Did I say zfs import? My bad. I meant "zpool import". Sorry about that.

rwp: it seems unlikely filesystem corruption would cause CAM/AHCI errors, no?

unixwitch, True. Agreed.

and no need for zfs import here since their root filesystem (the one with the problem) is UFS

They might want access to the data now from the live boot disk. I would. I could then copy things off from it.

The only things I've done with it recently were copy lots of bits off of the zfs mirror to my Mac…

…and before that, upgrade stuff, maybe to 14.0.

I don't think / has anything on it that I haven't copied to it from my main machine.

Instead of buying a new disk, singular, perhaps it is time to buy two disks and then you could mirror them, avoiding this failure problem in the future. :-)

Like, my FreeBSD-specific fish history isn't _that_ valuable.

You said you had root on UFS and the ZFS array was data that was elsewhere. If the root disk is dead then all of / the OS part is gone. And the ZFS array would have whatever you put on it.

Right. My giant pool full of backups is OK. I assume.

an alternative would be, if you already have a mirrored zfs pool, you could just put the root on there rather than using a separate UFS root

It depends upon how those disks are partitioned, and if the hardware boots UEFI or not.

I have a question about the base vi/nvi. From vi(1): Ctrl+W "Switch to the next lower screen in the window, ...". What does this mean exactly?

It's a newfangled ProLiant MicroServer. The new boot screen takes like three minutes to totally boot. It's…intimidating for a non-sysadmin like me.

I would buy 2x Samsung EVO SSDs of the smallest size you can get, probably 250GB now, for US$50 each and mirror them.

base vi does not support split editing, correct?

topcat001, Amazingly vi has split windows now.

oh really! :split does not work though

what about :hsplit or :vsplit

You are talking with an Emacs user here so I will have to read the vi man page now.

well, NORMALLY I'd just boot up my FreeBSD machine and try the base vi, but…

amazingly vsplit works! However I cannot find doc for it

hsplit does not work

The man page says "vs[plit] [+cmd] [file] Edit a different file by opening a new screen to the right of the current screen.

I know real vim had only one kind of split for a while.

I found it now. I was searching for "split" :)

thanks

Come to the Dark Emacs side. We have cookies! :-)

oh I have used emacs for a long time too :D

adiabatic, You only have this one FreeBSD system? (Shock! Horror! Nausea!) Create a VM on your other system or something!

I used to have a Digital Ocean droplet running FreeBSD but then I realized I wasn't using anything fancy about it so I dropped back to a normal boring webhost

The vi man page is a little confusing because to me it does not seem to use consistent terminology dealing with split screens, or windows, or buffers, or whatever. Makes it hard to search just for commands that deal with multiple screens. Have to read each and every word, determine what it is applying to, decide it is not multiple buffers, and move on to the next item.

now ctrl-w works as expected. It is quite interesting that only vertical is supported, and I am glad splits are at all possible

adiabatic, Not even on your own desktop or laptop? No need for a cloud host.

I got Debian working on my old MacBook Pro but I realized I spent more time updating it than using it

so I gave it away

topcat001, It's emacs envy. Can vi emulate emacs yet? Emacs has been able to emulate vi for decades. :-)

Although now that you mention it, my webhost provides chroot jails from FreeBSD…

`uname` says `FreeBSD`.

well rwp technically emacs has many more features :)

I actually searched for screen as well but since it has many hits I missed the vsplit. Another pair of eyes helps :)

Curious what "uname -mr" says there.

Me too!

13.2-RELEASE-p4 amd64

So mostly up to date. Pretty recent. Close enough when every reboot annoys your paying customer base.

I troll vi/vim users but I used vi for a long time in the past. I never picked up vim and vim has different keys for many things over vi and I trip over the differences.

Which actually encouraged me to use vim less, because it just wasn't the same and I tripped over things with it. And the colors they chose by default were abhorrent!

I certainly have emacs envy for it's extensibility, however I couldn't get used to its key chording sadly.

Here, $EDITOR is /usr/bin/vim, my go-to terminal editor is Helix, but my main editor is 21st-century Emacs (Visual Studio Code).

Except Electron based editors would be "eight gigabytes and constantly swapping" :)

That's one of the reasons why I call it "21st-century Emacs" — it's highly extensible bloatware.

except it consumes 260MB here and does not swap anything

When I put a new hard drive in my laptop why didn't freebsd or gpart recognize it ? I only saw it when I turned it into an mbr ?

Camcontrol saw it

topcat001, On 14 here, both "split" & "vsplit" work to split the buffer in 2 directions

jb1277976: new disks don't usually come with a partition table, you have to create one yourself (e.g. with gpart). freebsd should still recognise it though, e.g. in 'camcontrol devlist'

(if freebsd didn't recognise it, you wouldn't be able to create the partition table...)

topcat001, s/buffer/window/ # rwp, hold your horsey there. I am myself not consistent with the terminology

unixwitch thanks. i used diskinfo -v and installed lsblk i got it. love this os

to test i attached ada1 which is my m.2 sata ssd i just installed in to my pool and boom it reslived really quick lol zfs dosen't play

s/reslived/resilvered/

How do I list dependencies for a port?

CrtxReavr, Run "make pretty-print-run-depends-list"; s/run/build/ as needed

parv, there any way to make that print packages names, instead?

I need a port with no package, but I want to install packages for the dependencies.

There is some "make" variable to install packages instead of compiling ports that I do not remember

tmux pkgconf libvent

whoops

CrtxReavr, See /usr/ports/Mk/bsd.port.mk; search for USE_PACKAGE_DEPENDS & USE_PACKAGE_DEPENDS_ONLY

rwp: good news — $50 Samsung 870 EVOs are 500 GB, not half that

CrtxReavr, As for printing depend-package list, shove the output of ""make pretty-print-run-depends-list" to "pkg search": make pretty-print-run-depends-list | sed -E 's/^[^"]+"//; s/" to run.$//; s/ /\n/g' | sort | xargs -n 1 pkg search

hi! bit of a funny question, but... is ng_tty(4) nonfunctional since the "new" TTY layer in 8.0?

and if not, how can you install the NETGRAPHDISC line discipline on a tty? tty(4) says that TIOCSETD is nonfunctional since 8.0

Hi all! I have a FreeBSD 14.0 CURRENT host that I installed a few weeks ago. I am still doing some tests on it - I noticed tyhat I cannot install packages via `pkg` anymore, even ones that I know exist. What happened?

jail question.. I set one up as a vnet jail using the directions in the handbook, and it can't reach the host - and only the host. it can ping other devices on my LAN and the internet in general.

deepthaw: does the host have an address on the bridge?

it looks like the handbook instructions are missing a step - they add IP settings to the bridge in rc.conf, but forget to apply them right now

oddline: would that be bridge0? looks like the same as re0 (192.168.1.66)

oddly - I just restarted wireguard on the jail and now it can ping the host. :\

ah, you've got more complicated networking happening inside the jail then?

Yeah. The jail runs a wireguard vpn. I'm trying to mount an NFS share that's on the host.

i'll say this - i'm learning a lot about networking

ok - I'm a little rusty on my freebsd networking, so take this with a grain of salt, but I think you shouldn't have an address set on both the bridge _and_ the bridge's member interface (which you say is re0 for you?)

maybe having it set on the member too is harmless and does nothing, idk

I think I have all my jail networking figured out even if a lot more interfaces have the same IP address than I'd like.

And I neded up realizing I can't mount nfs within the jail anyways so I swapped to nullfs

deepthaw: the IP address should only be on the bridge interface, not the member interface. having it on both will probably cause strange problems

ifconfig_re0="up"; ifconfig_bridge0="inet 1.2.3.4/24 addm re0 up"

If I don't have it on the epair20a interface, it's very inconsistent as to whether the jail can talk to it. But since I'm not using NFS anymore, it's not relevant so I'll make sure both interfaces don't have the ip.

the problem must be elsewhere, because that definitely isn't required for jail<>host communication

Yeah - I suspect it's a pretty delicate setup already. It's a vnet jail running wireguard, which was copied from a different physical server.

but now that it works i won't touch it and will just revisit it later

hi

for some reason i was in the wrong room

i’m unable to boot x isn’t starting and i’m getting all kinds of odd errors

how can i get to a prompt so i can see what is going on?

You need gpu drivers

Also, ctrl+alt+F2

should get you to tty2

There you have terminalismus secondus

Greetings. I want to dualboot Windows 10 and FreeBSD 14.0-RELEASE. I have installed Windows 10 first. I booted FreeBSD with my USB stick, I went into the "LIVE CD", but I was not able to extend the EFI partition(Created by Windows's installer).

How can I extend it? I have ~600GB left.

pr-asadi: why you want to enlarge the EFI partition? Windows normally makes a 100MB EFI partition, the FreeBSD EFI loader is <1MB.

elirco: I know it is enough. I just want to know why I could not, besides having ~600GB space left.

pr-asadi: maybe there is no way to change that. it is quite common that the Windows installer decides what is "best for you" :)

elirco: Hmm. I am gonna remove Windows 10 and install FreeBSD 14.0-RELEASE first. Then I will see it is about Windows or not.

afaik FreeBSD creates a smaller EFI partiton if you install it first.

the real question is: what are you trying to achieve?

To know why it did not let me. gpart(1) says "No space left on device"

But I have ~600GB space left.

I can shrink it, but cannot extend it.

most probably there already is a partition right after the EFI partition (i.e. a Windows partition).

There is.

In that case you have to move the other partition first and create space for extending the EFI partition.

it is possible, but a more risky operation then just enlarging a partition at the end.

Is it the same in all OSes? I remember I could extend my partition on Antix Linux.

you cannot extend a partition if the space after it is already allocated to a different partition

it requires shrinking AND moving the other partition

Knoppix should be able to to this easily

with a GUI even

I'd take a backup though before moving NTFS partitions. That always can go wrong, you never know.

s/Knoppix/GParted liuve CD/

if it is a fresh install it would probably be faster to reinstall Windows

Remilia: elirco: Well, I cannot enlarge/extend a partion when there is another partition after it? It's the same on all OSes?

pr-asadi: obviously it is?

how would you be able to without shrinking and moving the partition right after?

there is a whole lot of metadata in there and partitions were not even designed to be moved

There are Tools like "GParted" that allow you to resize it anyhow, but GParted internally splits that "resize" into "Resize Windows partition, move Windows partition, enlarge EFI partition.

Well. I did not know this. Or perhaps I had just forgotten it. Thank you Remilia, elicro.

I won't resize it, I just wanted to know why I could not. But I found out by your help.

elicro: ^

pr-asadi: MBR and GPT partitions must to be continuous and they contain metadata in OS/filesystem-specific locations

even if the OS tells you there is 600 GB of free space, that free space is inside that partition

Remilia: MBR and GPT... -> Understood. Thanks for the information.

you often can quickly and easily shrink a parittion, which will give you space *after* that partition, but to expand a preceding one you have to move *every single block* of that shrunk partition over to the end

some tools know to skip unused blocks

but in your case, with a fresh install of Windows, it would most likely be faster to reinstall and specify partitions manually

while GPArted can do that moving of *every single block*, it may take hours to complete and if you are unlucky you need to repair your Windows installation afterwards.

so +1 for fresh install

(note that Windows needs a recovery partition)

elirco: OH yes, that's why I could extend a partition on Antix Linux, GParted moved those blocks.

It took a few hours though.

pr-asadi: that explains a lot

elirco: I have also another question. I want to make my home separate. I should make it `/usr/home`, or just `/home`?

see the man page for hier

and also ls -l /

namely /home is a symbolic link

I know /home is a symbolic link.

I had used FreeBSD.

do you plan to replace /usr/home with a symbolic link then?

But, I want to make it separate. First time, I created "/", "/var", "/tmp", "/usr". My home was under "/usr".

I do not want it to be under "/usr" this time. Should I make a partition with "/usr/home" mountpoint or "/home" mountpoint?

so you want a separate partition for home, right?

Yes.

which filesystem?

UFS

will / also be UFS?

or ZFS?

either you use /home, deleting the existing symlink and /usr/home and creating /usr/home symlink

or you do not do anything, choosing separate home partition in the guided partitioner in bsdinstall

elirco: All partitions are UFS>

unless that option was removed (sysinstall had that)

Remilia: I have not seen "separate home partition" in the installer yet.

see my line above, starting with 'either'

I'd use the FreeBSD installer with "manual partitioning" create 2 partiton with labels "root" and "home" and mount points "/" and "/usr/home"

huh that is almost like single root

Guided partitioning used to require the whole disk, is that still the case?

I have not tried it on systems with existing partitions

elirco: Well. My partions would be "/", "/tmp", "/var", "/usr", "/usr/home" then. Is it correct? All of them UFS.

elirco: Yes. AFAIK it needs the whole disk.

I just do not understand the desire to mount at /home when you can mount at /usr/home and leave the /home symlink intact

Remilia: The Traditional Split File System Partitions are "/", "/var", "/tmp", "/usr". Home is under /usr by default

I want to make it a separate partition. I just want to know, should I make the separate partition's mountpoint either "/home", or "/usr/home".

as of 14.0, the installer will use /home rather than using /usr/home and symlinking to /home

[11:01:41] <Remilia> either you use /home, deleting the existing symlink and /usr/home and creating /usr/home symlink

or you use /usr/home and do nothing

up to you

+1 for /usr/home

you can also mount it at /boot/usr/var/log/tmp/home

and create two symlinks

or at /123/456/789/this/is/not/really/home

might be fun too

the enemy will never know where to find it

putting home directories in /usr was a bit of a hack from back in the days when /usr would often be a separate, larger partition

from the changelog for 14.0:

> The pw(8) and bsdinstall(8) programs now create home directories for users in /home by default rather than /usr/home. The default symbolic link for /home, referencing /usr/home, is no longer created. bbb2d2ce4220

you can still use either, but /home imo makes more sense, especially if you're making it a separate partition

Remilia: It seems I could not explain well. I did not mean that. I mean, by default if you do not make a separate home partition, the home will be /usr/home and /home would be a symlink to /usr/home. I just want to make it a separate partition. I just need to know which mountpoint should I give to my separate home partition.

pr-asadi: the point people are trying to make is that doing it either way will work just fine

oddline: Thanks for the information you shared.

I suggest mounting it directly at /home. others have suggested the other way (mount at /usr/home and keep the symlink).

oddline: the point people are... -> Understood.

elirco: Remilia: oddline: Thank you all

you can mount it anywhere, symlinks are for backwards compatibility and /home or /usr/home are locations typically checked for home directories

actual user home path is obtained from the password database

Hi

i'm planning to upgrade to 14

Remilia: Understood. Thank you for your time.

does the upgrade process automatically create a boot env ?

or i need to create it manually ?

freebsd-update will create boot environments automatically for each step

ok, so if anything goes wrong, i can revert

are boot environments enabled as default?

i never used them

but I receive the following

elirco: try "bectl list"

root@elfreebsdo1:~ # bectl list

libbe_init("") failed.

elirco, do you use zfs ?

as rootfs ?

yes

then i don't know, i don't remember enabling anything to get it work

hmmm...

i will check on my other PC if the bahvior is the same

Title: BootEnvironments - FreeBSD Wiki

it seems that you use a non standard dataset layout

that bectl cannot manage

'libbe_init("") failed' sounds like a userland issue

oh that is also possible

elirco: try "bectl check"

root@elfreebsdo1:~ # bectl check

root@elfreebsdo1:~ # echo $?

255

my other laptop has the same problem

both are dual boot with Win and FBSD

here is my partition layout: bsd.to/YwFp

Title: dpaste/YwFp (Plain Text)

dataset layout ?

freebsd version ?

how did you dualboot ? which bootloader are you using ?

root@elfreebsdo1:~ # zpool list

NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT

root 816G 56.2G 760G - - 3% 6% 1.00x ONLINE -

root@elfreebsdo1:~ # freebsd-version -kru

14.0-RELEASE-p3

14.0-RELEASE-p3

14.0-RELEASE-p3

Bootloader is loader1.efi

root@elfreebsdo1:~ # ls -lh /boot/efi/EFI/freebsd/loader.efi

-rwxr-xr-x 1 root wheel 645K Nov 21 17:29 /boot/efi/EFI/freebsd/loader.efi

root@elfreebsdo1:~ #freebsd

elirco: for boot environments the GPT layout does not matter at all

what does zfs list show?

they use ZFS datasets

tsoome: bsd.to/gZLM

Title: dpaste/gZLM (Plain Text)

what's the output of `mount`?

output of mount: bsd.to/fg8A

Title: dpaste/fg8A (Plain Text)

zpool get bootfs ?

root@elfreebsdo1:~ # zpool get bootfs

NAME PROPERTY VALUE SOURCE

root bootfs - default

and zpool list ?

root@elfreebsdo1:~ # zpool list

NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT

root 816G 56.2G 760G - - 3% 6% 1.00x ONLINE -

btw my other dual boot laptop has the same issue. so it may be the way I am setting up dual boot. (despite the fact i do it pretty straight forward)

well, the beadm/bectl is expecting dataset layout <poolname>/ROOT/bename - where bename is custom name ("default" is used by installer) and poolname is also custom ("zroot" by installer).

how are those datasets created? shouldn't the installer do this?

then bootfs zpool property is set to "<poolname>/ROOT/bename" -- this property is used by boot loader

the installer *does* this

installer should, yes

when using guided partitioning

is it a problem i chose "root" and not "zroot" as name for my pool?

no

but if you upgraded from before BEs were a thing

the problem is that you do not have ROOT dataset for root file system

you appear to have OS files stored directly to pool root

can guided partitioning do dual boot now? I use "manual partitioning" since ever since that has been the only way for dual boot to work.

tsoome: i set up the device the way I do it since UEFI exists. so obviously due to not being aware of boot environments caused me doing it wrong.

what you need to do, is boot from alternate media, import your root pool and use alternate mountpoint; then zfs create -p root/ROOT/default and copy your OS data to root/ROOT/default

then you can use bectl to activate that be and create new one etc.

"copy you OS data"... how would be the best way to do it? tarcp?

and wouln't I have all files twice then?

once you are happy with setup, you can clean up the pool root

you can use `zfs send`

it will be much faster than tar

what is zfs send? can i add additional hard drive with that?

and what if i have 128gb ssd

can i only make redundant raid with another 128gb one?

I will check that out in the evening (CET). now it is time for a walk in the sun at outside.com/realworld soon. :)

or can it be 200gb hdd? with72 gb wasted?

the os itself does not take much

i have 128gb with most space used

can i transfer everything to 200gb hdd for example?

I guess you have around 100GB of other data there;)

sure

> now it is time for a walk in the sun ← I agree, I should go for a walk and buy tea in the process

you absolutely can transfer everything to new disk - even with simple mirror setup - and once you detach smaller disk from mirror, you can resize partition on larger disk and grow your pool

cooooool

once your pool is grown, you can not attach smaller disk any more, of course

but what about several smaller disks?

like 200gb one disk and 128+80 gb another?

mirror does map blocks 1-1

therefore, for redundancy, 200GB disk needs another 200GB disk

that's only one thing which is sad

% zpool list

NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT

zroot 103G 97.1G 5.87G - - 64% 94% 1.00x ONLINE -

hmmmm

i need some sort of zroot1 now?

V-T60: to use a 200GB disk + 120 + 80 GB disks, would would need to stripe (or concat) the 120+80GB disks and then mirror on top of the stripe, which is sometimes called raid0+1. this is almost never useful or desirable (except for unusual edge cases like this) so for simplicity, zfs doesn't allow it, you can only stripe over a set of mirrors (raid10)

you could probably hack it up with gconcat if you *really* wanted...

Hi! I'm replacing a faulty disk in my raidz pool. I had four disks in it (now three, since one is now "offline")

one thing confuses me a bit though

zpool status zroot shows me ada{0..3}p4

why the p4?

how do i mirror my 120gb from zero?

unixwitch: 128gb + my 200gb unused HDD

(as i understand, 128gb part of it for the begining)

sjk: The default layout installs efi on p1, freebsd-boot on p2, freebsd-swap on p3 and freebsd-zfs on p4

beginning*

than i wanna "grow" it somehow as i understand what was proposed before

"resize partition on larger disk and grow your pool"

I'm in the process of upgrading from 13.2 to 14.0 (RELEASE). I had Poudriere build the new packages, but when running pkg upgrade there was an error after [170/469] packages were fetched. Now I can'

fxttr: ah, so I'd have to recreate that layout on my replacement disk before running zpool replace?

e.g. gpart backup ada0 | gpart restore ada3

Now I can't finish the install as I get "pkg: An error occured while fetching package"

and "DBG(1)[87844]> release an advisory lock on a database'

unixwitch: gstripe would work

How do I go about releasing the advisory lock?

as an underlying device

but that would kill bootability in case the larger drive goes out

sjk: aside from gpart backup/restore you will likely want to copy the other partitions; the naming suggests you have GPT which means you want to at least copy the ESP over

(but you do not need to copy swap contents of course)

Remilia: ah yes indeed, I do want to be able to boot from it etc. What's the proper way to go about this? dd?

sjk: you can dd the ESP itself after creating the layoyut, sure

layout*

but I am curious as to why you have *4* EFI partitions

5 seemed a bit much

what

sorry

I mean, typically you would have 3

I am not sure either

ESP, ZFS pool, swap

do you have an old install with a separate bootpool?

gpart show -> efi, freebsd-boot, freebsd-swap ond freebsd-zfs

Schamschula: 'release an advisory lock' is a debug message, you can ignore it if I recall right

oh

not that old, hmm, maybe three years?

oh wait, actually not even that, more like 1-2 years old

I am guessing you boot in EFI mode but during installation you chose support for legacy BIOS?

that explains it

not sure!

is it bad?

no

phew :)

it can sometimes cause your system to catch fire (this is a lie, it cannot)

sjk: does `zpool list` show bootpool OR is there bootpool in `zpool import` output?

unixwitch: :D

Remilia: zpool list only shows zroot

zpool import says there are no pools available to import

then you are probably fine

so I'm guessing I'd want to copy efi and freebsd-boot over to the new drive?

probably a good idea

dd if=/dev/ada0p{0,1} of=/dev/ada3p{0,1}?

dd does not work like that

you have to invoke it twice

aye

honestly I am conflicted about freebsd-boot

what is its size?

is freebsd-boot a UFS filesystem or something else? (if so, it might make sense to use gmirror... but i can't remember how this works since i haven't had a separate boot partition anywhere for ages)

it is not ufs

It's 512K

oh

yeah that is legacy bootcode

efi 260M, boot 512K, swap 2G and zfs 5.5T

for non-EFI booting

you can dd it over

I set this up through the default freebsd installer

yes, the BIOS+UEFI mode

so, it's there in case the machine does not support efi booting?

yep

ah ah

though in *that* case you also need to install bootcode

pmbr + gptzfsboot

Remilia: Thanks. So the issue seems to be one of fetching. I get a 403 error, even though the file is there. It looks like there is a rule in niginx.conf that matches!

sjk: check the man page for gptzfsboot for instructions if you want to install bootcode and protective mbr

Schamschula: does it happen for pkg itself?

No it didn't

so you were able to bootstrap pkg?

Yes

403 is intriguing but I do not use nginx haha

I tend to deny a number of files re. script kiddies et al.

for poudriere I use h2o with this: paste.ee/p/HOULv

Title: Paste.ee - View paste HOULv

and it works fine except I never solved the pkg bootstrap issue

so I just manually upgrade pkg first

<Schamschula> I tend to deny a number of files re. script kiddies et al. ← I do not see a reason to block anything within poudriere because it is all static files with nothing of interest

That was never the plan. I found the rule. It was missing a leading ^

Thanks guys!

hi there, how can I specify set of flavors for specific port in poudriere ?

list them in your build list

Remilia, Could you please show me a syntax sample

I am sorry I do not understand

graphics/pecl-imagick@php81

list the origin twice with different flavours

this will build only graphics/pecl-imagick for php 8.1

yes

i need this for 8.2 and 8.0 too

what prevents you from adding graphics/pecl-imagick@php80 and graphics/pecl-imagick@php82 to your list?

just separately list them ?

I am sorry I do not understand, are you using a text preprocessor?

because flavours are *different* packages

Now it's clear

thanks

I thought it require some special syntax something like: graphics/pecl-imagick[php80,php81] ...

a single poudriere build job will only build one package

it is impossible to build at once for several php versions, as you are linking against different headers etc.

Yes, but I didn't understood the methodology for adding multiple flavor packages

thanks

nerozero: there is one special feature for flavours, you can use @all to build all of them

Remilia, Thank you! Can it be the case that I have tried with all caps @ALL and it didn't worked ...

flavours are always lowercase; uppercase is options

yah, my bad, thanks for make this clear to me!

nerozero: poudriere% bulk lunatear devel/php-composer@all \\ Building lang/php81 | php81-8.1.26 \\ Building lang/php80 | php80-8.0.30 \\ Building lang/php83 | php83-8.3.0

works fine

Thank you

is there a command to test if all system files have the correct expected hashsum?

something like `dpkg --verify` on debian

Uh ‘freebsd-update ids`

iirc

any openvpn users here?

Title: FreeBSD OpenVPN low upload speed | The FreeBSD Forums

Hello!

evening

so if i don’t have ports installed how do i add it since portsnap has been removed

is it safe to use pkg just to install git?

have you tried bsdinstall ?

Im spitballing here as im learning at a helluva speed catching up

what’s that? the installer? i didn’t install with the system because it seems like when i did that last time it messed stuff up

aybe just stick with pkg for now. i have to say the documentation for freebsd is horrible lots of stuff in the handbook is wrong

maybe*

Wat

Im messing around with updating a port. other thing you can do, if you have git installed just grab from github

like with this it tells you to use pkg to install git to install ports but is that safe since now you use have to update git with pkg instead of using a port tree?

cutler that’s my catch 22 i don’t have git

Pkg is the default package manager. If you can install something using pkg and don't need to compile it with some flags for some reason use the pkg

It's the choice between using binary packages compiled by freebsd maintainers or compiling everything from source.

It also installs from "checked/accepted ports"

Something like git im guessing would be Ok.

i did read somewhere that if you used both ports and pkg it can screw things up as you have two different upstreams

i guess i’ll just stick with pkg

ports tree you compile it yourself but not everything there necessarily works properly. If it has a pkg version then it works properly but you might want to remove or add something at compile time, then you do ports tree and fix it

Yes

ok

But I agree about it being confusing for beginners about the ports, portstree and pkg (since it includes ports as well)

its been a bit confusing.

but can honestly say I have leraned more about the internal in FreeBSD than the last 4 years in Linux

*Learned

I meant only the ports/pkg/ports tree part is

Otherwise it is good documentation

Jip

I mean, look at pacman and AUR, very clearly documented that you should prefer pacman, AUR is risky but if you have no other choice use it.

agree.

And AUR isn't mixed into pacman,

What runs moused by default, since the system default is moused_enable="NO"?

Ditto that question, I am starting to getting annoyed seeing a mouse in my FreeBSD tty

cutler, well, it's a lot more mature and less convoluded.

cutler, plus. . . it's an OS project. . . whereas linux is a kernel from here, a libc from there, a compiler from somewhere else, an init system from yet another place, ad nausium.

It was a welcome realisation when after a just 2 or 3 days I was actually annoyed with Ubuntu that is running my Fiber Laser.

Next step will be Freecad and some recompiling and my CNC will move to FreeBSD as well

I doubt I will get the Fiber Lazer guys from Lightburn to make a FreeBSD build

but the other machines will move - soon

If the src is available and sane with a Makefile, it might build and work easily.

And even if not, you might be surpised at how much help you can find in getting it working/ported.

There's also the linux subsystem.

It is a closed source product.

github.com/LightBurnSoftware they have a repo tho

Title: LightBurn Software · GitHub

With the update to 14.0 I had to leave MySQL 5.7 behind. Unfortunately, an update that I had done twice before, failed this time. I've got duplicate .ibd files. In order to get rid of them I would first have to start mysqld, which, of course it won't.

Why are you rushing to 14.x?

Rushing? I don't have time during the semester so I thought I'd tackle it over winter break.

This is my home server, however I also need to know what I'll run into when upgrading the production server on campus.

Dumb question, why did you have to leave MySQL 5.7 behind ?

Apologies for my ignorance, Im really not up to date with whats happening on the MySQL front

14.0 doesn't provide openssl 1.x.x., hence the package doesn't exist.

Cool. Im following

I get error like [InnoDB] Tablespace ID: 86 = ['mysql/time_zone.ibd', 'sys/mysql/time_zone.ibd']

Have the svnup servers been disabled?: bpa.st/EBLA

Title: View paste EBLA

cutler: Isn't 5.7 past its end of life?

Yes and no. There have been recent upstream updates.

but openssl-1 is EOL. so mysql lagging

IIRC, you can still get support for it by paying for an extended contract. Maybe the MySQL people consider that support enough?

How do I make the /usr/src dir git-able?

I've got git installed.

I've also got the 13.2-R src tree from the install media.

if you git clone .... it is gittable

You want to turn your /usr/src into an entire git? Or you want to git clone to it?

I just want the full 13.2-R security branch src there.

Just a thougt, if you not going to make changes or need source control, just download master as a zip and extract ? hell of a lot easier than mucking about with git

'git pull' wouuld be easy AF if it worked.

so im having weird networking issues and i think i found the issue what is l_magic? it seems like it’s a vlan device which i dont have or use

because neither wifi or ethernet work

ethernwt was working but stopped as well as wifi

so it seems like freebsd thinks im using a virtual adapter or some type or vpn

if it is not on your disk yet git clone will get it and setup what you need from there you checkout the branch you need

CrtxReavr, Zyxer, The console mouse needs separate disables for the USB event system. moused_enable="NO" plus moused_ums0_enable="NO" moused_ums1_enable="NO" for the USB event driven system.

The reason for both ums0 and ums1 is I have a USB mouse plus a Trackpoint keyboard mouse. You would need ums2 as well if your system had a third.

Thank you! I will fix after I do some grocery shopping

rwp, I just didn't need a mouse point on my mostly-headless server console.

s/point/pointer

CrtxReavr, Although I sometimes do use my desktop vt console the mouse pointer there was just a distraction and I didn't need it running there either.

Wifi be unstable on FreeBSD

s/\ on\ FreeBSD//

Schamschula, A time honored strategy is to put legacy services into a legacy jail and continue running it in the jail sandboxed workspace.

You could put your MySQL server into a jail and side step the current problem for a while.

rwp, I just set moused_nondefault_enable="NO" & moused_enable="NO"

moused_nondefault_enable? /me goes to look that one up.

moused_nondefault_enable="YES" is the default.

It's on a USB KVM. . . which might explain something.

Anyways, once the OS is installed and it's on the network. . .

rwp: that thought had crossed my mind.

I'm a bit of a noob in terms of jails. Currently I'm only using them for Poudriere.

Schamschula, It's actually nice having that type of setup when it comes to testing upgrades because one makes a new jail of the newer OS version and then tries it and if it doesn't work it doesn't affect the previous jail, them all being independent.

The one limitation is that the jail should not be newer than the hosting OS because of course system calls from older compiles are known on newer kernels but a newer system may introduce new system calls that the older kernel will not have implemented.

Meaning that though I may want to test out a service on the newest everything I can't do that until the host system is also running it, so I might as well have a newest everything testing machine then.

But for sandboxing up older legacy services it's perfect!

I'm trying to set up a jail by following docs.freebsd.org/en/books/handbook/jails

Title: Chapter 17. Jails and Containers | FreeBSD Documentation Portal

CrtxReavr, (Zyxer) I did not know about moused_nondefault_enable previously. I just tested having only moused_nondefault_enable="NO" and it is a one-stop-shop for disabling the console mouse here. I converted to it. THANKS! :-)

However, when I run freebsd-update -b /usr/local/jails/containers/classic/ fetch install it wants to install 14.0 RELEASE. I want to kee 13.2 RELEASE in the jail.

s/kee/keep/

I would fetch the base.txz of the version I wanted to keep and then use it. You might already have it on your previous 13.2R system in /usr/freebsd-dist/base.txz

OK. That's what I installed. I just want to make sure that when a new patch level comes out that I can update to it.

Looking at the handbook section 17.4.1 now and I see that command but I don't think that should jump forward to a new major release number but just along the patch line. Therefore I think it must already be a 14.0R version already.

My base OS is at 14.0 RELEASE, but I need the jail to remain at 13.2.

Note that you can use "freebsd-version -j jailnamehere" to determine the userland (-u) of that jail. And of course the kernel running (-r) is the host kernel. And nothing installed (-k).

First I need to figure out how to add the .conf file to /etc/jail.conf.d

My hosting system is 13.2R so let me create a 12.4R as a test and try that command on it because as I said I didn't think it should try to jump major releases but just along the patch level.

Schamschula, I reproduce exactly your reported behavior here too. It upgraded the jail to the host release.

When I try to start the jail I get: ifconfig: interface em0 does not exist

Schamschula, D'Oh! I forgot about the freebsd-update --currently-running option! That's the missing piece. Use: freebsd-update --currently-running 12.4-RELEASE -b /jails/testclient26 fetch

Are you trying to run a vnet jail or a non-vnet jail? (I like vnet jails so much that I pretty much always use them.)

I also saw that in the man page! I guess you don't need to specify the version.

A regular thick jail.

Jails can be thick or thin or otherwise and they can be vnet or otherwise. vnet jails have their own network stack entirely and get their own IP address.

I'll assume you are starting a non-vnet jail given the response. Which means that everything will share the hosts IP address space.

Which means that it is important to set sendmail_enable="NONE" in the jail's /etc/rc.conf file or the jail will try to start sendmail on port 25, which is probably already busy, and things like that.

Either would be fine.

I figure out the problem: I was using the wrong interface name. It should have been bge0

rwp: I got that and it doesn't help

The mouse nondefault thing

Schamschula, I have been working through an example here. I needed to do that anyway. Today was just motivation to do it.

You might have completely figured out everything on your end already. If so then awesome! If not I put together this self-contained example. bsd.to/sNkT/raw

Title: sNkT

This is using a vnet jail. I think it makes things simpler. All of the setup and teardown is in the jail.conf file for this example. Because I like to be able to step through the steps in order to understand things.

Zyxer, I removed all other moused_* lines from my /etc/rc.conf file, added moused_nondefault_enable="NO" as suggested, rebooted, and no moused on my vt consoles. Worked for me.

Nope! Not completely there yet. However, the jail is up and running, just nothing installed. I'm off chasing another issue: Mediawiki is giving me 500 errors. However, since it's just for personal use it's running under sqlite3. However, it's not a php vs. sqlite issue, as that's working for my own code.

I see in my paste that I cut off the -p9 and it says 12.4-RELEASE-p but it should say 12.4-RELEASE-p9, honest! I had forgotten that --currently-running was needed before.

Hopefully that example will help jumpstart things. I am just hoping it will be helpful. I am going to be afk for a while. BBIAB.

full of help

One step closer: Mediawiki is back up, so are my notes! (somehow php82-pdo_sqlite didn't get installed).