.

I think the big draw for the appliance is the ease of configuration backups/restores for re-installs, though. That does come in handy.

any idea how to watch traffic with dwatch and ignore traffic for a specific port? Now I am using "dwatch -X tcp" but I would like to ignore ssh traffic

rtprio: did you find a grafana graph for unbound? i had jgoguen.ca/posts/2024/07/20/openbsd…bound-and-statistics/#pretty-graphs sitting in my bookmarks which hooks up the two

meandrain: hm, -t <test> seems relevant, but I have no idea what format is expected for the test and if it would get the port

maybe 'tcps_lport != 22'

nimaje: not working, maybe there's no way to filter ports with dwatch

nimaje: this works but it's weird: dwatch -X tcp -t 'args[3]->tcps_lport != 22'

(don't ask me how I've found out ;) )

well, after seeing a example in the man page, I suspected that <test> to be a dtrace filter expression, but no idea for which probe and hoped dtrace would make unique fields of arguments avaible as just the field name

bugs.freebsd.org/bugzilla/show_bug.cgi?id=293243 update, i was able to reproduce it in a vbox vm

OlCe2/freebsd-src #2 also btw did anyone saw a phab review for this? i couldn't find any

TIL: This is a really cool tool to validate your outbound E-mail setup: learndmarc.com

hi, while setting an adidtional floating IP I do this:

ifocnfig_vtnet0_aliast0="inet X.Y.Z.W/32" in /etc/rc.conf shall that suffice? thanks

ifconfig_vtnet0_alias0

the interface is configured previously with DHCP

uskerine: ifocnfig_vtnet0_aliast0="inet X.Y.Z.W netmask 255.255.255.255" (if you actually want it in a /32?)

ifconfig_vtnet0_alias0*

uskerine: Is this vbox by any chance?

I'm on my new OS freeBSD, installation went great and now KDE on wayland is working great! thank you to everyone that helped me yesterday with my questions prior the installation process

mirens: Glad to hear it!

mirens: Great to hear and enjoy

thanks

Im setting up network with another PC at home. I need kio-extras for playing videos over a network ?

btw samba share works faster than on linux or I am imagening things

this is lightning fast

Always been partial to, and still using xfce

balsamic-oval: i never looked, don't have time for any build out at the moment

rtprio: no worries. maybe worth a bookmark for when you circle back. im in the same boat

`tbh, having the temporary on/off switch and block lists would be priority

the fewer linux VMs to run, the better

how do I add a route to say to reach IP address 172.31.1.1 use interface vtnet0 ?

do you have multiple gateways?

Not really, you can see the "goal" routing table in the link, everything shall use 172.31.1.1 as default gw, but that IP goes through vtnet0, and vtnet0 is configured with an IP of a different network

it is a VPS, so the routing is a bit in that way

iirc i think you want dhclient and a supercede option? perhaps someone else has an idea

nulltaz: You mentioend using an irc bouncer. Do you have a favorite that you like to use?

uskerine, Are you hosting at OVH? OVH has some "interesting" details that make things very particular about network configuration.

No, it is not OVH

heston76: weechat relay is great for mobile usage

I got one server burned there, literally BTW (in OVH)

other android apps + znc don't work reliable when you on a move

in my opinion

and OVH management GUI sucks!

uskerine, For one OVH uses IPv4 /32 and IPv6 /128 address assignments (IPv6 is actually allocated as /64 however, but assigned as /128) which means that to get to the router it is also required to add a host route to the network device for the router.

Since everything is dual stack the way I worked through the problem was to keep IPv4 working then to debug and finally get IPv6 working. Then log in with IPv6 and reverse that debugging IPv4 and getting IPv4 working.

add a host route to the network device for the router. <- you mean I have to tell the system, to locate this router IP (172.31.1.1 in my case) use this network card?

I know but I want to know how to do that

This is the routing line I do not know how to create in /etc/rc.conf (nor in command line): 172.31.1.1 link#1 UHS vtnet0

I am making a pastebin...

uskerine, bpa.st/XNMA I crafted that using example.net addressing so that it would be more generic. Let's hope I didn't screw it up!

The structure is exactly what I am using at OVH. I don't know if this is your issue or not. I didn't quite understand the problem statement. But it sounded like it might be the same problem. You will need to decide.

Oh I screwed up one thing. In my comment I typo'd /31 which should have been /32. Oops.

I am adding static_routes and route_lan

When messing with networking always make sure you have an alternate path to the system. Network console. The other side of the dual IPv4/IPv6 stack. Whatever. Because it's not if we screw up and break it but when!

it works now and the routing table is the same

Just fix one of them at a time. Rework it until one of the two is working. Then use it to fix the other.

I use the virtual console from the cloud provider

So does that mean we were successful? \o/

so you never can lost access

Excellent!

yes but I would like to understand the meaning of static_routes="lan" and routes_lan, because everything else I get it

why do we need static_routes="lan" ?

The "lan" part is an arbitrary string for the variable name. It could have been "herman" and then the variable name would have been "route_herman" just for example. I used "lan" because that made the most sense to me.

In the rc.d init script it will look for strings listed in static_routes and then use that string to construct the route variable name that needs to be set.

so you can not just define the route without defining a static_routes "domain"?

Right.

I see

So for example: static_routes="herman" route_herman="-host 104.18.5.1 -interface vtnet0"

thanks for your help rwp this helped

yeah I got it

That would have been a valid combination too. I use herman because it is an unusual name and therefore we can identify it in the construct. Should have used "foo".

actually last time I deployed this I manually added the second IP via ifconfig (not very clean)

I think this is much better, it autoamtically brings up both IPs

Yay! \o/

The first time I encountered this problem it took me a while to figure out that it needed a host route for the gateway router address because it is not located on the subnet, because there is no subnet configured there since it is a /32 and /128 configuration. That took some frustration for me before I finally figured it out.

I already had encountered this, so I more or less was clear on what to do, but I did this before in Linux where you can force routing to an interface , I did not know how to do it in FreeBSD

I have been told that OVH allocates IPv6 at the /64 block prefix level but still assigns them at the /128 anyway because on IPv4 they do the same.

Cool. Anyway... I am getting in a time crunch and must run off. Glad this helped! Good luck!

and /etc/rc.conf has its way too, I have actually half-read how the booting process works, the handbook has a good description and there is even an old "paper" describing it in detail and its rationale

thanks rwp it helped

Nie mam rozsądnych komentarzy. Także: zajebiście niedojrzały, i chyba najdłuższy kawałek h-h w naszym polskim języku: youtube.com/watch?v=I6VG5rdwIeM

Ooof, apolofies! This was meant for #freebsd-pl ...

apologies*

I recently upgraded from 13 to 15. Now when i start x, the keyboard input is nonsense; seemingky random characters and control keys, not consistent (so likely not a bad map). Any tips on a likely culprit?

When the x server comes up, if i immediately type “pwd” followed by enter i get “ p2w3d45” - rherea pwd in there - but then it deteriorates.

anth, happends in all X programs?

All i’ve tried.

No idea, but I am afraid to upgrade my own from 14 to 15.

It seems like mouse clicks start getting eaten, too, but the behavior there is less clear. They might only be eaten on the root?

There is an X program for tracking mouse events, but I forget its name. If the problem is reproduced on a freshly installed v.15, you can report it in the bug tracker with the reproduction scenario. I don't know how to help you. Maybe try reinstalling X?

...or force-upgrading?

xev. I tried that, but I haven’t gotten it to produce any output. I think I’m invoking it wrong, but I have to reboot between attempts, and it is slow and irritating.

it shouldnt be slow but it is irritating

Slower than just re-running a command!

anth: what desktop environment / window manager are you using

you move the mouse over the xev window and type something, and a ton of shit should scroll past

Ah, then you could try it with barebones startx and twm.

i _think_ it spits out on stdout, so start xev from xterm or your other terminal app

It does.

I’ve gine back to the stock twm to remove variables.

Yeah, i’d started it directly from .xinitrc which doesn’t work.

Can you share your not-working .xinitrc ?

...or delete it, and then the global xinitrc will be used.

no, that would not

?

There is a global xinitrc that just works (unless you meddle with it).

/usr/local/etc/X11/xinit/xinitrc

A .xinitrc containing only the shebang line and twm exhibits the problem.

And the problem is what -- broken keyboard and mouse input?

Broken keyboard, for sure.

I can’t tell for sure, I think the mouse is working properly but gets confused because it thinks a keyboard modifier key is being held down.

OK.

Well, you might try digging <man.freebsd.org/cgi/man.cgi?query=libinput&sektion=4> , including reintalling it.

I saw a note about the sc device being deprecated; is scfb still the right X device to use? It seems to be from the docs.

Hmmm. I thought sc and scfb are not related to X...

X uses its own graphics drivers.

I agree, I think it’s just coincidental naming for the frame buffer driver

It's the terminal/console/tty driver that's deprecated.

Anyway, thanks. I’ll have to poke it this more later; I have to pick up my kid from school. Next step is trying to get cev started in a way where I can actually see it’s output.

xev?

You can redirect its output, includng with tee.

How do manage to run xev, if you can't type its name?

You could run it from .xinitrc ...

Yes, that’s what I’ve been doing, but I need to have it run from cterm or similar, otherwise I don’t see the output.

I can’t even force quit the ex server, so it really is a reboot on every attempt.

“ex server”. No, dictation, I can’t make it “ex”, that’s the problem!

Don't forget that Control-Z will put ex into the background and returning you to the command line prompt.

how come in freebsd if you dont have any ipv6 config for an interface, you dont get a LLA

i don't recall that being true

that doesn't sound right