i asked every day for the last three days but all I saw was an obsolete forum link on a slightly different question

So what, if anything happens if you install gptzfsboot boot blocks on a UFS system?

i can make a zfs pool with a disk drive on a usb adapter and i can then transfer it to a disk shelf when i get one and keep it in the same pool without losing data?

darwin: did you try tput already?

no one mentioned it

darwin, Try "vidcontrol black white" to reverse the colors and "vidcontrol white black" to return.

Using vidcontrol was found in the forum link that meena posted. Works for me.

thanks

how do I enable 12 ttys? I added more lines in /etc/ttys but it didn't work for 11 & 12

and where can I set global command-line aliases (for all standard shells)?

i don't think that's a thing? i think each shell has its own config file, but maybe most of them read .profile

is the syntax different for default root & user shells then?

It looks like MAXCONS is defined as 16. Perhaps you ned to remove others or alter that. But I am purely speculating.

Title: syscons(4)

i didn't say 16; I said 12

darwin: on freebsd the root user gets a tsch by default, regular users get sh. you have to install bash or what not yourself

i don't need bash right now

The total is 16.

That likely includes all the other types of consoles in /etc/ttys

i see

But again, guessing here.

Check that oage. The kernconf can increase MAXCONS

check that page even

Perhaps removing some of the serial consoles is an option, but I have not tried.

i looked at man syscons. Does this mean I'd have to compile my own kernel to increase MAXCONS?

That was my understanding.

But perhaps you could comment out some of the other terminals, like ttyu3 or something

Then add more ttyv9, ...

Erhard: do you know much about why vt replaced syscons

No clue, sorry

i don't even see a serial console

but I don't need ttyu*

Oh, my /etc/ttys has the ttyu*s

Your does not?

of course it does. I said I didn't see one described as serial console

13.2-STABLE here

I see.. mine has a comment that says #Serial terminals

i see it now

that is solved, but what about global aliases?

Don't know there... You want alies for the different consoles?

if sh & (t)csh alis differently then surely I can just put in a global file something like if $SHELL=sh, or if $SHELL=csh, etc., then do different sections of aliases

if they alias differently

What are you trying to alias?

Make some command have an alias in all shells?

various commands

yes

I guess you would put the alias into the defaults for each shell

But perhaps there is a global .login or something

well it has to be global regardless of user

OWuld have to check the docs for all the shells.

i thought it might be in /etc/profile or something similar in /usr/local

I would think, yes

actually the console thing isn't solved; apparently it's not only configured in /etc/ttys

i commented out two serial consoles but still didn't get an 11th or 12th tty

even after rebooting

YOu added the extra ttys ?

ttyv9 "/usr/libexec/getty Pc" xterm onifexists secure

Something like that

yes

i just copied a previous line and changed the number

Not ttyv8, though, right?

That is for X

YOur new lines have getty

?

now it's for 12. They do

Then I do not know

Anything in the logs?

maybe grep for tty or getty (which I guess includes tty)

oops, I forgot the first is ttyv0 not ttyv0. I redid this and maybe it'll work this time

'not "ttyv1"'

I wonder if you have to create the device files

oh...

Just thinking out loud

they should be added up to 20 since some newer keyboards have Fn for n=1 to n=20

Mine has through /dev/ttyvb

And I don't think I ever created them by hand

oh, they're in hexadecimal?

yes, it would seem

According to that page I linked.

And my /dev/directory

And to go over F12 you use Shift-Alt-F1, F2...

you don't need shift

If you define those

Supposedly Shift-Alt-F1 gets you /dev/ttyva

Which doesn't make a lot of sense

Since alt-F12 does too

According to that page

But Shift-Alt-F3 gets you ttyvc

*shrig*

shrug even, lol

And use the print screen key to cycle through all terminals

i was able to get ttyva but not ttyvb

Try Shift-Alt-F2 ?

Or the print screen trick

is that because it's defined for X?

Oh, yes, probably

If you are not running X or have it installed

i have it

something I liked on GNU/Linux is I could remove a password or (for blank) asterisk fro /etc/passwd then login would accept that username and immediately login without prompting for password... is there any similar thing I can do on *BSD?

from

yes, i believe so

i removed asterisk for blank one but it didn't have this effect

# pw usermod example_user -w none

I think you can leave the password blank in the password file

And then there is none

i forgot I had to use 'pw usermod'; thanks--that worked! :)

I think you could delete the password form vipw too

from

BUt it is more reliable to use pw usermod

Or at least harder to mess up

After "vipw", need to regenerate the database via "pwd_mkdb(8)"

Sorry, no, not according to vipw(8); it calls "pwd_mkdb" itself

That was pretty useful: ozgurkazancci.com/freebsd-protect-y…ver-against-php-shells-and-malwares

Title: FreeBSD: Protect your web servers against PHP shells and malwares – Özgür Konstantin Kazanççı | Blog

hi!

i actually switched to freebsd yesterday

welcome, then :-)

I just came back after a ubuntu disaster

thanks a lot!

its been a bit frustrating from time to time since im not that used to building programs out from back othe linux days and such, but after fixing upa few things ive got xfce and librewolf working

however, i have a small problem concerning internet access on my pc

They aren't working straight from pkg?

they do work straight from pkg, thats what ive done to install all the drivers and big packages such as xfce and browser as well as neovim

my issue is that i cannot connect to the internet

pkg install <file name> is easy

it really is

i could connect to the internet yesterday

aha...

but now i cant, i first got an error about the ethernet connection having no carrier, i then moved the cable to the other port on my computer and it changed to active, but i still cant go online

Have you imported settings into the second nic?

not really

Just swapping cables over won't make it work, lol. What does ifconfig tell you?

im on linux right now so i cant just give out what ifconfig told me but it pretty much said that it was active and also that the ethernet was labeled ig0

before swapping the port the cable was in it was saying no carrier instead of active

for reference, my motherboard is an asus rog gaming x570-e, which iirc has both intel and realtek ethernet ports

No carrier indicates a problem with ethernet cable or modem. Yes, the ig0 is an Intel ......

yea

and yesterday the same cable was workinng

Are you using DHCP or static ip addresses?

dhcp

its a bit hard for me to set up because all the forum posts i see are for static addresses]

but at the same time my address changes like every 2-3 months or so

not every time i restart the router

That's what DHCP is for ..... Dynamic Host Configuration Protocol. Without your pc running it's a bit difficult to assist. DHCP is pretty foolproof

i see

i guess ill then try to log into freebsd and log into here (irc) on some laptop then

how have you configured it in rc.conf ?

either that or kiwiirc, wait a bit

Both my nics are static, I'm not using DHCP

i see

Check rc.conf, it probably has the old nic selected, change that to your new one. It's midnight here, hope you get it sorted.

okay, just booted up freebsd

ok, so what's in your rc.conf?

RhodiumToad, good luck, I'm out.

I’d say if I should just reinstall freebsd all over again but I really don’t want to do so again

ok, so what's in your rc.conf?

you should not reinstall

I don’t remember exactly

And

oh.

I partitioned my ssd while freebsd was running and got this error after rebooting to freebsd in order to get rc.conf

while it was running, or while it was not running?

while it was running

partitioned it how, with what?

Used gpart and took away like 128 gigs from the storage partition of the ssd

I fixed my mounting issue

I have a small question though

Do I need a webserver to use poudriere? The whole thing concerning http servers for getting the packages has made me a bit afraid of using poudriere (especially since my router is locked down) and I want to know if there’s an easy way to use poudriere with the packages stored locally

you don't need a webserver

storing the packages locally is easy

I see

How should I configure poudriere then?

though having a webserver makes it easier to keep an eye on progress

Seems about right

(though even without one, you can get goot progress info from control-T in the terminal)

Nice

I actually have a vps but I’d rather not use it for poudriere since I already use it to host websites and services

Plus it uses debian

configuring poudriere is just a matter of looking at poudriere.conf and fixing anything that's not right for you

Thanks!

the main thing to watch for with poudriere is how much tmpfs space you let it use

How much should be okay for an 1tb ssd and 16gb ram?

it can be extremely profligate with ram+swap use if you use tmpfs too much

(Hopefully expanding to 64gb… after I get my hdds)

And seems aborut right

I have 4gb of swap

USE_TMPFS=no will avoid using tmpfs at all. the default of "yes" will use it for wrkdirs, which for some ports can be quite large.

USE_TMPFS=no will avoid using tmpfs at all. the default of "yes" will use it for wrkdirs, which for some ports can be quite large.

So i should not use it for let’s say, firefox

Speaking of Firefox, it and a few other really big packages such as xfce break when I try to build them as ports

What kind of stuff would you NOT recommend me to build from source, aside from low level stuff like drivers

I build everything from source, but I don't mind it taking a long time.

building llvm, gcc, ghc, tex*, and rust is a very long process

Yeah I just installed gcc and llvm through pkg as we speak

Should I get portmaster alongside poudriere? I know it’s mostly deprecated but heard it could make some stuff faster as well

(I also build the whole OS from source)

nah. postmaster isn't really worth using these days.

*portmaster

Damn

And yeah that makes sense

I used to use it a lot, but when I switched to poudriere I abandoned it completely

I guess it’s just poudriere and rust then before I start to set up xorg and xfce

do you have a particular need to compile X?

Or arte there any other big packages I should fetch as well if I want to compile xfce succesfully

isn't there a package for xfce?

Yeah ik

But its a bit too bloated for me

ah ok

I absolutely don’t need ristretto or parole

Or svn on thunar but no git

And if I use okg autoremove the xfce extras I do want like the screenshot tool also go áay

if you want to mix packages with stuff you build in poudriere, I don't know the best wat

*way

Yeah ik

the thing is that poudriere doesn't use the installed system's packages because it builds everything in clean jails

Back on debian/guix I remmever mostly using packages for the very low level stuff

I see

So I’m better off going full one or the other

And uninstalling all compilers on my installed system

I think the -devel version has support for fetching packages

I see

but I haven't used it myself yet

Ooh

So this means that unlike portmaster having gcc12 installed in the base system won’t speed up building time

right. but portmaster has its own ways to waste time

Also any root tips?

I know root should be used as infrequently as possible, but what’s the best way to handle that? Creating one account for all root things and another user with no access to root whatsoever?

(Desktop wise at least)

just use sudo

Adding an otherwise non-root user to "operator" & "wheel" group goes a long way; e.g could reboot, shutdown the machine

what's the distinction between operator and wheel?

I’ve decided to use this switch to fully switch to doas but yea

In my planned case I have two users on wheel

wheel group membership lets you use 'su' to become root

One has all the superadmin powers, the other can only really shut down and reboot the machine as well as load xorg

Inb4 I find a way to do rootless xorg

(and is often used in sudo configuration to control who can become root with sudo)

you don't need root to run xorg

depending on what GPU, you may need the user to be in the "video" group

Nvidia user here, I’ve added my users to video

And should I use sudo/doas passwordless for poudriere when I su into my non root super user?

up to you

Ty

btw, do you use a light or a dark background for terminal windows?

Background image

Usually at like 60% opacity with a dark background though

it's just that poudriere uses bright yellow for some of its output, which is painfully unreadable on a white background

(I have to fix that using COLOR_SKIP=${COLOR_BROWN} and COLOR_WARN=${COLOR_BROWN} in the conf)

I see

Setting up doas rn but I get some error at line 11 message

Frankly all the colored text from wherever ought to have background color specified too for enough contrast if the source of the output would not find out the existing color & adjust accordingly

I see

Going to install xorg but I get this error

dns lookup not working?

does drill pkg.freebsd.org return anything?

Nope

what about other dns lookups?

Lol I fixed it

Turns out I had not restarted netif yet

I also uncommented the dns servers I got on install despite using dhcp, should I comment them again?

usually you should use the ones from dhcp.

dhclient will normally replace /etc/resolv.conf accordingly

So it should be fine if I keep the name servers I got from dhcp uncommented in /etc/resolv.conf?

Or should I comment them again

I also noticed that nameserver 127.p

127.0.0.1 was noted in red while the dhcp ones were not

What should I do to fix that?

The ".cshrc" file in FreeBSD has a line like "# A righteous umask". What is a "righetous" umask? Is this a joke? It sounds comical somehow.

just don't use csh

(root's shell switches to sh by default in fbsd 14)

077 is /the/ righteous one

better to set umask in the login class anyway

parv, I beg to differ. On single-user computers, it doesn't matter, and on multiple-user, it comes down to who has physical access to the computer.

path, umask, and environment vars should be set in the login class so that they apply regardless of shell and also to cron jobs

Ok🤷🏽‍♂️

On a multi-user host 077 makes even more sense

umask is only but one, possibly minor, hurdle

"minor" (and effectively nonexistent when others than you have physical access to the computer - or can set the boot device on a VPS) was exactly my point.

Right

About to locally configure poudriere

What should I make URL_BASE if I don’t have a webserver?

doesn't matter

Oh, so just localhost would work then

it literally does not matter

I see

So if it literally doesn’t matter I might as well just comment it out right

I don't have it in my config file at all

this is mine, if you're curious:

Title: dpaste/UZGoA (Plain Code)

NO_ZFS because my builder vm doesn't use zfs, you wouldn't want that setting if you are using zfs

my FREEBSD_HOST is there to disable fetching of distfiles which I do manually; you'd want to remove that

I use zfs

And why do you think I would like to remove that, exactly? Simplifying the process?

I can probably remove PKG_REPO_META_FILE now too

Damn

for my builds, I fetch all the source files manually with make fetch-recursive in the ports tree. That's a pain in several ways, including the fact that there are bugs in the -recursive part

I see

(bugs which I've fixed locally but which are not yet fixed in the main ports tree)

I installed the ports tree alongside freebsd so i don’t think it should be much of a provlem

if you don't want to mess around with that kind of thing, better to let poudriere fetch the stuff itself

Yeah

RhodiumToad just set up my poudriere.conf

Where should my poudriere package lost be

like, if I set my basefs to /home/user/poudriere, should my packages to make list be there

Hi all, I'm trying to understand why does the free size on my SSD does not increase after I delete / move files. For example, I moved 19 GB to another computer, but the free size (df -m) did not reflect the move and remained the same. I use ZFS as the file system, FreeBSD 13.2. Anyone familiar with this issue?

I found an interesting regression somewhere between FreeBSD 12.1 and 12.4. Got a host that takes 5 hours to boot with a random USB drive plugged-in, but if I go back to 12.1, no such issue.

bolik: do you have any snapshots?

@rtprio: not that I'm aware of (at least not intentionally). How do I see that?

zfs list -t snapshot

yep, I see 5 lines in the output.

how can i get rid of them?

in a safe manner of course. Additionally I probably need to disable the creation of new snapshots (if that's advisable)?

bolik: zfs destroy pool/slice@snapshotname

rtprio: thank you. Will try that

rtprio: the amount of free space looks much more reasonable, thank you for your help

I was about to compile some packages in poudriere for the first time when I got a dependency loop error thrown in

Is there a way for me to quickly resolve these issues? I dont want to spend 40 minutes checking and unchecking what dependencies I want or not want to use

How do I reset options I set with poudriere

Nvm I found out

poudriere options -j ... -C category/port

@otis thanks a lot! Im very new to freebsd and poudriere so im still making a lot of mistakes, any best tips to avoid such errors from happening?

dependency loops in ports should not happen; any such is a bug in the port. I'm guessing you used some non-default port option?

Yep, multiple

I wanted to configure lightdm alongside its gtk greeter and xfce

Both desktop and goodies, and tried to at least disable zsh completions in some build dependencies as well as cups support

disabling zsh completions isn't really worth it unless you actually use zsh _and_ don't want completions

Oh I see, I generally use bash and fish, what should I disable then?

Nothing much unless I dont want bash completions?

generally no need to disable anything. shell completions files do _not_ pull in any dependency on the shell in question (or at least aren't supposed to)

I see

you only need to disable port options if they're bringing in some heavy dependency that you don't want

Ive just forced myself to install lightdm from pkg though

did the message indicate where the loop was?

Yeah lemme uninstall lightdm off pkg and try building with my current setting-

Oh wait I think I deleted the options file

Yeah I removed the directory where all the options were in

lightdm has options DOCS VAPI, both of which look mostly harmless

perhaps it was an option on something else?

Yeah im sure it was xfce related

Involved avif, one python package and I forgot what else

Decided to just build lightdm first

After that I’ll set up xfce and then librewofl

The biggest thing I want to avoid is pulseaudio really

graphics/libavif ?

so here's a trick you can use

I c

Tell me the secrets

you can put port options in make.conf rather than relying on the options dialogs

(note that poudriere has its own set of make.conf files in /usr/local/etc/poudriere.d)

Nice nice,

and where do I store the make.conf files

if you put OPTIONS_UNSET_FORCE=PULSEAUDIO in there, it applies to all ports you build

That same place you mentioned?

Cool

the normal place for make.conf is in /etc, there's where ordinary "make" of a port will look

And then I can just force all packages i can use sndioon there instead right

So just in /etc

poudriere ignores the one in /etc and only uses /usr/local/etc/poudriere.d/make.conf (and variations on that name based on jail name and porttree name)

man poudriere for the full details

you can configure options for individual ports with catname_portname_SET= and catname_portname_UNSET=

Cool!

read /usr/ports/Mk/bsd.options.mk for all the things you can define in make.conf

I guess I’ll start building my ports with poudriere then

I should start building the smaller packages i need in order to get used to it, right? Or can I just go config and build xfce immediately after pourdriere is done compiling lightdm?

up to you

If anyone's got some input on this, it's welcome over here: forums.freebsd.org/threads/hang-on-boot-12-4-not-12-1.89239

Title: Hang on boot, 12.4, not 12.1 | The FreeBSD Forums

I'm fully stumped.

RhodiumToad thanks a lot! Also, for reference, how long should some larger packages build on poudriere?

Such as Firefox and lobreoffice

depends on your cpu speed and whether you enable ALLOW_MAKE_JOBS

let me check my firefox build log, sec

my last firefox build took an hour

not sure how many cores that was on (I build in a VM and don't always allow it all cores), but probably 3 or 4

(of skylake i5-6500 @3.2GHz)

there may well have been other builds running at the same time though

I see… I have a ryzen 5800x and am running on bare metal

So it should be around 2-3 hours right?

also that's without using tmpfs and a little short on RAM, though usually these things are cpu-bound

I think it was around that, but on portmaster

that's just for the firefox port itself, obviously, not all the dependencies which can take a while themselves

I need to fix this crap. I upgraded my router (gateway) from 13.1-STABLE (1301506) to 13.2-STABLE (1302505) and cleaned up configuration files in the process. It has the typical WAN+LAN combo (wan0, lan0). I have an internal DHCP server as well for my LAN. wan0 gets two addresses: one that is DHCP-assigned by my ISP, and another one that I

statically assign. Problem is that at boot the router is works fine for a day, until it renews the DHCP lease on wan0, then it gets an internal address from my LAN DHCP server and stops being able to route properly. I can't remember how I fixed it before (I want to say it was either a rule in /etc/pf.conf or configuration block in

/etc/dhclient.conf, or maybe both), so now I'm trying to figure out a "proper" solution for it.

spork_css: you're booting from ufs, what partition type (MBR or GPT) are the boot drives?

Here are the approaches I can think of off the top of my head: A) Configure /etc/pf.conf to block broadcasting DHCPDISCOVER on lan0, B) Configure /etc/dhclient.conf to ignore lan0 when broadcasting DHCPDISCOVER, C) Configure my DHCP server to ignore DHCPDISCOVER from lan0 MAC address.

_xor: weird, the router ought to be trying to renew the lease from the same server before trying any other, I think

Oh yeah, true. That didn't even occur to me.

Yeah back when I tried portmaster I directly installed llvm and gcc just to cut down the compile times

Hmm, is that recorded in /var/db/dhclient.leases.whatever?

it shouldn't be sending DHCPDISCOVER after the initial boot

it ought to be, I think

let me check

puddinghead: rust is the other main time-consuming one, also it's not unusual to need several llvms

Yeah seems about right

But I guess since im already building lightdm I should just build xfce next isn’t it

All I want to do is to disable cups support, mousepad and a decent amount of the goodies

…although at that point im prob better off installing the goodies and extensions i want one by one

RhodiumToad: gpt

RhodiumToad: So one thing I did during the upgrade is that I renamed my interfaces to be by convention, which in this case means em0 became eth0 and em1 became eth1 (I just called them wan0 and lan0 above to be clear). I just noticed there's /var/db/dhclient.leases.em0 and /var/db/leases.eth0.

RhodiumToad: this is the same server we discussed a month or so ago, btw. :)

spork_css: what I'd do at this point is boot 12.1 and replace the partcode on the boot drive's freebsd-boot partitions with gptboot rather than gptzfsboot

RhodiumToad: I'm going to remove them and reboot anyway, but do you see that potentially tripping it up? Haven't looked at rc to see when netif manages the rename, but could it be somehow trying to load both?

spork_css: and see where that gets you

(brb, going to get disconnected, rebooting router)

spork_css: (it might just defer the hang until loader is running, but...)

spork_css: oh, it might already be in loader at that point. hmm. maybe replacing it with a non-zfs loader build would help.

RhodiumToad: I just tested some additional ISOs and 12.1 boots fine, 12.2 boots fine, 12.3 ISO and 12.4 ISO both hang.

spork_css: how about 13.2 ?

That's ambitous!

RhodiumToad: going to boot 12.2 one more time because a message in the boot process looked interesting and I missed capturing a screenshot by a millisecond

I also noticed thwt thunar doesnt seem to have the volume manager plugin in xfce4-goodies

does this mean it domes with the default way thunar is packaged on freebsd?

RhodiumToad: i.imgur.com/FuyIW4X.jpg - this is interesting - the 12.2 loader is a bit more verbose. the zpool it's complaining about does not exist anymore.

puddinghead: I think thunar's volume manager is not supported on freebsd (depends on some linuxism)

I see

puddinghead: but that's from googling, I have never tried it myself

I’ve been struggling with moving my external hard drive’s contents to freebsd myself

how is it formatted?

Ext4

there's a fuse filesystem for ext* in ports (don't try and use the builtin one)

I suggest mounting it read-only in any case, for safety

Yeah I tried the builtin one

Said it needed journaling because of “huge files”

And yeah im gonna mount it read only

the builtin one supports ext2 fairly well I here, ext3 less so, and ext4 only barely (not well enough for usability)

so go with the fuse one

Seems about right

Surprised on why I didn’t back it up as ext2 back when I backed up my Linux stuff, and its just a 1tb drive

RhodiumToad: and in 12.2 a "zpool import" shows me this, and I'm lost on where it's getting that pool info from (specifically "hdpool" which does not exist, the drives that it was on are gone, new, blank hds are in their place) i.imgur.com/syhNYeP.jpg

spork_css: zfs's insistence on tasting everything means that it can discover labels for vdevs and zpools that are at different offsets from the real ones due to partitioning

e.g. da0 vs. da0p1 will have stuff offset by 40 sectors or whatever

are you sure that no disks from "hdpool" are on the system now?

RhodiumToad: Guys went to the colo and pulled them yesterday, popped in new ones. I had them double-check, old ones were HGST, new are WD. I can re-re-check that they're WDs...

RhodiumToad: also 13.2 booted fine. It had some messages during boot, trying to screen record now so I can actually read them

I'm pretty sure it'd be ok to steal 13.2's /boot/loader and use it

Not sure if it applies here, but I know ZFS keeps multiple copies of disk/partition metadata, so maybe that data is still there in other drives in the pool. I've had that happened to me before with similar results where it sees an old pool that's not there anymore. I think it's mentioned in one of the zfs-* man pages (though I just checked

zpool-labelclear and it doesn't have details) or maybe gpart.

zfs's assumption that it can discover everything by tasting is one feature that I'm not entirely comfortable with

Okay, finished building lightdm with poudriere

now how do I install the package

did poudriere say it was creating a repository?

RhodiumToad: Referring to feature flags or something else?

_xor: I mean things like device paths, partitioning, etc

oh

Yeah iirc

RhodiumToad: this bit about zfs is on-screen for like 2 frames, but looks like the same error as 12.2 i.imgur.com/jVU3B5H.png

Just moved to that repository, what should be done after that?

puddinghead: whatever dir you configured poudriere to put the repo in is now a repo. You can configure pkg to use it by creating a file in /usr/local/etc/pkg/repos/*

or, you can use the quick and dirty way, and do pkg add filename.pkg

Damn

Does it come with any risks

no

it doesn't do the dependency graph stuff that pkg install does, but it doesn't clobber anything

So I can just use pkg add lightdm and nothing will go wrong right

Inside the directory where the pkgs are located in

the All/ subdir, yes

RhodiumToad: I was also planning to use my unplanned downtime of this server to convert to zfs on root, it's one of 2 machines that I run that uses UFS. I think I'll try installing 12.2 w/zfs on root, then updating to 12.4 w/o replacing boot blocks and see what happens.

it will add any dependencies that are not already installed and which exist as packages in that dir

if anything fails, it won't do anything that can't be undone with pkg delete

RhodiumToad all/ subdirectory?

whatever, the directory with all the *.pkg files in

Yeah I found out

spork_css: What's bootfs set to on the zpool?

@RhodiumToad what’s the config you use for poudriere package autoinstall conf?

I feel like id rather do pkg autoremove than check every dependency manually after pkg add

when you do pkg add foo.pkg, then foo has its autoremove flag off, but anything that pkg add added as a dependency has the autoremove flag on

I see

pkg query -e '%a==0' '%n-%v %o' will list packages that have autoremove off, change to ==1 to list ones with it on

I personally do not use autoremove at all because I install using metaports, so almost everything gets %a==1

I see

_xor: not sure because the pool doesn't actually exist. :) I have another pool of SSDs (I boot off UFS) and I'm booted into the 13.2 install DVD right now - is there a way without importing the ssdpool to query the bootfs property?

Not sure actually. I don't think so with zpool, but possibly with zdb (though I don't know how).

I am now building xfce

But I want to clean up the packages I dont need in order to free up space after installing them, what would y’all recommend me to do for that

honestly its usually easier to keep them

Yeah seems about right

And then update them right?

saves time if you need to rebuild anything

I c

Should I keep the /usr/ports folder I got when I first installed freebsd or nah?

poudriere will rebuild as needed, e.g. if you change options or update the ports tree

if you want to keep up to date, then you'll need to fetch the ports tree regularly

I don't know what options currently exist to do that, I personally do it with git

I see… if its through git, what do you use? Git pull?

yes

So its just git pull then some command to tell poudreire to update everything in bulk

clone the ports tree (shallowly if you like), checkout the main branch or the latest quarterly branch, and pull as needed

if you then run poudriere with a list of your desired ports, it'll rebuild anything that changed (and only that)

(though it'll also rebuild dependents of anything that changed)

I see

How I could automate the list of desired ports though

Im too lazy to check manually

put a list of the ports you actually want to use in a file and use that :-)

I see

Thanks

Where can I find the oorts tree poudriere uses though? I think ive forgotten it

The same main directory the jails are located it

poudriere uses whatever you specified with poudriere ports

I see

So its a conf file thing

poudriere ports -l will list ports trees known to poudriere

or -lv to see more detail

poudriere ports actually has an option to get the tree from git

in which case I assume poudriere ports -u will do the pull for you (haven't checked)

yes, it does

Heh, I had a one-liner somewhere that used lsof + miller (or frawk?) to spit that out.

caveat: poudriere does a shallow clone. if you force push to your ports repo, it'll unshallow it

There's a switch to change that too.

why want you know the highest fd currently open?

I'm improving on posix_spawn. part of that is a higher-level wrapper that, amongst other things, defaults to closing all open fds > 2 except as specified by a list of fds to process otherwise

what I want is a way to put a size bound on that list, without limiting the ability to inherit an fd which is (a) currently open, and (b) above the current fd soft limit

(just closing the fds is not an issue, we have closefrom and close_range for that)