hey

hi

ho

after disk corruption my pkg local.sqlite file has been corrupted

how can i recreate this file?

your pkg or your database got corrupted?

only /var/db/pkg/local.sqlite

I removed that file (which becames unreadable

and now pkg tells me that there is no installed pkg

pkg rebuild

unknown command :/

$PATH properly set?

from man page of pkg

Restore a backup database:

% rm /var/db/pkg/local.sqlite

% xzcat /var/backups/pkg.sql.xz | pkg shell

that would not work if pkg is not found

there isn't pkg rebuild for me either

titou: unknown command was for `pkg rebuild` and not for plain `pkg`, correct?

ah yes

its register, but thats a one by one pkg thing

you got all in stalled pkg data still in /var/db/pkg ?

i have a raspberry pi running 13.2-RELEASE, is it safe to upgrade straight to 14.3, skipping 14.1 and 14.2 ?

jozan: probably yes, but you must have freebsd.org/security/advisories/FreeBSD-EN-23:12.freebsd-update.asc to update from 13 to 14 with freebsd-update

oh thanks ivy

sorry i had to leave quickly

ridcully: yes it was for pkg rebuild

iRobbery: thanks a lot i will try

it worked! Thanks a lot !!!

you rocks

\:D/

upgrade to 14.3 from 13.1 was smooth, i did have the include/c++ error tho

it seems to be running fine for now

titou: good news :) i had to nap

anyone know why a thinkcenter m710q might be fussy about an nvme as boot?

I transplanted one just now, but the bios doesn't want to play

friends

friend

Internet stranger.

how cold

i mean not feet

i turned to MAKE_JOBS_NUMBER_LIMIT=1 in my rust build

:p

Well, I *am* a cold-blooded critter. :-)

i'm truly puzzled what it might be doing if it wants nearly 2g

and i dunno if the synth developer is reading here but i have question about his software. I need it to compile some packages with custom compilation parameters and share produced package between machines

synth grabs 2/3 of the needed packages from the FreeBSD repository but not all of them (except those who have custom compilation options).

In particular: gcc, rust, llvm...

I don't know why

I don't care about them, they are just used to compile other packages and official FreeBSD version would be enough..

it's really boring to recompile them each time i upgrade ports

moreover the LLVM version correponds to the one rebuilding... i don't understand

+in my ports

Hey there! I'm on FreeBSD 14.3-RELEASE, and it seems that freebsd-update says that my Xeon E3-1220L is not a tier 1 arch?

claims I'm on amd64, which should be tier 1

wipt: can you pastebin.com what you are seeing?

titou: haven't used synth, but poudriere too is broken now and doesn't fetch any prebuilt packages

zi: seems that it's a default error message for freebsd-update if it can't connect to to a server.

Seems there isn't a "true" issue, just was internet connectivity. Bit of a jump for an error message, though.

wipt: yea--it's kinda dumb with error handling

Hence the reason freebsd-update is going away. :)

re

has anyone tried the misc/ollama port?

BHYVE framebuffer device detected

BHYVE Get Graphics Mode: w 7320, h 32494

something doesn't seem quite right here

ek: if you don't mind, what is replacing freebsd-update?

aic: pkg ugprade

hmm so setting a different default gateway and setting a /32 IPv4 does not work

arp no dice

arp cant see the router hmmm

probably because the router interface has no IP

what if I add a RFC1918 address to it and then set this as a gateway for the connected device with a public IP address (/32)

then see if arp works

polarian: what are you trying to do?

zi: someone ages ago spoke about individually assigning a bigger block of IPv4 as /32 to save the network and broadcast address

trying to figure out how they did it

my router holds a /32 address for NAT within a different block

so I set that as the gateway, but arp cant see the gateway

I assume because the port the device is connected to on the router has no address...

you will need a 'point to point' interface between the two devices--likely a /30 for simplicity. then, you can route another netblock over that

right

why not a /31?

id obviously need more information about the setup and what you're trying to do

it doesnt really matter

but the above is the basic jist

but can you do the ptp with a rfc1918?

you can, but traceroutes and some other things from the outside wont work quite right

also question, if you need ptp then how come ISPs work then, they allocate /32 addresses within a bigger subnet and *that* works

but then again they probably have a public IP on the interface all the connections go into

i think youll find that most of the time, you might be getting a single ip, but it's part of a larger block (/24 or whatever)

indeed

but my point is the gateway is still out of the subnet

so check the netmask on whatever you get from your isp

it probably isnt a /32

i doubt that

nope

0xffffffff

and the gateway is completely different

and it works for them

:P

you're gonna have to try to type up some information about the setup/devices/etc and what you're seeing

but like I said, in order for arp to work I assume the interface your device is connected to must also have an IP

okay how my ISP works is that for PPPoE you are given a /32 netmask 0xffffffff and then a gateway

that works, obviously otherwise I wouldn't be connected right now

my question was simply if you must use point to point for this to work, then how come this works for my ISP

pppoe is a little different than just ethernet

so my question was in order for arp to work on Linux, the iface I assume always needs an IP address

zi: sure...

ok going back to the original question then

take a /28, you want to preserve the network and gateway address, to allocate them as /32 without needing to allocate the subnet to the router, I you need to ptp it?

take 192.168.0.1/30 and 192.168.0.2/30

.1 router, .2 device

set .1 as the default route, then the public IPv4 passes via the default route

is that what you were suggesting zi?

it depends how it is setup and who is in control of the devices

im in control of both the router and the device

one option is to have your isp route the /28 to .2. you could use every ip in the /28 in that scenario

nah ignore that the separate IPv4 block just goes via the PPPoE route

I am not asking that ignore the ISP

I have additional addresses, I dont want to waste them on a network address and a broadcast

I can keep them router side, allocate a ptp address and NAT to each address router side

but a friend which is sharing this asked why I cant simply give a public /32 address to the device and set the default gateway to the router address, but that doesn't work because arp cant see it

polarian: you're gonna need to make a diagram if you need more info

ugh

it sounds like you want me to argue with your friend, through you

unless im misunderstanding

no

my friend simply asked why I dont do that, so I thought "hmm I will try"

ok

lol

wait I didnt actually realise I was asking this in freebsd

its between freebsd and openbsd :P

window 3

denied