i need to create a virtual network interface

i want to create a internal virtual lan

any hints how to do that on fbsd?

nvm i found the solution

rwp: correct, it is every time. let me check the debug execution if it shows something.

rwp: error happens exactly after the bash completion user dir.

rwp: + BASH_COMPLETION_USER_FILE=/dev/null

bash: SRCDIR: unbound variable

bash_completion_user_file is set to /dev/null (but that is the default, I never touched that).

cybercrypto, It should be the line after the BASH_COMPLETION_USER_FILE=/dev/null line because that one reported and then after that you got the unbound variable.

But again, don't set -u, no one else's shell code is expecting to run with -u or -e active. If you feel you must set it then set it last after everyone else's code.

Find where you are doing a set -u and remove it.

With today's FreeBSD update it's unimportantly nice that "freebsd -kru" reports all three the same again. It's a silly thing I know.

i need a complete tutorial to ipsec ^^"

i got openvpn sorted but i want ipsec

the ipsec tutorial in freebsd is completly worng

i managed to set few first steps, then it's completly useless

i can manage to set the tunnel up and that it

rwp: Yes, when I mannualy set the variable, the error message disapear. I will investigate if somewhere it is me using -u (not that I changed mannually)

rwp: thanks for the help, keep you posted if any progress.

mane: have you considered wireguard?

i can do openvpn

someone else told me about wireguard already

i will try wireguard too

I have moved to wg completely (Free/OpenBSD, Linux).

freebsd documentation for openldap is horrid

and outdated

i need someone to tell me how to set up ldap

or at least hand me a proper howto

ok there is something usefull in the handbook, finally

Hi guys, what is best practice to symlink a script, say ruby one, but it depends on a module in the original folder, when symlinked to /usr/local/bin, running it caused error, cannot find the denpendent module, only running from the original folder works... Anyone has any tips?

Is there a list for zfs or storage ? I don't see any. The only thing I'm finding anything about zfs is the handbook and the forums

Aw found a list for filesystems

yep, freebsd-fs@

Thanks

mane, Last time I hated upon ipsec in the channel I apparently offended some ipsec users. But honestly I think ipsec is a terrible protocol. Never use it. Use openvpn or wireguard instead.

Or just ssh.

I also find sshuttle very useful.

how would i configure openldap client?

Hi

how much time could we expect to wait to see 14.1 coming ?

is that safe to instal 14.0, or should i wait for 14.1 ?

mane: for ipsec use strongswan

eoli3n: a friend of mine is running a few 14.0 instances without problems, you may want to look at freebsd.org/releases/14.0R/errata/#errata though

Title: FreeBSD 14.0-RELEASE Errata | The FreeBSD Project

jmnbtslsQE now i rather need to know how to properly set up openldap-server-26.6.6 and the clienets

eoli3n: looking at past releases, it's anywhere from 6-12 months until the first point release, weighted more to the longer side

eoli3n: we're already at p3, I hope we've got most wrinkles ironed out by now. but if your use case / hardware / configuration is very special, you might be the lucky one to discover even more bugs. that's just how software works

14.0 should be fine though for a fresh install. Generally I wait until a point release for upgrades from previous versions, as there's generally not rush

or rather, ALPHA1

ok thanks, then i'll try the upgrade soon

a scary note on the upgrade page: After upgrading, it is not possible to use freebsd-update rollback to return to 12.4-RELEASE or 13.2-RELEASE.

of course there is always a possibility for zfs rollback :) so do not forget to place a snapshot prior upgrading

To quote Macbeth "I am in upgrades / Stepped in so far that should I wade no more, / Returning were as tedious as go o'er"

btw I should host ports also for 13.2 bsd versions, my question, will poudriere work as expected on 14.x ?

if you use a 13.2 jail

nerozero: ^

poudriere do build ports in jails so ... ?

nerozero: Create a poudriere jail for each major version+arch combo for which you want to build packages

no, my question was - will this ^ work as expected? Because if I try to build a higher version BSD ports on a lower version BSD machine in jails, poudriere worns about it ...

s/worns/warns/g

the running kernel only creates an upper limit for the version in a jail, not a lower limit

nerozero: Is your builder box running 14?

vkarlsen, I still hesitate to upgrade to 14.0, but I would like to test it on a spare machine, so ... as an option I need my ports ( extra configured ) ...

so, how do i log in with nis credentials on a nis network client?

After upgrading to 14.0-RELEASE I noticed my wifi is terrible. regdomain isn't used as specified in /etc/rc.conf for example. Defaults to US instead of NL. Startup and trying to get it to even work fully takes several netif restarts. I tried searching for it but can't seem to find anything. Anyone know what the issue might be?

After a netif stop/start it might get a IP adress and I can ping stuff on my LAN, but not the internet. Another restart makes that work somehow.

Quick poll: Which do you prefer, readlink or realpath?

_xor: realpath

well, they do diffrent things, but when are you really interested where a symlink points and not the absolut path after resolving symlinks and stuff?

There are instances where that information might be relevant (e.g. symlink should point to some object by convention, and if it doesn't, then create it).

yeah, but then you exactly know which one you need

readlink -f

_xor, actually stat, it support output formatting

well updated my home page, working on a project since yesterday :DD

Title: Mirosław Leśniak

when i'm trying to compile a port (bitlbee w/ libpurple), is there a way to have it use the existing pkg pacakges for things like autoconf, xmlto, etc instead of trying to compile them all in ports?

Let's say, that I open a socket with IPV6_MULTICAST_LOOP set, send a packet with sendmsg() to multicast address and receive a reply from the kernel addressed to unicast address. Is it possible to receive those replies? If yes, then does it require some kind of specific receive socket configuration?

Demosthenex, "make install-missing-packages" in the port's dir. See ports(7) for more.

V_PauAmma_V: neato!

Hi folks, I have asked this question several days ago and had a few suggestions but none worked. I started the following thread forums.freebsd.org/threads/upgradin…from-source-is-not-successful.91263 to try and get help from forum and also got some suggestions but they also didn't really help. Basically I tried to upgrade from 13.2-RELEASE to 14.0-RELEASE both via source and via binary updates. Compilation and installation (both world and

kernel) work fine. However after reboot, when issuing freebsd-version -k gives 14.0-RELEASE-p2, freebsd-version -r gives 13.2-RELEASE-p3. I've tried to use various boot environments, but the situation persists. I alo found the following thread serverfault.com/a/602263 but not sure whether this is applicable. Any help is appreciated.

Title: Upgrading 13.2 -> 14.0 from source is not successful | The FreeBSD Forums

Title: Conflicting information about the running kernel version in FreeBSD - Server Fault

speaking of install-missing-packages, why does it include build-dependencies of dependencies? shouldn't it let pkg resolve dependencies?

any ideas what happened to this zfs pool? host memory corruption maybe? bsd.to/CA0p

Title: dpaste/CA0p (Plain Code)

unixwitch, First, I don't know, but are there any kernel errors logged to /var/log/messages?

rwp: no

And then I would check the SMART data from each of the drives, starting with the ada0 and ada1 drives in the cache. smartctl -l error

Then I would look at the "zpool status -v" output as -v will add information about specific files that are affected.

rwp: i was assuming it wasn't a disk error due to lack of checksum errors, but i guess i'll check that (i didn't get any error mails from smartd though)

i included the zpool -v output in the paste, no files are listed there, it seems to be only metadata corruption

i wouldn't mind losing the data on this zvol (or its snapshot? as the zvol itself seems fine) since it's backed up anyway, but i'm worried the zpool won't import properly if i reboot now

Oh, that was -v there, sorry I didn't see it because I didn't see any affected files. I guess that is good.

smartctl -l error doesn't show anything concerning, a couple of disks have a small number of corrected errors but that's expected given their age

Here is an example error from my system from earlier in the year that shows the files listed with -v, just for an example of what it might have said: bsd.to/PnYw/raw

Title: PnYw

yeah, i saw that last month when i had an unrelated instance of data corruption due to the ossl(4) bug :-)

(which has been fixed and i don't have ossl(4) in the kernel anymore anyway, so it's not a repeat of that)

In my case it was a hard crash of the system. In my case the affected files were all temporary and log files that could go away. I was able to delete the files and snapshots and then scrub things clean.

i'm going to try deleting the affected dataset entirely a bit later (data/vm/media/disk1) but i need to wait until people aren't using it first. i suspect that might fail though because it seems like this prevents zfs from enumerating the snapshots at all

It's listed as an I/O error though. I don't see how it can be an I/O error if the kernel is not also listing it in a log file as an I/O error.

And if it is an actual I/O error then I expect to see a physical hardware manifestation of the problem. That's good and bad. Good because there is redundancy for the devices, bad because the controller is shared. But the actual problem has yet to be determined.

And if it is not an actual I/O error then it does seem like deleting the affected data, scrubs, and such should clear it out. But if it is not a hardware problem then why did the array get into that state?

I know you have asked yourself all of these questions already. I am really just confirming your analysis.

i think EIO in this case is just a general error code when ZFS can't read the data for some reason, not an actual disk I/O error

i'm waiting to see if scrub fixed it, that's going to take a while though: 2.23T / 18.8T scanned at 3.20G/s, 519G / 18.8T issued at 745M/s

unixwitch, I'll just mention some random thoughts and you pick out any that might apply. Are backups otherwise good and current? Do you have a spare disk on the shelf just in case? I run a SMART short selftest daily on every disk, I stagger the times so only one test is running on one disk in the array at any time.

rwp: i have a cold spare disk but since there's no indication anything is wrong with the disks, i'm not sure what i'd do with it :-) only part of the pool is backed up because there are very large, but low-value files (~10TB worth) that i don't want to pay backup costs for... i thought that was a reasonable compromise, but now i'm running into monthly data corruption issues, i might rethink that. and yes, smartd is set up to run ~weekly self tests

Depending upon the drive I used to be able to run long selftests but on the drives I currently have the long selftests are several hours long and degrade performance for all of that time and I stopped running long selftest on my drives. The short ones are okay still though. YMMV.

i have it staggered so each disk runs a long test at a 2-two interval, so each disk is tested once per month. the test takes ~18 hours on these disks though so it still hurts performance a bit... i've been meaning to look into running sectioned/partial tests

s/2-two/2-day

(aiui, that allows to e.g. test 2TB of the disk per day, so you can run all the tests during off hours... but i don't know if SCSI disks support this or if it's ATA-only)

There is no canonical right answer. You have to look at your own setup and make a judgement call individually. It's all different depending upon arrays and the collection of drives and the controller used.

I am still fixated on the zfs array reporting being an I/O error but the kernel not logging an I/O error. That's shaking my model of the machine.

Your data is probably okay though.

rwp: zfs returns EIO whenever it can't read the data, which includes corrupted pools. it doesn't always correlate with a physical I/O error

it *should* at least correlate with a checksum error but in this case it seems like either a ZFS bug, or some sort of memory/cpu hardware issue

(i'd say controller hardware issue, but everything is either mirror or raidz, so a controller issue should have been caught as a checksum error)

I am just working from the experiences I have had so far. But I am very happy that I have not hit every storage array problem! Happy to avoid them. But have hit a few over the years. So just relating based upon those. (The saying is, We are always fighting the last battle.)

I look forward to hearing the results of the currently running scrub. Let's expect that the scrub will repair everything unless and until we learn otherwise.

Good luck!

Hmm I'm reinstalling freebsd and it's taking forever to boostrap pkg

Anyone having issues with the server ?

jb1277976: try Ctrl+C and restart the bootstrap?

"the server" is a CDN

Yea did that and tried another they same issue

That's probably why

Anyway to fix ?

Even if I ssh to it it still gonna try to bootstrap

jb1277976: does your networking work? can you 'ping' anything?

oh is this during bsdinstall

Yes I'm in the network chatting here

It finished let's see how installing xorg goes

It's stalled

Can I change the ftp or something

you can edit /etc/pkg/FreeBSD.conf, i think you're supposed to copy it to /usr/local/etc/pkg/repos.d/FreeBSD.conf and edit it there though

'repos' not 'repos.d' sorry

Can I change the mirror there ?

jb1277976: yes, just put in the URL to the mirror you want to use

like i have this: url: "pkg.eden.le-Fay.org/${ABI}/latest" (but that particular example won't work for you since that's a private mirror)

Where is the original freebsd.conf ?

/etc/pkg/FreeBSD.conf

unixwitch where did you find that mirror?

jb1277976: i own that particular mirror, but you can find a list of public mirrors here: pkg.freebsd.org

Title: pkg0.nyi.FreeBSD.org

(in the 'other mirrors' section at the bottom of the page)

Thanks

unixwitch can you pastebin your freebsd.conf I'm getting no srv records

Want to make sure mine look the same

jb1277976: bsd.to/TTmW

Title: dpaste/TTmW (Plain Code)

jb1277976: note that only pkg.freebsd.org has SRV records, you don't need those if you pick a specific mirror (pkg.XXX.freebsd.org)

Ok

Brb

back up

what was the issue/fix?

unixwitch: The "scrub of death" is a bit of a myth.

unixwitch: Here's some reading for you: jrs-s.net/2015/02/03/will-zfs-and-non-ecc-ram-kill-your-data

Title: Will ZFS and non-ECC RAM kill your data? – JRS Systems: the blog

(Funny enough, I was just listening to a podcast about misinformation and disinformation: bigpicturescience.org/episodes/skeptic-check-naomi-klein

)

meena: tl;dr? i have ECC RAM so i don't thin kthat applies to be

and i have no idea what "scrub of death" refers to

I'm not mason

oops, sorry

mason: ^

Also, i should learn a romance language