can I set my git-daemon keeps lurking and if something happens to my sshd it takes over the port?

this libmd sovers bump is *really* way more annoying that it should be with pkgbase

hah

i dealt with that transition years ago :-D

kevans: what? it was only committed like a month ago

was there a make.conf option for it?

no, i did all of that work two years ago and only just now got to land it

=)

so the first problem is the poudriere bug where it claims to upgrade jails but actually doesn't, which i really should report

then even after you get packages rebuilt, you have to 'pkg upgrade -f' everywhere to actually install the fixed ports

and for bonus, 'pkg' will be broken while you do this, so you have to know about pkg-static, which i do, but many users do not

that latter point is true of any soversion bump in base with or iwthout pkgbase

i agree, it's true, but it should not be

it's simply not feasible to identify all of the packages and bump PORTREVISION like you would with an soversion bump in ports

we need a way for ports to declare dependencies on pkgbase packages/libraries

although, I thought bapt was saying somewhere that the hash should've changed and it would pick up that it needed to be reinstalled

maybe that was ideal-world and it was broken

especially now that 15.0 is no stable there is never any need for normal users to run stable/release builds, we need to fix pain points like this in current

s/no stable/so stable/

kevans: while you're here, ever seen a build error like this? 0x0.st/XlCr.txt - i am much confused

funky, but I think I saw a transcript much like this just the other day, actually

i was going to mail current@ but i accidentally deleted ${WORLDTMP} while debugging it which fixed the problem

the sample the other day was building a poudriere jail, sjg noted that you can run -DWITH_META_ERROR_TARGET and bmake will drop the aborted target's .meta file in ${ERROR_LOGDIR}

run it with

'make will exit 6 when it sees an abort token'

oh is this going to be a meta mode problem, maybe i should ask sjg directly

it happened on the second build after i moved the build to a new host though which makes me think i did something wrong

maybe, though from the sound of it he wouldn't be able ot tell you much more without understanding which target precisely failed

sounds like a good opportunity to improve diagnostics

sed -ie 's/@//' Makefile*'

ivy: if you could pick one resource that helped you the most to learn FreeBSD, what would it be?

jauntyd: "Understanding UNIX", second edition, Stan Kelly-Bootle, Sybex, 1994

ty

i would not necessarily recommend this book to modern readers though considering it is 30 years old

i almost bought it just now lol

despite its age i do think it does a good job of introducing the basic Unix userland, which hasn't really changed much in that time

but it won't teach you how to install FreeBSD on your laptop or how to be a sysadmin

i mean, its $10. would you pay $10 for it in modern times knowing what you know about it?

if someone gave me a FreeBSD shell account somewhere and i wanted to learn how to use it, yes. if i wanted to install FreeBSD on my laptop, no

gotcha

i can't really offer any more up to date recommendations since i learn Unix in the 90s when that book was current and i don't get paid to review books about Unix :-)

lol i understand. I have 4 or 5 specifically on FreeBSD, but they just gloss over things

anyways, thanks for sharing your brain :)

any Unix book would be good, I would think. Unix Power Tools from O'Reilly would be a good one.

great! i will add it to my cart as well

I basically learnt by reading the man pages, and then had Unix System Administration Handbook (1st ed), and Unix Power Tools.

Yes I'm dating myself with all that, late 80s and early 90s.

jauntyd: have you used any other Unix type system?

Well, I used FreeBSD before the split....since then off and on. I had a Mac for a few years, but mostly just BSD

I find the books I have lack the details I crave

i remember "Unix Power Tools" being good but it's more of a second good

it's like, okay, i know how to use 'ls' and 'cat', now what?

^

s/second good/second book/

what details are you craving jauntyd ?

the best reason to obtain a copy of "Unix Power Tools" will be the CD-ROM that came with it

that must have a bunch of really interesting ancient software on it

mns: Anything that can help me level up. I am a solid user, but I would like to know how FreeBSD works internally.

ivy: remember user group disks? :D

the first one had roughly 100 pieces of software on it, for 7 different versions of Unix (HPUX, AIX, SunOS, Ultrix, etc.)

jauntyd: if you're looking for internals than the book would be The Design and Implementation of the 4.4 BSD Operating System. There is a version for 4.3 BSD and for FreeBSD

jauntyd: that was out of my baileywick at the time, but i remember Fred Fish and PD libraries

I have lua54 installed. I want to have a /usr/local/bin/lua but doesn't seem like any of the lua packages provide a symlink. I know other packages provide it, like python.

ok build worked this time, let's reboot then we are going to have such fun with sr-iov

ONE HUNDREEEEEEEEEEEEEEEEEEEEEEEED AND TWENTY EIGHT VIRTUAL FUNCTIONS

sr-iov?

mns: sr-iov lets you expose "copies" of a PCI card to virtual machines. so you can connect all your VMs directly to your NIC and their traffic doesn't have to go via the host = better performance

virtual machines as in jails?

no, as in bhyve

ahh ok

(or kvm, if you speak linux)

hah it worked

pci10: <unknown> at device 0.11 (no driver attached)

pci10: <unknown> at device 0.15 (no driver attached)

pci10: <unknown> at device 0.19 (no driver attached)

this never worked with my Intel card, Chelsio is fine though

let's see if it works in the VM though... need to configure the switch first i guess

Oct 23 04:46:43: fatal; pci passthrough not supported on this system (no VT-d or amdvi)

hmm

how do i do PCI passthrough on vm-bhyve, i feel like i'm missing something here

that would be an interesting question for me too, since im just setting up my new server and want to pass through my soundcard

i looked at vm-bhyve.log and it says this

Oct 23 04:50:10: booting

Oct 23 04:50:10: [bhyve options: -c 2 -m 1g -Hwl bootrom,/usr/local/share/uefi-firmware/BHYVE_UEFI.fd -S -U 4d47f56a-6519-4039-9368-d63c0fabeeef -u -S]

Oct 23 04:50:10: [bhyve devices: -s 0,hostbridge -s 31,lpc -s 4:0,nvme,/data/vm/lily/disk0.img,sectorsize=4096 -s 5:0,passthru,10/0/11]

Oct 23 04:50:10: [bhyve console: -l com1,/dev/nmdm-lily.1A]

but this just produces a usage error if i actually try to run it

# bhyve -c 2 -m 1g -Hwl bootrom,/usr/local/share/uefi-firmware/BHYVE_UEFI.fd -S -U 4d47f56a-6519-4039-9368-d63c0fabeeef -u -S -s 0,hostbridge -s 31,lpc -s 4:0,nvme,/data/vm/lily/disk0.img,sectorsize=4096 -s 5:0,passthru,10/0/11 -l com1,/dev/nmdm-lily.1A

Usage: bhyve [-aCDeHhPSuWwxY]

and i really can't work out what i'm doing wrong

even if i disable passthru0 in the config file, the options it prints don't work

maybe this just doesn't work on AMD?

apparently AMD does support this but i need to find ivhd0 device somehow

okay progress: bhyve: Found 1 AMD-Vi/IOMMU device(s), use hw.vmm.amdvi.enable=1 to enable pass-through.

this is a loader tunable though, so another reboot...

holw fuck it works

what file does regdomain attribute locate in freebsd14

root@lily:~ # ifconfig cxlv0

cxlv0: flags=1008802<BROADCAST,SIMPLEX,MULTICAST,LOWER_UP> metric 0 mtu 1500

options=6ec07bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCSUM,TSO4,TSO6,LRO,VLAN_HWTSO,LINKSTATE,RXCSUM_IPV6,TXCSUM_IPV6,HWSTATS,HWRXTSTMP,MEXTPG>

ether 06:44:3f:e7:60:30

media: Ethernet 10Gbase-Twinax <full-duplex> (10Gbase-Twinax <full-duplex,rxpause,txpause>)

ok, it doesn't actually work: cxlv0: a looped back NS message is detected during DAD for fe80:3::444:3fff:fee7:6030. Another DAD probes are being sent.

kevans: hello frand

Waiting for PIDS: 29703 97832 1242 4323

90 second watchdOct 23 06:46:10 hemlock init[1]: /etc/rc.shutdown terminated abnormally, going to single user mode

Oct 23 06:46:10 hemlock syslogd: exiting on signal 15

vmnet8: link state changed to DOWN

vmnet7: link state changed to DOWN

Waiting (max 60 seconds) for system process `vnlru' to stop... done

Waiting (max 60 seconds) for system process `syncer' to stop...

Syncing disks, vnodes remaining... 0 0 0 0 0 0 0 0 0 0 done

All buffers synced.

Uptime: 19m16s

panic: VERIFY(avl_is_empty(&sk->sk_dsl_keys)) failed

i have to say, i've never seen a panic after the 'Uptime:' report, this is now

new

bug @ bugs.freebsd.org/bugzilla/show_bug.cgi?id=282281, hope kevin can produce a fix for this soon

do i get some sort of bonus for finding a cxgbe bug and a zfs bug in the same day? i wasn't even trying

help me kevin evans you are my only hope

installing chromium from pkg in freebsd, i can't use proxy of chromium, i tried "proxychains chrome" firstly , it didn't work, then i tried [chrome --proxy-server="socks://127.0.0.1:10608"

], it still not work

k

Define "does not work"

It errors out? Does not use the proxy at all? Tries but fails?

Alver: yes , and chromium page proxy setting don't work yet ,not setting option, when i click it , then it forward to PC setting panel

exit

Alver: it works , i forget shutdown chromium window opened in the past

ast

I'm trying to manage (create, mount, unmount, destroy) ZFS "legacy" datasets inside a jail. I can create them, but mounting gets "Operation not permitted". Anyone have a hint?

I tried with devfs_ruleset=4, devfs_ruleset=0 - no difference.

ivy: i've always wondered what that last message means when I shutdown my system.

in particular the 10 zeros

Alver: i do not know but there is a book on the topic of ZFS. a whole textbook!

Alver: i have not played with it iun a long time but have taken two approaches: 1.) Mount the zfs on teh host and then nullfs the dataset 2.) You can also "jailed" the data set upon creation and then have a mount point setup

Alver: search.brave.com/search?q=mount+a+Z…p&summary_og=bfa1495c3befe43c13a304

bah

>brave

well i mailed support⊙cc about my cxgbe issue, basically to say hi, i have your card and it doesn't work, will you help me?

i hope they take pity on me

voy4g3r2: that is already done; I can mount the datasets using zfs nicely. But, for some reason, *legacy* mounts - which is what podman uses - does not work

Alver: ah, my area of expertise on that is non-existent.. hopefully someone with more knowledge can help

It's silly because podman uses legacy mounts - no idea why, honestly - and yet those are not working inside a jail, whereas zfs mounts do work nicely.

I read somewhere that vfs.usermount=1 on the host might be needed to allow legacy mounts inside a jail, but no dice.

Alright, reverted the jail to devfs_ruleset=4, securelevel=2 and vfs.usermount=0.

Does anyone have a hint on how I could get "legacy" mount working on a specific ZFS (jailed) dataset inside a jail? I can already create/destroy what I need, but mounting fails when using mount instead of zfs.

hello

hey

i have a ssd with zfs that is not mounted, the pool is called zroot, just as my real zroot on another drive

i need to mount or view this ssd

how can i mount it?

or export it whatever

i just need the data

sopparus: have you tried `zpool import -d tmproot zroot/the-id-of-the-thing` ?

what is the id?

how do i find it

sopparus: `zpool import` should show it

hi, can you display all cores with top? I am reading in linux it is done with -l

uskerine: top -P

uskerine: or configure installing 'htop' and use thst instead, which imo is a better interface

s/configure/consider

thanks

Alver: do you have those allow mount controls set for the jail?

allow.mount and allow.mount.zfs for example

if I get my $WORK stuff finished today or tomorrow, I have a whole day of podman funsies planned... its hard juggling all the things

If I want to create a custom freebsd image bootable from optical or usb drive, what tooling should I look at?

Essentially, I just want it to be able to run a custom shellscriptafter it booted.

mfsbsd

henlo, thanks :3

it might be overkill, but it basically boots DHCP and then can do anything you want if you stuff it with rc.local

phryk: eradman.com/posts/automated-freebsd-install.html has a reasonable walkthrough

AMA

What are the images at mfsbsd.vx.sk and are those required?

dch: yes, they are set. I can mount zfs using the zfs command, but 'mount' fails.

phryk: those are pre-generated ones from a fbsd committer, you can grab that and try it.

Which is even more strange because the jail is also mounting a lot of other things using 'mount' - but zfs does not work

phryk: mfsbsd just boots an in-memory freebsd that supports DHCP and sshd directly. You can do the same thing with memstick image ofc too.

its just you wont have remote access. eradman's one is nice as it runs in a tmux too. I remember doing this for a closed-source customer prior to covid...

Alver: IMO its very confusing to follow your stuff. It's snippets of info but never enough to reproduce your exact steps. Can you post a pithy summary of your setup so I can try it?

dch: mind if i query you?

mounting normal mount via fstab in a jail should only require the jail.conf settings above

phryk: ofc

dch: long story short, I have a zfs dataset which is created with -o jailed=on. The jail.conf settings are set and are working for everything - except when the jail creates something with mountpoint=legacy, and tries to mount it 'anywhere'. Let me see if I can cobble up something which is not an ongoing stream of discoveries :)

;-)

so its the zfs canmount=legacy option that seems to screw things up? that should just allow using fstab, iirc

so long as /dev/zfs is available i would expect that to work

dch: I think the relevant bits are all in pastebin.cz/en/p/BTHEPJf

dch: I cannot use 'mount' to mount anything zfs on a chosen location. Using 'zfs mount' works fine

I'm not sure this is something 'by design' or I'm just doing something stupid. It's pretty deep water for someone who hadn't touched FreeBSD until 3-4 months ago. :°)

Alver: do you get an error or is it just not available?

Tenkawa: see the pastebin link - 'Operation not permitted'

It's available, I can mount it the 'standard' way all I want - but not the 'legacy' way using 'mount'

And unfortunately, podman's zfs storage backend only supports the legacy way

Yeah thats odd

Alver: oh. I just found searching it might be trying to protect part of the zfs container

thats why the zfs cmd allows the way around it

Aha - got a link?

Just a sec

just a wild theory though: wiki.freebsd.org/RootOnZFS/ZFSBootPartition

look at #10

I'm not sure I'm getting it

That area of the drive is locked unless the sysctl is set to allow

Hrm

The error message is the same, but I'm attempting to do things with a disk which is not even related to the "OS" part of the jail

That's why I said it was a stretch but it was the only thing coming up..

But something is thinking it's a bad idea my jail does that, that much is certain

Alver: I think this will happen for anything provosioned with zfs because it contains metadata

s/provosioned/provisioned

I need to hook up a few more zfs drives on my box here

(Once I fix my kerel bug lol)

Tenkawa: tested for shits & giggles with the kern.geom.debugflags but no change

bummer.. well it was worth a try

I can't wrap my head around it - I always kind of assumed that 'mount', when called for a zfs FS, would just call on zfs itself

Yeah its going to be something hidden, wrapped in an enigma like most things are in today's world.... lol

trying to find parallel package in freshports

uskerine: freshports.org/sysutils/parallel

˜/35

is it normal on STABLE to get "BOOT LOADER IS TOO OLD. PLEASE UPGRADE." at boot time? I've copied over loader.efi per LOADER.EFI(8)