go to 2600 Club! :)

darwin: Why are you spamming that?

you'll sometimes find UNIX/*BSD experts there

And this is reason to spam it across all the channels you're in?

i didn't do that

Of course you did. I saw it in at least three channels.

i'm on 150+ channels mostly where it's not relevant to; I didn't do that

one of them was GNU, run by FSF, and FSF themselves mentions 2600 Club

[21:12:30] <darwin> go to 2600 Club! :) someone using your nick did

how do I find my wireless device or maybe need to load a certain driver?

i found it's iwlwifi0 in dmesg but 'ifconfig iwlwifi0 up' says it doesn't exist

if club2600 doesn't help, try handbook: docs.freebsd.org/en/books/handbook/network/#network-wireless

Title: Chapter 7. Network | FreeBSD Documentation Portal

yeah no one else showed up tonight there. :\ Thanks; I looked in handbook:netowkr which shows example of Intel adapter, which I have... it says the particular driver iwn is already loaded or in kernel, but after showing in dmesg, my device somehow disappeared...

i have Comet Lake PCH-LP CNVi WiFi

it says to create wlan device for your wireless adapter

the easiest way is using rc.conf, or you could look up correct command in ifconfig(8) man page

okay; I based it on the command in that manual chapter this way: 'ifconfig wlan0 create wlandevice iwlwifi0' which said 'ifconfig: must specify a parent device (wlandev) when creating a wlan device'

maybe I'm supposed to use the driver name instead of device name from dmesg?

well, I tried that with same result now

i think 2600 has been rather on-topic in UNIX discussion areas since 1984 on nntp:comp.unix.freebsd.misc ... the system administrator (sysadmin) who taught me starting FreeBSD 1997 - '8 academic/college year writes articles for 2600 magazine and I've seen it discussed before

'... FreeBSD starting 1997 -...'

Repetition for emphasis?

i corrected my statement section where I had edited beforehand and put words in the wrong order

i see my wireless problem was solved at least six years ago on forums.freebsd.org but the manual is wrong. The correct command is 'ifconfig wlan0 create wlandevice iwm0'... but apparently I have to do this on every boot?

probably not, I am sure there would be an rc.conf entry that would help

wlans_iwlwifi0="wlan0"

try putting that in rc.conf

thanks

:)

might not be exactly what you are after

let me look in to it a bit more

depends on your wifi driver

but something *like* that

in your rc.conf should do it

what driver makes for iwm0?

looking in to it, seems that iwm is the correct driver

so something like

wlans_iwm0="wlan0"

`man iwm`

it was iwm0 on one and iwlwifi0 on the other, but 'wlandev' (not wlandevice)

so you need both?

yes. I setup these, rebooted, and they still show up, but when I try 'WiFi Networks Manager' in XFCE it asks for administrator password then says no wi-fi interface is configured in rc.conf

can you use wpa_supplicant.conf instead?

i don't know if I have that, but I'll be connecting to public networks I don't want to configure that every time

I do not know

I just use wpa_supplicant.conf

I would suggest using connman instead of nm

but I do not think FreeBSD has connman?

John Constantine?

Title: ConnMan - ArchWiki

I think it's a NetworkManager bug

what installs nm, darwin ?

rubygem-aws-sdk-networkmanager OR pc-networkmanager ?

i don't know

it is probably net-mgmt/networkmgr

NetworkMgr is an open source, Network Manager based on the look of the Linux

Network Manager user interface. It use ifconfig and netif if make all work.

FreeBSD Manual ch34 says use /etc/hostapd.conf ... when I 'service hostapd forcestart' I get errors on both laptops

see if you have that installed

not just using the xfce4 applet

if that does not work, try net-mgmt/wifimgr

ch34.5.1.3.1 ... well I'm trying this now and also installing networkmgr but don't like it

do not like it?

why?

often had/caused problems on GNU/Linux

eh?

it is not the same

does darwin use pkg or ports?

yes

which

networkmgr didn't appear in menu and on command-line just halts with new line

i always use pkg and depending which computer it is, sometimes ports

let me see

what I can get it to do

if you lose me ... I'll be back

i got it working on command-line and took notes for what to do with public wi-fi, though GUI would be better for that

idk darwin

I think there is an issue with my systray

darwin, same thing is happening here with me for networkmgr

does not even allow me to pass flags to it at prompt

I do not know about this networkmgr

try wifimgr

When I try to run a bhyve VM with ppt I get "MPtable is incompatible with multiple PCI hierarchies."

What is happening? Where can I read more about that?

dautor: probably something wrong with your vm configuration? you can read more in usr.sbin/bhyve/amd64/mptbl.c searching for that error message (and it also mentions -Y switch?)

Yes, it does mention it, but I want to know what I'm dealing with.

i'd guess your configuration has non-0 bus number for some device?

You're guessing right :)

I'll try with all of them on bus 0.

yuripv: It starts fine now. Thank you!

i wonder though if mptable is still required or it's a backward compatibility thing today

Is it possible to pass through a NVME device with a working installation of some other UEFI-capable OS installed on it? I'm using BHYVE_UEFI.fd as bootrom but the VM just hangs after some initial output before the OS gets control.

Here is a screenshot if it helps: dautor.xyz/files/vm_hanging.png

Solved it! It just wasn't outputing to serial device.

lovely people of #freebsd, hello

there is no chromium in packages?

looks like it's in quarterly but not in latest for some reason

oh

did not expect something like that can happen

seems like it is only in some repos freshports.org/www/chromium/#packages as there are no fallout reports portsfallout.com/fallout?port=www%2Fchromium%24 probably because some dependency failed to build

Title: FreshPorts -- www/chromium: Google web browser based on WebKit

I assume more in latest than quartely ;/

ye, just checked it in freshport and saw that

latest is less tested, more likely to have build failure

generally if someone builds once in a quarterly branch it will build forever as the branch only gets minor package updates

so it can like appear and then disappear?

in latest

i guess so, i don't use the freebsd.org packages much. but since the package isn't there the answer seems to be yes :-)

okay, makes sense

oh, "build/timeout", but no idea why there is no fallout report

latest is like the -current of ports so, some breakage is to be expected... i would stick with quarterly if using packages although a fair number of people use latest

ye good point

so latest is build like every few days or so?

I wast yesterday in infinite captcha loop in firefox, so wanted to try it but

I saw all hydrants and traffic lights on the earth probably

and a lot of bicykles and buses

afaik for latest and quarterly incremental rebuilds are done in a loop and a full rebuild every two weeks, but latest gets more changes, so more needs to be rebuild on incremental rebuilds

nimaje: do you know if the build logs are available anywhere? there's a 'logs' directory in /FreeBSD:14:amd64/latest but it's a 404 so i assume it's a symlink to somewhere

start from pkg-status.freebsd.org/builds?type=package the commit hash brings you to a mirrored site that supports both ipv6 and ipv4 and the bomb icon brings you directly to the builder, in case the mirror doesn't work for some reason, but some builders only have ipv6 (no idea if there are some with only ipv4)

yhsnkd

uh, thanks

are the nginx port defaults the same as the ones that come with the pkg version ?

last1: all packages are compiled from ports with their default options.

gotcha, thank you

To expand a bit more on that, packages are built from the ports framework using poudriere, which is a small utility that exists to create and destroy jails used for building things cleanly (ie. no environment pollution, et cetera).

This, ie. using poudriere, and the use of the ports framework, is also what makes it easy for other people to build custom package repositories for themselves (or others), or even create overlay repositories that can provide individual custom packages with different options than the default.

found a good fix for my alc(4) problem: ix0: <Intel(R) X520 82599ES (SFI/SFP+)> port 0xd020-0xd03f mem 0xfc280000-0xfc2fffff,0xfc304000-0xfc307fff at device 0.0 on pci8

I periodically run into a situation on my FreeBSD NFS server in which nfsd somehow (apparently) loses its registration with RPC and needs to be restarted. it also seems to happen after a reboot. I feel like I'm missing something?

I'll try to mount a share from a Linux client and get "mount.nfs: Protocol not supported" ... and if I run showmount -e on the server, I get "RPC: Program not registered" until I restart nfsd

tsundoku i can confirm, happens to me too, 13.2

I'm on 14.1, but this system started on 13.2 and I think it happened there as well

er 14.0, sorry.

rmacklem@ is actively working on NFS, so my best suggestion is to try and replicate on 15-CURRENT, and if not then file a report on bugs.freebsd.org

Title: FreeBSD Bugzilla Main Page

hm

If you can't upgrade your production system to 15-CURRENT, you might do it in bhyve?

you could share with zfs over nfs. I am guessing you are using /etc/exports now. but I wouldn't know how to troubleshoot

oh I can definitely set up a VM, I'm just thinking about the situation

and I'm sharing ZFS as it happens

ah hm

"Sharing with ZFS" just adds lines to /etc/zfs/exports

right, I noticed

It's all the same to nfsd, which reads both.

but doesn't require restarting nfsd

it would be nice if I could get it to stop complaining about /etc/exports not existing, on a related note

and least upon initial share

scoobybejesus: I'm not sure that applies to this issue.

tsundoku mounting with v3 protocol?

v4

oh wait it *is* v3

I want it to be v4

tsundoku: if you're using V4, you need "V4: /" in /etc/exports which'll solve your issue.

that's weird. does FreeBSD's NFS server default to v3 only?

That's the only line you need.

so how do I do that through the zfs command

If you don't specify "V4: /" in /etc/exports

You can't, you need to add the line to the file.

I hope I don't have to manually clean up /etc/zfs/exports after issuing a zfs sharenfs command

You don't.

that's... a little rough

Howdy.

After my failed atempt to upgrade 13.2 to 14, it was suggested that I should try a boot environment, a snapshot of which was probably created by the new upgrade process. However, it seems that I only have one available. Now I'm running a 14 kernel with a 13.2 package set, freebsd-version says 13.2-RELEASE-p9. bectl list gives libbe_init("") failed. freebsd-update fetch tries to pull files from

You never touch /etc/zfs/exports

14.0. What's the best way to proceed? I think I want to have the binary install start over, basically.

wait, this seems like a catch-22. I never touch /etc/zfs/exports, but the only way to serve v4 is to manually edit the file after running the command?

tsundoku: the only way to get nfsd to do V4, as far as I know, is to add the line to /etc/exports and leave /etc/zfs/exports alone.

weird

zale23: /rescue contains a very basic environment that you should be able to bootstrap from

I feel like if that's the case, all of the FreeBSD docs saying "just run this zfs command to share a dataset!" should have the caveat "only if you intend to do NFSv3" attached to them

debdrup: The system is booted currently with the 14 kernel and 13.2 package set. /rescue/bectl list also produces the same error.

I see ldd reports no dynamic linking, so I assume somehting else is going on.

tsundoku: the documentation is part of the FreeBSD project, so you're welcome to submit a patch via GH, BZ, or phabricator.

right

zale23: /rescue is statically linked (and is technically just one single binary, via crunchgen1)), so it can be used in situations where even dynamically linked libraries are broken. Not sure what bectl is attempting and failing with init.

I need to do some research and decide how to set things up, but I need coffee first

I'm a little disappointed that it doesn't just default to v4 like every other OS I've used in recent years, though

tsundoku: I run NFSv4-only for my servers, so I don't have to deal with RPC at all - but that's because I don't have any Linux boxen, as they don't have support for NFSv4 ACLs.

debdrup: that's good to know about, thanks for the info. Do folks usually boot into that, or just execute those paths? Not sure how to make use of it to solve my situation.

It doesn't default to NFSv4 because Linux doesn't implement NFSv4 ACLs among other reasons, and FreeBSD attempts to have interoperability defaults.

Is there a way to force freebsd-update to start over on an upgrade? Or force it to pull from a specific list?

I've been over Solaris, FreeBSD, and various Linux distros lately, and they all do v4 for everything out of the box unless you specify v3, so that's interesting to hear

"Linux doesn't implement NFSv4 ACLs" is news to me

zale23: you can boot into it by setting init_path via the boot loader prompt, which I've sometimes had to do.

I don't think the setting up ACLs is actually necessary to use it in the first place?

It is if you're doing a shared environment.

and Linux just straight up doesn't support them, you say?

isn't it also possible to use traditional access control with NFSv4? I'm not following what you're saying.

tsundoku: nope.

As in, nope it doesn't support them.

Not sure what "traditional access controls" are, in this context.

posix?

yeah

There's POSIX ACLs, but that's not part of the NFSv3 specification, so not every client implementation will support i t.

just setting the user/group and octal permissions

I've never encountered a client that doesn't

The ACLs are pretty boring unless you back it with an authentication system and some crypto.

Windows, for example, doesn't really support it, and I don't know about macOS.

macOS does

NFS+Kerberos isn't exactly unknown.

Windows, I have no idea

debdrup: also not trivial.

trying to serve shares to Windows clients with NFS sounds kind of insane

Anyway, NFSv4 ACLs are also cool because they let you do both NFS and SMB sharing on the same share.

tsundoku: eh, it's fine. Windows has a NFS client that works quite well.

yeah, I mean, I know NFSv4 ACLs have their purpose, but in my simple environment I've just been serving NFSv4 without setting up ACLs and relying on POSIX permissions (and specifying hosts allowed to access the share) for everything

so hearing that NFSv4 doesn't work without ACLs and therefore the default needs to be v3 runs counter to what I've seen other OSs do and what I've done so far

That's not really what I said..

*and* hearing that Linux doesn't support them is... really odd

Posix isn't an over the wire acl. I think you two are talking about different things.

I said that FreeBSD defaults to being interoperable, because among other things, not everything supports NFSv4 ACLs.

NFSv4, set up properly, doesn't need rpcuserd or any kind of RPC configuration and can be done over a single TCP port.

Is there a way to know what version /boot/kernel.old is?

right, that's part of the reason why I prefer to use it

And with the work by rmacklem@, it can even be done over TLS.

zale23: strings(1)

Nice.

or what(1)

but I've so far not understood ACLs to be a prerequisite for running NFSv4. they're just an optional extra thing you can do if you need more sophisicated access control than traditional POSIX ownership and permissions can express

(as in 'what /boot/kernel.old/kernel', which just prints the version)

Yea, what(1) can do it too, but strings(1) works on arbitrary binaries, so I tend to remember that better.

also I'm still hung up on "Linux doesn't support NFSv4 ACLs." that seems completely improbable to me. I'm trying to find out more

s/sophisticated

tsundoku: good luck. There's been two different attempts at implementing it, approximately a decade ago - neither succeeded for the usual reasons.

ACLs aren't a pre-requisite if you're just a single user with a NFS that you're accessing..

Just because ACLs don't fit in your use-case doesn't mean FreeBSD has to switch the default..

my understanding, which may be incomplete, is that NFSv4 without ACLs set up can do all of the same access control that NFSv3 can.

Sure, unless you're doing NFSv4-only.

so I don't understand how defaulting to v3 improves interoperability

tsundoku: you mean posix permissions, correct?

yes

Then that isn't an NFSv4 ACL.

I know

If, for example, you adjust vfs.nfsd.server_min_nfsvers to 4, you probably won't be able to use NFSv3 either, but I'm pretty sure you _still_ need to add the line to /etc/exports

Posix permissions in v3 rely on local passwd database to confirm Id of the user.

I'm trying to understand the scenario where a client doesn't support NFSv4 ACLs and therefore it's necessary to use v3 instead, because doesn't v3 *only* have POSIX permissions?

zale23: right, and you can use v4 that way if your needs are simple enough, too

I'm not sure why you'd try to understand that use-case.

I already outlined one: Windows NFS client.

so I'm not grasping which kind of situation would make it appropriate to use v3 but not appropriate to use v4 without ACLs. as I understand it they are functionally the same when it comes to permissions

because v4 without ACLs will just rely on local passwd database like v3 does

You can't do NFSv4-only if you have a Windows NFS client.

tsundoku: not if it's Windows NFS client.

so it's basically for Windows clients

What do we suppose the chance of booting a kernel from an older zfs snapshot by rsyncing from the snapshot to /boot?

There are probably others.

the defaults I'm used to, for the record, are NFS servers that default to v4 but will serve v3 if the client only supports v3, or asks for v3

Also, the others that "support NFSv4", do they actually default to NFSv4 for the shares?

yes

So you lose out on Windows clients and others.

I think most vendors expect you to use SMB for Windows clients

SMB is slower, though.

Samba is, at least.

..it's also a pain.

i think a more pragmatic reason for not requiring nfsv4 by default is that 98% of unix 'admins' don't stay up to date on this stuff and expect nfs to work the way nfsv3 did, even though nfsv4 has been around for 20 years

I mean, fair

but SMB is the Windows standard, so people serving Windows clients usually use it

zale23: I'd say backup your stuff and just re-pave...

Gah, so much!

that's an interesting difference in perspective on the part of FreeBSD, if all of this is true

Yea, it sucks. :(

just trying to understand the situation

zale23: I wish I had a better recommendation

tsundoku: defaulting to interoperability doesn't seem that outlandish.

If I compiled from source, would the install be successful?

I'll be honest, my needs are simple and I have just been using NFSv4 the same way as v3 all of this time (on Solaris and Linux)

debdrup: I don't disagree, but other platforms I've used default to serving v4 or v3, depending on what the client asks for. not v4 only

so I'm not following why in FreeBSD it needs to be either-or.

or maybe it doesn't need to be, and they just decided to do it this way for other reasons

It's just as likely that nobody has seen the need to make it default to V4, because there isn't any real reason for it.

Unless you do NFSv4-only, you still need to configure RPC for NFSv3 clients.

Its a little faster iirc.

that's what I thought too

zale23: it's quite a bit faster if you do NFSv4-only, yes.

But I have a server doing both v3 and v4. Pretty sure.

In my testing, the biggest speed-gain was had from doing NFSv4-only.

debdrup: was that because of delegations, or something else?

I'm used to things doing v3 and v4 out of the box with no configuration necessary beyond exporting the share, unless you want to restrict access beyond what the POSIX permissions do

unixwitch: that's an excellent question. Next question. :D

tsundoku: Your client isn't upgrading the connection.

with v4 being the default for clients who don't ask for v3

upgrading?

And does 'mount' client side report v4 mounts?

tsundoku: one issue here is that freebsd requires the V4 root to be specified in /etc/exports before v4 will work, so it can't be enabled by default. you'd need to default the root to / or something, which may not be a good idea (not sure)

unixwitch: the root share, not necessarly /

If everything you've shared is in /srv you can specify that rather than /

interesting. and this is specific to FreeBSD? or other platforms have a default?

debdrup: isn't that what i said?

zale23: yes, it did

you have to specify "the V4 root", which doesn't need to be /

unixwitch: fair, I missed that.

Then you don't need v3 tsundoku, correct?

I don't.

Do we have docs on v4 only?

tsundoku: read the NFS protocol specs for details I suppose.

nfsv4only is an optin in rc.conf(5), I think?

I'm talking about on my old server environment that this FreeBSD server is replacing, and others I've used. all of my mounts were v4, but if I had a client requesting v3, it would get v3

but I might as well set it to v4 only for this

nfs4_server_only*

occasionally I connect an old system that doesn't have v4, but that's rare.

tsundoku: Right, support what your clients need. The fstab line will tell you which protocol.

s/optin/option/

thanks debdrup

weirdly nfsv4_server_only isn't mentioned in nfsv4(4). should probably submit a patch for that

I don't really care whether it's v4 only, I just want clients that don't specifically request v3 to get v4

unixwitch: I'll approve. :)

and I'd prefer to be able to use the zfs commands to export shares, without having to remember to ignore that whole feature and write /etc/exports by hand

I don't know how other nfsd implementations manage to oppotunistically default to v4, and I don't know if nfsd in FreeBSD can do it - but what I do know is that unless a copyfree-licensed doesn't do it, rmacklem@ will have to figure out how without reading any code..

tsundoku: then you want something that isn't nfs. Like I said, nfs doesn't upgrade the connection from 3 -> 4. the client makes a request for a specific version and gets that version.

zale23: yep.

I never said anything about upgrading a connection

unixwitch: yes, yes you should.

I was not expecting that.

When you say "to get 4", that's what I heard.

same.

then you misinterpreted

nothing about "get" inherently implies "starting with one thing and changing to another"...

so fine, clients are requesting v4 and getting v4, in the setup I'm used to and want to end up with.

Mostly what I'm saying is the client is making a request for a specific version. The server either supports it or it doesn.t

what I'm used to for defaults is that almost all clients will request v4, but if a client requests v3, it will get v3

Not much good can come from trying to attribute blame for miscommunicating, since there's at least 4 chances of error to occur in information transfer.

yep, no finger pointing from my perspective. Just want to be accurate about what we're speaking too.

okay, so I'm just trying to understand where FreeBSD is relative Linux and Solaris, basically

because in all the Linux and Solaris I've used, for a long time, the default is to serve v4 and v3, depending on what the client asks for

and if you don't set up ACLs, v4 will just use POSIX permissions

like v3.

but it sounds like nfsd in FreeBSD is different and maybe can't do both at once, so the default has to be v3 for edge cases with non-*nix clients?

See my comment at ~17:15:09 UTC..

Anyway, dinnertime!

debdrup: well i did! freebsd/freebsd-src #1008

Title: nfsv4(4): mention the nfsv4_server_only setting by lexiwinter · Pull Request #1008 · freebsd/freebsd-src · GitHub

okay, i am looking at zfs send -I and -i.. what the heck is the difference? utlimately i want to take my baseline daily snapshot from t-2 and APPLy t-1 daily snapshot in an incremental fashion to it

voy4g3r2: -I sends all the snapshots that exist between the two you specified, -i only sends the two snapshots you asked for

for backups, you probably want -I, so all your snapshots are copied

so -I has more "logic" to figure out what is needed, where -i is more explicit?

i definitely do not want to send over a 180gig music drive everyday.. as i just added more concerts to the library.. in order of 3.4gig

makes sense, now to figure this out, thank you unixwitch

might as well learn this, since it is snowing.. and can not do anything.. first snow fall of the season

and i really need to get in habit of reading stuff like this: docs.freebsd.org/en/books/handbook/zfs/#zfs-send-incremental

Title: Chapter 22. The Z File System (ZFS) | FreeBSD Documentation Portal

voy4g3r2: imagine you snapshot music@monday, music@tuesday and music@wednesday, then do 'zfs send -i music@monday music@wednesday', you will end up with music@monday and music@wednesday on the destination. if you specify -I instead, you will end up with music@monday, music@tuesday and music@wednesday. the only difference in the amount of data copied would be if music@tuesday had a significant amount of referenced data which had been deleted in mus

ahhh!!

minus my time machine backups, second is definitely number in priority

i get more and more why there are

"tools" to do this.. this is not easy to get "right"

it is not... fwiw, i quite like syncoid, which is part of sysutils/sanoid. it will automatically create snapshots and sync them to a remote host over ssh

the perl program.. i saw the presentation on that one.. looked real cool

unixwitch: cgit.freebsd.org/src/commit/?id=fbbdfa2b8a42

Title: src - FreeBSD source tree

i am in the phase, okay i understand how it "works" let me use a tool, that can do it better than i can.. plus i can offload "support/maintenance" to an expert

debdrup: ty!

i hate my inbox and want it to suffer, so i have subscribed to freebsd-desktop@

Only one list? ;)

I'm _very_ glad neomutt has proper list support, including reply-to-list functionality.

debdrup: well i'm already on current, stable, ports, questions, fs and hackers...

but desktop seems to have more mail than all of those combined :-)

probably because it gets all the bugzilla desktop@ mail

desktop@ is one of the lists I'm not subscribed to, but follow via mlmmj-webview.

you know, it would be nice if there was (read-only) IMAP access to the list archives. i have absolutely no idea how difficult that would be to support though

unixwitch: mail from bugzilla-noreply@ doesn't count in my world, as that gets filtered into its own mailbox by my MUA.

ah, i just dump everything into one freebsd folder

if i had a separate bugzilla mailbox i'd just forget to check it until it had 10k mails in or something

I have a freebsd directory with subdirectories for doc, current, hackers, status, and a few others.

unixwitch: You can send a mail to postmaster@ with a recommendation, but I can make _absolutely_ no promises.

i don't think i want it enough that i'd consider actually asking for it, i'm sure it would be non-trivial

i'm sure postmaster has better things to do

i was worried about -current, with that one checksum issue i saw.. it is relatively quiet in comparison to my past experiences with listserv

most of the lists are fairly quiet, the general ones like fs@ and net@ are quite busy because they also get bugzilla traffic

although really not *that* much traffic

wpow it is still around: lists.psu.edu

Title: Welcome to LISTSERV 17.0

oh, i used to use listserv on freebsd. aside from not coming with source code it's quite a nice bit of software

i preferred it to mailman, never tried mlmmj though

debdrup: do you want me to close the nfsv4(4) PR or will you leave it open until MFC?

unixwitch: go ahead, the MFC notice is a tooling thing for the repo of record, because there's a cronjob that scans and keeps track of all commits, and sends the relevant reminder when it's time.

Github is still _just_ a mirror (though it now accepts PRs, which we grab via gh(1) using a git-remote(1)).

oh man, i use to work on this server a lot when i was at uni: reddit.com/r/PennStateUniversity/co…shutting_off_its_open_source_mirror before they instituted bandwidth limits, damn you napster!

The warstories from its operators at the time are _something_.

debdrup: nothing like dating ourselves.. i remember prodigy on my ibm pcjr

i STILL have that computer.. not even sure if it works

Hopefully there's no battery in it, or it probably won't.

i get the evil eye from the wife, when she sees it and we are rearraning our storage room

i remember ftp.cdrom.com, but i always got my BSDs from sunsite.doc.ic.ac.uk because it was closer

plus you could mount it over NFS :-)

i honestly do not know if there is a battery in it

but i do have fond memories of using cassette tapes to backup gw-basic programs and "forgetting" to NOT have it by the monitor

sad: ftp: Can't lookup `sunsite.doc.ic.ac.uk:ftp': Name does not resolve

Yea, sunsite is gone :(

i guess oraclesite didn't have the same ring to it

tsx-11.mit.edu is dead too, i wonder how many interesting old files have vanished in the last 20 years

at least TUHS is still around

is there a cache somewhere where pkg keeps its info ?

I've updated to 12.4 and pkg install pkg complains that Bootstrapping pkg from pkg+http://pkg.FreeBSD.org/FreeBSD:12:amd64/quarterly

but that URL no longer has info about 12.x

last1: 12.x just went EOL

darn it, is there an archive site I can use temporarily ?

i managed to crash freebsd 14-RELEASE by using the serial port... any ideas how s_ttyvp ends up as 0x38? 0x0.st/H6Dg.txt

last1: you could always use poudriere to build some packages on a newer system with COMPAT_FREEBSD12... i think that's supported

Does ngctl crash for anyone else on SIGWINCH?

last1: although you'll probably have to use ports 2023Q4 since main has had all the 12.x-compat stuff stripped out

I found the opnsense repository

used that one to bridge the gap

why are you trying to support 12

ascreen: forwarded this to another chat. I think you should open up a bug

How do I re-bootstrap pkg without a NIC, after upgrading to 14.0?

ShinyCyril: download pkg-<version>.pkg and install it with /usr/sbin/pkg add

I managed to bork my FreeBSD box upgrading from 13.2-RELEASE to 14.0-RELEASE. My NIC uses the if_re.ko driver (not part of base install). After rebooting into 14.0, this driver is no longer loaded, presumably since it was built for the 13.2 kernel. I have a newer, which means I have no network.

(i can grab the url for you from packagesite if that helps)

when running pkg add however, I'm getting a ld-elf.so libssl.so error

that's probably because you're using /usr/local/sbin/pkg and you didn't do 'pkg upgrade' before the 3rd run of freebsd-update

which means freebsd-update removed the old 13.x libraries that packages depend on

I did freebsd-update install - reboot into 14.0 - freebsd-update install

I must have misread the manual — I was thinking I had to do a pkg upgrade *after* the second freebsd-update install

ShinyCyril: it should be after the second and before the third, unless i'm getting confused

yeah this is definitely my bad

anyway

no you're right - I just re-read the docs

download this: pkg.freebsd.org/FreeBSD:14:amd64/quarterly/All/pkg-1.20.9.pkg and install it with pkg-static

then download this and install it with pkg: pkg.freebsd.org/FreeBSD:14:amd64/qu…ly/All/realtek-re-kmod-198.00_3.pkg

then kldload if_re (or whatever it's called) and you should be sorted

(is it kldload if_re_updated? re and ix use different module names here and i can't remember which is which)

ok i checked, it's "kldload /boot/modules/if_re.ko", the _updated is for ix

interesting: pkg-static add pkg-1.20.9.pkg is complaining about pkg-1.20.9 already being installed

oh, use -f

it is installed but it's for the wrong abi, i think pkg(-static) add doesn't notice that

ah that did the trick. And same message for the kmod - presumably I also need to add -f that, for the aforementioned reason?

yep

then rather than mess with kldload i would just reboot and if you network works, pkg update; pkg upgrade should update everything

(i use pkg upgrade -f here but someone recently pointed out that pkg upgrade shouldn't need -f and i think they're right)

awesome, network is back up. Now to just pkg update/upgrade

ShinyCyril: if you remember, did freebsd-update warn you when you ran the 3rd install that you should update ports/packages first? because i think you're the third person this week who had this problem and i feel like there needs to be a warning, or if there is one, it should be bigger...

thank you very much for your help unixwitch! Any chance I could buy you a beer?

I don't recall seeing anything, but given I didn't read the docs clearly, it's also possible I didn't read the freebsd-update prompts clearly either ;)

very glad that there's a statically-linked version of pkg - definitely saved my bacon :D

unixwitch: you get the instructions after the 2nd install

but some people like to live dangerously

3rd is equivalent to `make delete-old` and like, don't

i still didn't find out what's wrong with my tunnel ^^"

i suspect there is a kernel setting i need to change but idk which one

and what hte value should be

(freebsd/freebsd-src #1007)

oh this is some github-specific thing