that's what storage guys always say

heh

I shouldn't say anything but... I heard "Windows clients" and past history made it impossible for me to consider that the problem might be elsewhere.

this is fair

I can't imagine it being anything on the FreeBSD side

there just isn't much to go wrong

but I also know that Windows can be extremely picky about things, so maybe it's not getting something from the FreeBSD iSCSI target that it wants. *shrug*

it "works" but as I'm being told, Windows client's NTFS filesystem riding on the zvol occasionally loses file indexes or something

how are they seeing occational file corruption when NTFS has no facility for detecting that?

I'd have to check again with them, but what I heard was that they get "indexing errors" accessing files

when running freebsd-update is there a way to force freebsd-update to always override /etc (I don't want to merge) ?

i think you can ue IGNORE_FILES

if I remove /etc from UpdateIfUnmodified and MergeChanges is it enough?

(in freebsd-update.conf I mean)

does freebsd-update uses mergemaster or etcupdate?

ah, freebsd-update. it uses its own config update, i remember

It used to have only mergemaster. Now it has both, but I understand mergemaster is or will be deprecated.

It doesn't use either mergemaster or etcupdate, it has its own facility.

ok..

so could I simply comme,t

oops

so could I simply comment MergeChanges in freebsd-update.conf and remove /etc from UpdateIfUnmodified ?

Can I ask what you're trying to accomplish?

I'd like to upgrade 50+ jails automatically without merging things (I'm using Saltstack)

so overwriting things is not a problem as there is a minion in each jail and the config will be reapplied after

Seems like it's easy to set up a test case then. :)

:)

Oh, I misread. freebsd-update, not freebsd. *wanders off until caffeine hits brain*

I wanna use lua shebang, but I noticed there is no lua symlink

FreeBSDs lua is private.

what does it mean?

It means it's not accessible outside of the source tree.

interesting

There's plenty of private things in the FreeBSD source tree ;)

but it's in /usr/libexec/flua

Also, it's probably good, because flua is, iirc, kept at an old version because the lua folks decided to make a POLA breaking change at some point.

And running it, it says "Lua 5.4.2" which is not that old

The Makefile still mentions that it's internal.

Demosthenex: You should be in #freebsd-social as well.

i used to be

mason: did you find a driver in ports?

debdrup: There isn't one, unfortunately.

Aw.

debdrup: That said, there are issues with the card under Linux too, so it's maybe not a fantastic target.

Best I could accomplish on it was Fedora or (where I left it) Debian with a backports kernel and still-bleeding upstream firmware.

mason: net/wifibox?

VMs on laptops aren't my favourite thing.

Alternatively, can't it be replaced?

That I'm not sure. Possibly? Hm.

What model is it?

my7961 anyway, and there are a few recent laptops out there with it now

That's the card. The laptop itself is an Asus Vivobook

It's effectively the same issue as forums.freebsd.org/threads/issue-wi…rom-memstick-on-acer-aspire-5.86036

Title: Issue with installing FreeBSD 13.1 from memstick on Acer Aspire 5 | The FreeBSD Forums

Ah, Asus don't publish service manuals.

mt7961, typo'd

Dell, HP, and Lenovo all publish service manuals for their commercial lines of laptops.

This was an impulse buy more than a researched decision. Needed something like this for my wife's birthday, and noted this on sale.

Oh, didn't finish my sentence. Best I could do was Fedora or Debian with bleeding-edge stuff plugged in, and I was still measuring packet loss with that. (Bleeding edge firmware made it very little packet loss, but still some.)

Wifi was a mistake. :P

100% agreed

I need to make sure my next apartment actually has ethernet ports

could always make use of a wifi-client device

I've mostly got ethernet where I want it here.

There are a couple rooms that still need to be wired.

It's very weird living in a country that still uses fax for things but having no wired connections for internet

Make up your minds

Wiring ethernet in an already built house is a bloody pain, but it is doable.

Thankfully, I'm done.

Hi. I'm using ZFS but I don't understand how to read this part of zfs list: paste.scsys.co.uk/1986

Are default and 12.3-RELEASE-p5_2022-07-17_155100 both mounted as / ?

Title: magnet_web paste from "ferz" at 217.168.150.38...

uh, it would appear so

the 12.3-RELEASE-p5_2022-07-17_155100 is only 8k tho

?

ferz, Please run this command: zfs list -o name,canmount,mounted,mountpoint

And this one: bectl list

that's the mountpoint

do zfs list -o name,mountpoint,mounted

Those will show that a boot environment file system clone has been created, with a mountpoint of / but that it is not mounted at this moment.

probably one of them will say "no" for the last property

sorry, that was for ferz

futune, We both arrived at the same time and responded to ferz. But we were saying the same thing so it must be good advice! :-)

oh yeah hehe

ferz, Also, zfs list is listing zfs datasets, but don't forget "df -h" which lists mounted file systems. Unmounted datasets won't list with df.

are there freebsd ports that use npm but don't try to save the npm cache as a distfile?

is the portbuild cluster allowed egress?

look at textproc/kibana8, for example

i've spent days to figure out how to do npm build properly

the whole npm/gomod/pip/composer pattern has gotten out of hand

otis: this is sadly way more simple than the one i was looking at

network access is only allowed in the fetch phase

it is? damnit

i think their npm is broken and can't be fixed in the port

you could look into using npm as fetch dependency, no idea how to get it to allways fetch the same stuff, so it can be checksummed and then use npm in an offline mode at build time (it should have something like that I hope)

i could try that but there's ~200 packages and that's basically duplicating the process of npm, with urls and checksums

Any idea why posix_fallocate(2) would return something other than 0 or -1? It is returning 22 for me, but the man page leads me to believe this isn't compliant

I know it doesn't make sense to call posix_fallocate with ZFS, but regardless, still bizarre

22 is EINVAL, ... or the operation is not supported by the file system.

I know it is EINVAL, I'm just confused why it returns EINVAL rather than returning -1 and setting errno accordingly

it uses errno for other failure conditions

strangely, glibc's posix_fallocate doesn't set errno at all, it just returns the error code directly :/

The manual page (13.1) says it conforms to POSIX.1. Why POSIX.1 wants it to do that is more than I can guess.

ah, I'm dumb, I was looking at an older man page

seems like the sane design "It returns an error on failure, without setting errno."

Posix standard text doesn't even say. Rationale: none.

why would you set errno if you can do better?

POLA.

sometimes, POLA just means: We've always done it this way

pretty sure most cases where the result is just success/failure, 0 is returned for success and the error number or negated error number is returned on failure (maybe with some exceptions) and errno is used mostly when a function additionally has a normal result

I have a iwm wifi mini-PCIe card which is detected in dmesg - but I can't turn it into a wlan interface no matter what I do in rc.conf. It is an Intel 7260. Any ideas? Anybody running this card?

The same FreeBSD installation works find for other iwm cards (nvme).

s/find/fine/

sysctl dev.iwm.0.debug=1 doesn't seem to do anything in dmesg.