what was the binary package manager for freebsd 8?

it isnt installed when using memstick

i can try to use a vm once im up and running to figure out how to manage a legacy_install

pkg_add ?

im not sure where it swapped over to 'pkg'

What are the meanings of the -ifa and -ifp options of route(8)? The manpage isn't very helpful

artemist: where you seeing this used?

I'm not seeing them being used, I'm trying to figure out how to force a route to go over a specific interface

sudo route add -net <destination_network> <gateway_ip> -ifp <interface_name>

artemist: what have you got so far, and how is it failing?

Also, what xa0z writes looks plausible

I tried -ifp and it seems to work, I'm mostly confused what -ifa is for when nothing mentions it

ifp is using the interface, ifa uses the IP of the interface.

Okay, cool, makes sense

Thanks

that paragraph could use some work.

When using -ifp, you specify the interface you want to use. When using -ifa, you specify the IP of the interface you want use instead.

hello, i have a dns program listening on 127.0.0.1:9999, and /etc/resovle.conf is set to 127.0.0.1. now i want my local dns request to be redirect to 127.0.0.1:9999. like if i dig google.com, it will first look at /etc/resovle.conf, which is 127.0.0.1, then i want a rule to redirect it to look at 127.0.0.1:9999 , what pf rules should i write?

i tried this : nat proto udp from 127.0.0.1 to 127.0.0.1 port 53 -> 127.0.0.1 port 9999

but it does not work. dig timeout

wow, this works: rdr on lo0 inet proto udp from any to 127.0.0.1 port = 53 -> 127.0.0.1 port 9999

that is a bit odd, eh

emmm. i tried to use route-to , the upload speed is very slow.

pass out eth0 route-to utun6 proto { tcp, udp } all

pass out on eth0 route-to utun6 proto { tcp, udp } all

all up/down go through utun, download speed is ok, but upload speed is very slow.

anything i forgot to run ?

What is the easiest, and safest way to replace all the disks in a raidz array for an SQL server?

xa0z: how else would you do it othr than replace one, resilver, and so on until the pool is complete

what the server is running is immaterial

So you think doing one disk at a time and resilvering 8 times is best?

that's the only way unless you feel like rebuilding the pool, and migrating

I could actually probably do 4 disks at one time, then do the other 4 afterwards. The pool is 4 mirrors, 2 disks per mirror.

so it's a mirror, not raidz

yes, with care you could do multiple drives at once; is there some reason to hurry?

It's a striped mirror pool.

I'm in no hurry. I'm just concerned with more drive failure during the resilvering process.

you could add mirrors before removing the old ones

That should not be a problem if your drives are well at this stage

depending on the space in your chassis

xa0z: it would be something try on a test pool

AmyMalik: just started getting sector errors for one of the disks.

then I would replace that disk

They're all old by now. 85,000 hours.

haha, half my pool is in that neighborhood

AmyMalik: theres 8 disks to replace

oh.

do you have an extra disk or disks that you can run a backup to before you begin the procedure

If I'm going to replace one, I might as well work on replacing all of them, no?

AmyMalik: I have plenty of spares.

Right, yes, it would be good to replace all of them.

I'd probably not do more than two disks at a time.. but this is not based on any evidence really

if you're feeling paranoid, use zpool attach to add a third mirror to all of them, then zpool replace one of the old drives, then remove the othre old drive

rtprio++

with zpool detach

are the sector errors correctible or uncorrectible? if they've been corrected without data loss, you probably have time before you hit data loss, but

i wouldn't nec rely on that

AmyMalik: so far just got my first one on this server today, 1 offline uncorrectable.

at what, 9 years, probably not worth to wait and see

"time" by which I mean, time enough to get things moved befoe the fan is struck with excrement

but I'd implement rtprio's advice here

ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE

9 Power_On_Hours 0x0032 001 001 000 Old_age Always - 94411

i know

Since I use mysql on zfs, I was considering sending a snapshot of the pool to another local server's pool.

rtprio: dang, even my server "only" has about 84k hours

and it's been powered on pretty much 24/7 since I got these WD Reds back in 2014

This SQL server has been powered on 24/7 for a long, long time.

I have 1 WD RE4 and 1 HGST Enterprise mirrored in each of the 4 pools.

these are wd blue. the data raidz2 is wd black

xa0z: another backup won't hurt, i guess

but if they could fail at any moment i'd just add mirrors as fast as possible

I just did a full dabatabase dump on MySQL, which wasn't soo bad. It was 3.2 GB. Smaller than I thought it would be.

that's a heck of a lot of drives for a 3gb database

/var/db/mysql is 4G lol

8x 2TB disks.

not sure when I should really start worrying about this drive in my home server. I probably should just get around to a backup even though I somewhat recently backed up the stuff that matters most (~14 years of IRC logs)

tm512: my problem is when i made this pool so long ago, all the drives were 512 sectors. since then some replacements have been 4k which kinda borked up the pool

zroot 7.25T 1.47T 5.78T - - 42% 20% 1.00x ONLINE -

it's just irritating because my external HDD is in a cheap enclosure, and apparently the USB chip in the enclosure uh, cut some corners. it just doesn't support falling back to USB 2.0 data rates, seemingly

and my home server is a bit too old for USB 3

network speeds are pretty painful since my PC is on the other side of a powerline ethernet connection

This SQL server is a 3U Supermicro dual xeon E5540

oh I tried one of those adaptors

it was not good

It's OLD.

AumShivaya: the only nice thing I can say about powerline ethernet is that it's better than connecting to the wifi from this far away

that's not really saying much though

how many feet?

or distance*

I'm not sure, wouldn't trust a ballpark guess. it's like completely opposite ends of a 3 bedroom house, multiple walls in the way

And you can't run cat6?

not fibre either?

xa0z: I've thought about it, but it seems like it would be quite an involved process (or expensive, if I paid someone else to run cabling through the walls)

if you can't run cat6, you just aren't handy enough with duct tape

if you think that, you've never lived in an apartment with lathe and plaster

AmyMalik: it's fiber internet, but the fiber stops at the garage, it's then coax into the house from the ONT to the main router

I only have ethernet (cat6) ran to 1 room outside of my server room, and use it with 10gb SFP+ adapter in my 10gbe NIC

all well and good, but what about fibre (SFP+) ethernet?

although in one apartment i used comcast's coax to pull cat5 from the basement to the 2nd story

A buddy of mine lived in an apartment and ran CAT5 behind the baseboard trim.

AmyMalik: seems like the same situation as running cat5/6

not handy enough with gaffer tape eh

rtprio, perhaps not

fiber is too expensive to run, opposed to cat6.

not really willing to hack it up by running cabling through the entire house outside of the walls

ah, yeah. and not kinking either

one thing I have thought about, is that years ago to get TV into this room, we had coax put through the walls

I have 1 fiber sfp+ cable and it goes from my switch to my NAS. It's like 10 feet long and I think it was $40.

I've thought of running ethernet over that, but I'm not even sure where the other end of the cable is

if the coax is not connected to anything, you could probably run wifi? (yes, over coax; yes I know it's a bad idea)

(I just like to advocate for impractical jank whenever I can)

like connect the other end of the coax where an antenna would go on the router? that sounds cursed

yes

exactly like that

and then, idk, vampire tap the cable if you need a network somewhere (don't do that, you'll get all sorts of reflection badness)

this newest router we got doesn't even have an external antenna terminal, it's one of those ones that's entirely internal to look nice and clean or whatever

(you'll also torch the phys)

You mean, people don't use managed APs connected to routers/switches?

so with sendmsg(), either it sends the entire message, or i get EMSGSIZE and nothing is sent... right? or can a partial message be sent?

this is with SOCK_SEQPACKET sockets in AF_UNIX, in case the protocol makes a difference

how do you tell pkg what repository to add pkgs from?

oh, pkg install -r

What are the equivalents of /dev/drm, /dev/dri etc. when using Xorg in FreeBSD? or do you enable linux devfs for using X?

albertus: if you have drm-kmod installed, you should have /dev/drm and /dev/dri, these are literally just ports of the Linux DRI drivers

ah, so is drm-kmod a dependency of the X server?

I installed the X server in a jail, to no avail

no, since X doesn't require DRI/DRM, but if you want accelerated graphics on a supported card (Intel or AMD mainly) then you should install it

what would be used without it? the simple framebuffer thing?

any driver shipped with X that doesn't require DRI. so NVidia, S3, VESA, ...

I'm wondering which devfs-nodes I have to pass into my jail, or what else I would have to do, to make it start

you probably need to pass /dev/dri and /dev/drm into the jail

... if you're using DRI

... that's where my question came from

well, is drm-kmod installed and loaded on the host?

you can't run that in a jail - it's a kernel module

I have not installed drm-kmod, and only installed X server components in the jail, so probably no

but I was wondering if DRM is the usual way if I were not using a jail

it depends on the graphics card you're using, if it's Intel (i think yes since you're the person the odroid H3, right?) then yes you need drm-kmod

yes, it's the H3 with integrated graphics on Intel

okay, thanks

oh that pkg brings a lot of firmware pkgs :)

many "lakes" on Intel, but not the right one (Jasper), as far as I can see

albertus: intel tends to be a bit vague about the GPUs on these celeron/pentium chips, try `kldload i915kms` and see what happens

all of a sudden, the screen blanks and the font is a LOT smaller (or... the screen resolution actually uses the WQHD monitor!)

that means it's working (almost for sure... i guess you need to start X to be really sure)

I think I'll need to enable the /dev nodes in the jail first :)

I do have /dev/dri and /dev/drm now

i've never tried X in a jail but i don't see any reason it wouldn't work as long as the appropriate devices are available

also don't forget 'sysrc kld_list+=i915kms' so the module is loaded on boot

I only installed X because Kodi failed to start. but I think I'll just have to specify a devfs-rulseset for the jail and will see how it goes then

(i think drm-kmod tells you to do this when you install it anyway)

it told me, but when I started vim to look inside /etc/rc.conf, it had overwritten that message and I had not written it down

albertus: also re: font size, try this in /boot/loader.conf: screen.font="12x24"

that's about the right size for me on a 27" 4K display

(you might also want 'exec="gop set 0"' but you have to reboot into the loader and run 'gop list' to find out which number is which resolution)

(other fonts are available in /boot/fonts, maybe a slightly smaller one is good for 1440p)

okay, i want to submit a phab review for ports (not src)... do i still use 'arc diff --create main..' like

[00:00:22] Cleaning up

[00:00:22] Unmounting file systems

whoops

okay, i want to submit a phab review for ports (not src)... do i still use 'arc diff --create main..' like wiki.freebsd.org/Phabricator says? or do i need to tell it the diff is for ports?

Title: Phabricator - FreeBSD Wiki

welp i did the thing, hope this is right reviews.freebsd.org/D43637 ... never used phab before

Title: ⚙ D43637 audio/pulseaudio: convert JACK option to a package

how do i update an older release? pkg: Error fetching pkg.FreeBSD.org/FreeBSD:12:amd64/quarterly/Latest/pkg.txz: Not Found

hi, I'm trying to run npm with npm-node21-10.2.5 and I get: ld-elf.so.1: /usr/local/bin/node: Undefined symbol "uv_pipe_bind2" (FreeBSD 13.1)

BTW, this is in a 13.1 jail on a 14.0 host

Kobaz: you need to upgrade the OS before packages. try freebsd-update, although i don't know off hand if update from 12 is still possible

hmm, that would be unfortunate

i *think* it still works? 12 was only EOL'd like 2 months ago... but i've never tried it

debian is pretty good like that... even in 2024 I would be able to upgrade from debian 1.0 to the latest via apt

freebsd-update fetch looks like it's going

would you? i mean would it actually work? ... because i used to be a debian maintain and my experience is the upgrade scripts are *definitely* not tested across that sort of timeframe, so it's kind of random if it works or not

io: yeah, debian has every single release on archive.debian.org

lw: I successfully started X :) will tinker some more the next days. thanks so far!

Kobaz: so does freebsd, that doesn't mean upgrading from a 20 years old release to current release actually works

lw: why wouldn't it?

lw: it's the same process anyone else would go through to upgrade

Kobaz: in debian's case, because almost every package includes various pre/post-install scripts which are not tested for upgrading from a release that old

lw: i would assume 12 -> 13 -> 14 will be the same process years ago vs today

like, it's *theoretically* possible, but did anyone test it?

lw: what you mean no tested?

Kobaz: when did someone last test upgrading from debian 1.0 to whatever the current release is?

like, what would it matter if i upgraded from 1.0 -> 2.0 today... or 15 years ago... the script is the same

well, yeah, if you just want to upgrade one release at a time that should work fine in freebsd too

lw: i think you are assuming going from oldest to latest in one step.. which was never a supported operation

it's just the old ports packages aren't kept around for old releases

so you upgrade base to the current release, then update packages to whatever the current version is

(freebsd doesn't keep old ports package around in general because ports doesn't really have releases...)

ah

freebsd-update fetch .... Fetching 28 patches.....10....20.... done Applying patches... done. Fetching 72 files... <-- this part seems to be taking a while

that does always take a while, not sure why

ah, k

pkgbase will fix that i guess

although now i wonder how well pkgbase works if you're upgrading from an EOL base with ports installed... hmm

i always forget the freebsd ways of doing stuff for things like strace and ps auxf (there's no forrest option for *bsd ps)

ah, looks like it's ps -d

there it is... truss... hehe

okay looks like fetching is actually doing stuff (reads/writes)

yeah -d. they call it decendency, not forest

i switched my todo app and started migrating, and the new todo app wiped all my new todos

i need to restore my desktop from backup but now i realized i haven't been checking my backups as much as I should... this hasn't backed up since 350 days ago

and then it turns out email from backuppc hasn't been working since it looks like the default sendmail install doesn't work... so.... ugh

oops