This is the M.2 in question: ssd.userbenchmark.com/SpeedTest/4874/SAMSUNG-MZHPU512HCGL-00000

Title: UserBenchmark: Samsung XP941 PCIe M.2 512GB MZHPU512HCGL

Nothing by modern standards, but not bad.

It's the stupid card. Maybe I should get a modern one of those cards. THere are chepies for like $12.

But I do wonder how well it will work in this old i7 920 mobo

It's old spare parts junk. But it does seem to serve out the 5Gigabit over the wire. So I am happy.

With an old Sun Intel x520 card (ix driver)

my Dell laptop's screen flickers quite a bit with FreeBSD installed, and it becomes quite unusable; I've checked quite a few forum posts and sites with similar problems but there seem to be no real solutions. Tweaking the xorg.conf helps but only to a point. I must be missing a setting in the kernel that allows the AMD APU to effectively use the GPU cores and memory, or a driver that is needed for

that use. Any experience would help.

bsd-hardware.info/?probe=e54be582a7 <-- hw probe of the laptop

Title: HW probe of Dell Inspiron 5555 #e54be582a7

Holy craps, I think I just figured out how to replicate my root-on-zfs pool to an external disk reliably :o I'm a slow learner

its pretty simple

It's taken me over 2 years

The trick was to mount the pool with zpool import -o ro -R /mnt/altroot so the OS doesn't write to the disk making incremental snapshots sending break

Then send each dataset with send -v -I @$START_DATE zroot@$END_DATE | zfs receive -F zroot-backup

Using -F on the recv side fixes previous errors I had when the pool was mounted r/w

Send each snapshot, rather

I can't find zpool import -o ro documented anywhere though

shockingly, in the man pages

I've found -o readonly=on in man, but not the shorthand -o ro. Do you have a link handy?

man zpool-import

look at -o

Yeh, it doesn't mention any props, specifically ro

Comma-separated list of mount options to use when mounting datasets within the pool. See zfs(8) for a description of dataset properties and mount options.

See zfs(8) for a description of dataset properties and mount options.

and mount options.

Not in that man page either

ro is a mount option for filesystems, not specific to zfs

this is in zfsprop man page: "The values on and off are equivalent to the ro and rw mount

options."

Still not obvious to me that you can use filesystem mount options with zpool import

tehpeh_, With you. "-ro" could be an option for "mount"; zpoolprops does not list that

kitty terminal emulator feels kind of slow at startup in freebsd for some reason :(

All good, I'm just happy it seems to have fixed the issues I used to have :)

tehpeh_, Wait; there a mention in "zfsprops" for "readonly" property: "The values on and off are equivalent to the ro and rw mount options". So that would be why "zfs mount -o ro ..." work

tehpeh_, Aye👍

Yeh, that's how I understand it now. So I could also use "readonly=on".

I also couldn't find a way to _not_ mount the filesystem during import *shrugs*

tehpeh_: canmount noauto

ah, thanks grahamperrin!

<openzfs.github.io/openzfs-docs/man/…7/zfsprops.7.html#Native_Properties> more accurately, canmount=noauto but you get the idea :-)

Title: zfsprops.7 — OpenZFS documentation

actually that's a good point, I could have set that permamently on the whole pool

because it's just an external backup disk. No need to mount it if I'm only sending snapshots over

tehpeh_: you can also encrypt the file system, if you like.

grahamperrin: currently using geli for that, but will check out native encryption soon :)

geli oes not encrypt filesystem, it does block on your disk

filesystem lives on top of that

tehpeh_: PAGER='less "+/Temporary Mount Point Properties"' man 7 zfsprops

angry_vincent: same result for me

thanks llua!

right, it is just not technically correct to use filesystem encryption of zfs encryption or pool encryption ( while there is native one since OpenZFS 2.0 )

Not sure I'm following

ok :)

<grahamperrin> "Subsequent successful runs..." <- Follow-up to that part of chat on Saturday: I sent an e-mail to pkgmgr@ with a different example,

emulators/virtualbox-ose-kmod packaged for FreeBSD:13:amd64 but not listed at e.g. recent <beefy14.nyi.freebsd.org/build.html?…1amd64-quarterly&build=a1f9d8a68e23>

does anyone own a contabo VPS over freebsd ?

it says that freebsd is available, but i can't choose it in the interface

tehpeh_: I import the backup pool before running back-up and instead of not mounting stuff I use the altroot argument to zpool import, which also means the pool will not be automatically imported after a reboot, plus I can manually check the results via the altroot mountpoint

hi guys, would like to create a small freebsd infrastructure, and have only 1 node with access to internet, how can I patch the other FreeBSD machines who can only reach the machine which has internet access. Is there any tool/syntax which could help me create a repo clone locally ? I know that linux has reposync and createrepo, can't really find anything for freebsd or am probably looking in the wrong place. Any info on this would be greatly appreciated.

Are you looking to install software other than the base system?

L3Fr0g: sysutils/poudriere-devel is probably a good starting point

On second thoughts, maybe overkill

not all hosts will have the same software/middleware installed, would like to include those packages in this mirror as well

not possible to just rsync /var/cache/pkg ?

ah

some of them will be hosting database, while others web apps and so far and so on..

i would like to use 1 repo server which will probably also server as a jump host

L3Fr0g: pkg fetch $packages on the internet host, then rsync /var/cache/pkg

Hello folks! I need help troubleshooting an issue. About 1 hour after boot, my Zabbix Server host stops accepting any connection. Kernel only says "kernel: [zone: udp_inpcb] kern.ipc.maxsockets limit reached".

then on other hosts : pkg install $packages ?

not sure it would work

just wondering

i guess you will have dep problems

However, kern.ipc.maxsockets = 62671, "netstat -a" only shows ~500 rows, and "sockstat -n" some 350 rows. How can I hunt down the source of this problem?

L3Fr0g: docs.freebsd.org/en/articles/hubs ?

Title: Mirroring FreeBSD | FreeBSD Documentation Portal

specially docs.freebsd.org/en/articles/hubs/#mirror-pkgs

Title: Mirroring FreeBSD | FreeBSD Documentation Portal

it seems to be the way to go

L3Fr0g: another way ? espinosa.io/blog/2015-02-21-proxy-cache-for-freebsd-pkg.html

Title: Caching FreeBSD packages with a proxy server - Allan Espinosa

you will still need to pkg fetch from the internet host

but i would then be available directly from the cache to other hosts

s/i/it

nice

thank you very much !

this is what i was looking for !

problem is that hosts without internet will get packages from cache which will get packages on internet

not a problem for you ?

i know, this is not a problem

ok

also, just for my information, what kind of homeserver do you guys use for backing up your mobile phone devices ?

what do you want to backup, only data ?

data, contacts, os, etc..

for data : syncthing

.oO( I wonder if there is a way to kick off fetching dependencies on the remote host if local host is missing those ... )

for contacts : davx5 with your mail host

for os: i use calyxos on a google pixel, it embeed a way to backup the apps settings and all

locally, then i use syncthing to export it

.oO( ... I think "pkg" "should" ... )

isn't there any app who can cover all ? :-(

Title: CalyxOS Features

L3Fr0g: if there is one, i don't know about it

if you're not a syncthing user, you should take a look

What about Nextcloud ? i've read something about this, anyone using it ?

it allows you to sync datas between phones and desktops

nextcloud is centralized

centralizing is not a good solution

well, i just use it for my personal devices, lol

what does it change ? if your server is down, you don't get access to your data

i don't want to backup data in google or any other cloud, would like to keep personal data locally

with syncthing, data is replicated locally

nice, thanks !

L3Fr0g: check this eoli3n.github.io/2021/12/21/degoogling-android.html

Title: Degoogling | eoli3n

Syncthing use server discovering, you don't even need to open tcp ports or something

just install it on android, on freebsd, linux or whatever and connect hosts with qrcodes

(with syncthing over zfs, you can auto snapshot to avoir to replicate an accidental deletion)

avoid*

very nice, thanks a lot !

What hypervisor type do you recommend for a productive infrastructure ( bhyve/xen ) with HAST ? Also, are there any cool webUI's or tools available which would help managing the VMs better ?

Although i'm aware that webUIs are not the way when working on something like this, it's just for my own curiosity

I've living the last 20years without freeBSD in my life and now i want to come back to unix, have had enough linux, lol

at least the physical infrastructure i would like to set up on freeBSD, VMs can stay linux, don't really care about that

Also, any puppet or ansible like tools would be very much appreciated, not sure what you guys are using for your environments

ansible works with freebsd without any problem

i use bhyve with virt-manager, which supports bhyve nativly

for ansible, there are useful modules in community.general : docs.ansible.com/ansible/latest/col…community/general/pkgng_module.html

Title: community.general.pkgng module – Package manager for FreeBSD >= 9.0 — Ansible Documentation

Title: community.general.sysrc module – Manage FreeBSD using sysrc — Ansible Documentation

and about virt-manager : virt-manager.org

Title: Virtual Machine Manager Home

it works through ssh, i you have access to your bhyve server, you don't need anything more to configure

wow, thaught that vmm is only linux-ready, lol

hardware accelerated virtualization is based on CPUs (specifically, VMENTER and VMEXIT)

what is the MIB for kern.stacktop on 14.0-CURRENT? I've seen 1.40 on phabricator, but that's something else in the user headers

How new of a -CURRENT would it be in?

Mine's about a month old and doesn't have that OID or any descendents of it.

my mistake, looks like the changhes were reverted reviews.freebsd.org/rG7a38cb1e4e1bb7edba0ef3c5c93f2aef75242ee3

Title: rG7a38cb1e4e1b

'

Welp.

Boohumbug: you have a stray quote

', there, now the channel is unbroken

phew.

maxfx: command swapinfo -h

What's reported?

Remilia: interesting, the Discord CDN stuff. It seems that my everyday Firefox profile is somehow extended to play, instead of download, media.

Are you a Firefox user?

Found: <github.com/Rob--W/open-in-browser#readme>

Title: GitHub - Rob--W/open-in-browser: A browser extension that offers the ability to open files directly in the browser instead of downloading them.

V_PauAmma_V: <bugs.freebsd.org/bugzilla/show_bug.cgi?id=248562#c2> indeed. Sorry, I have been intentionally avoiding references to lists.freebsd.org URLs until one or two bugs are fixed (I can't predict whether fixes will be retroactive).

Title: 248562 – intro man pages need a review

Oh, right. I was wondering. Thanks.

I have a dedicated fbsd zfs machine that I am thinking of virtualizing with xcp-ng. Will zfs work ok if I do this, any pitfalls?

plan is to pass the raw disks to fbsd os and hardware does support vt-d

quit

Might help if they stick around a bit longer when waiting for an answer.

I only had commentary anyway. Moving from bare metal to a VM is likely to have reduced performance. And perhaps more limited RAM as a VM.

still here..

rwp: didn't catch your prev mesg but yea I would expect some reduced performance but would that be significant given its passthrough disks?

I think the only way to know is to "suck it and see" as they say.

yea :)

A lot will depend upon the difference in hardware of the two systems. Say the old one was a standard deskside with 3 Gbps SATA.

Then moving to a server with a fast multi-channel SAS/SATA 6 Gbps controller might be faster.

Or it might be the reverse that the bare metal has better I/O performance and more RAM than the shared hosting which must compete with other VMs.

hardware wise it’s the same server, whats different may be the resources I allocate to the vm

I look forward to your feedback on how it worked out for you. :-)

lol, we shall see.

I'm expecting it to be fine but again yea, I'm updating my backups before anything..

I'll just ask a generic rhetorical question as a just in case... This move does not create any circular dependencies among the entire set of everything?

Make sure that from a cold power off start and power on boot that everything can boot without becoming stuck waiting for something that is waiting for something.

I am sure this isn't the case but I have seen people decide to have the host system hard NFS mount and wait for an NFS server. And then have that NFS server be running as a VM on the host would not be a good thing!

I also have a personal rule that an NFS server should never also be an NFS client. Those two things should never overlap! Avoids a lot of problems.

I say this because when I hear about virtualizing a zfs server then it makes me think NAS system.

Good points yea. As of now there aren't circular deps of that kind no.

One of the main reasons for doing this actually is to run my router virtualized on the same server which does have the negative drawback of if server goes down, router, backups and everything go down.. But I am trying to save some power and that server has too much resources to just let them sit idle.

adonis: you want a disk controller/HBA that can be passed through to the guest, so that ZFS maintains complete control over the cache, otherwise you can get into a situation where ZFS thinks data has been flushed to disk but it hasn't.

I am currently using a classic Banana Pi as a house router. It's good for 750MB/s through it. It's currently running Devuan. 2 watts. Independent. Stable.

It's not so much a question of performance, as it is of the atomic transactional nature of ZFS requiring that level of control.

debdrup: yup I read up a bit on that. Supposedly xcp-ng allows full passthrough of an hba which I have 4 disks connected to as well individual disks as well.

My point is that you shouldn't trust the hypervisor with it, as hypervisors have made the claim of flushing data when they didn't.

So I shouldn't trust it saying its doing passthrough necessarily?

Depending on your location, availability of proper SAS HBAs that can be/are flashed to initiator target mode should be fairly decent.

I wouldn't, even if I knew the hypervisor well (and I don't know it at all).

Ok. As far as my hba yea it's flashed to IT mode.

Which one is it?

Just curious :)

that’s the way its setup in my current bare-metal setup.

hmm one sec

debdrup: It's a LSI SAS 2008

as for the exact model I'll have to open up the chassis for that :)

dmidecode maybe?

or ther other one is it pciconf -lv ?

pciconf -lv is what gave me the above

SAS2008 PCI-Express Fusion-MPT SAS-2 [Falcon]

is the exact output

You can use mprutil, mpsutil, mptutil, or mfiutil depending on the actual hardware model.

Check which driver that device is attached to, that should tell you which -util to use

how do I check the driver it's attached to?

grep -u sas2007 /var/run/dmesg.boot

see what Device:

it got

err 2008

and grep -i

:)

mps0

mps – LSI Fusion-MPT 2 IT/IR 6Gb/s Serial Attached SCSI/SATA driver

'man mps'

so mptutil

just based on the naming

SAS9211-8i

:)

:-)

mpsutil show adapter shows this

0-7 SAS Initiator

that’s IT mode right?

0-7 being the PhyNum

no idea :D

I would assume 0-7 are the 8 ports on the card

or well yes phy's

that's what they are yea

What is the syntax for jail.conf.d/* files? Same as jail.conf?

What else would it be?

Just the inside of the jail block possibly?

Do I need to do anything else so the files are detected? Just doing `jail -c foobar` says it's not found.

Is the .d specifically for rc.d?

mps - LSI Fusion-MPT 2 IT/IR

Says IT in it

;-)

Erhard: it also says IR, so that's not much to go on.

ccx: Well, I don't know exactly where it's from (I think Xorg?), but the .d is a convention used to indicate that a directory contains configuration files of a certain type.

You still need to follow the format of the configuration file inside them.

It's just there so that if you're managing a lot of jails but still want to use the in-base tools, you can now put one per directory with jail.conf being used to define defaults.

Erhard: daemon printed that out from the man page, that doesn't necessarily specify what mode my card is technically in

There's a way to use jail.conf to define basically every default, so that jail.conf.d/ only contains a couple of lines with the jail block and its IP.

Oh, I see

adonis: I don't have a system booted with that SAS controller and it'd take minutes to start up (server POST times...), but I'm pretty sure there's no real way of knowing short of having flashed it.

Does it list the firmware in dmesg|grep mps

mpsutil(8) might show it.

The version or anything?

dmesg definitely doesn't.

I saw it somewhere yea

MIne shows firmware version in dmesg

But does not specicifally mention IT

mpsutil show adapter shows it

20.00.07.00

debdrup: I don't think that's the case actually. It seems to be specifically for /etc/rc.d/jail which tries to guess proper config file for jail command and only one is used.

Yeah, that is the version I have in IT

mps0: Firmware: 20.00.07.00, Driver: 21.02.00.00-fbsd

from dmesg

I don't remember if there's an IR mode for that firmware version, though - so at best, it's inference.

That is the standard IT firmware version. Could also be an IR version perhaps.. But unlikely

I'll grant that it's unlikely, but epistemologically we're no closer to an answer :P

:)

Are we ever?

Erhard: for certain values of sometiems.

Erhard: weird for me is that dmesg | grep mps doesn't show firmware or driver..

If it printed IT mode somewhere, we'd be as certain as we can be - with is epistemologically more certain than we are know.

That is weird.

Mine is on TrueNAS.. maybe they added that

oh ok

I doubt it.

I won't claim iX upstream everything, but they've been pretty good about it.

This box has a different card, though: mps0: <Avago Technologies (LSI) SAS2008> port 0xe000-0xe0ff mem 0xfbec0000-0xfbec3fff,0xfbe80000-0xfbebffff irq 40 at device 0.0 on pci3

debdrup: so mpsutil show adapter saying SAS INITIATOR for all physical interfaces is not enough?

just wondering if that implies IT mode or not.

adonis: I think that's SAS Initiator port.

ok

SAS probably inherited that nomenclature from SCSI, which is also where the notion of initiator targets come from.

I won't guarantee that this card is in IT mode but I'm sure (within margins of human error lol) that I put the card in IT mode specifically so it can be used as an hba for zfs :)

Ah, so much fun. Get your centronic termintor right...

mpsutil can be used to reflash.

Do: sas2flash -list

Then look for Firmware Product ID : 0x2213 (IT)

Thought aparently that is only positively affirming, as some card won't show the (IT) even when they are ;-(

adonis:

just out of curiosity ... you did just try dmidecode (for if it could determine the precise model) right :)

Erhard: what port installs sas2flash?

No clue. It seems to come with TrueNAS

lol I'm not running a nas

you can steal the images from truenas/opnsense without much issue

being based on freebsd and all anyway

couldn't find a port that installs that actually

daemon: working on install dmidecode

maybe its base, slap it in a vm maybe can just cp it over

for some odd reason doing pkg install anything is taking a long long time

ok I know why. I added ipv6 and for some reason that’s not working too well.

Maybe DNS

I am not sure where sas2flash comes from ;-)

It may be from Broadcom

ipv6 works on my network, my config for my bsd machine probably needs to be tweaked a bit

yeah its a broadcom tool

as an aside there is apparently a thread on an 'easier alternative' truenas.com/community/threads/sas2flash-easier-alternative.30789

Title: SOLVED - SAS2FLASH Easier Alternative | TrueNAS Community

might be helpful

His real goal is to determine if he has IT mode firmware on the HBA

reading the thread it seems they come default in 'ir' mode

Yes.

Though he has the firmware most usually associated with IT

However, that versions also has an IR mode (~5% slower from what I find)

is there any safe quick test

Do: sas2flash -list

I will need a bit for this as I think I will need to restart my bsd box for networking changes but it's in the middle of a zfs send and I need to let that finish first.

Seems at least affirm IT if it says IT

from the thread it also seems the best way to play about with the sas2flash tool is in msdos

Oh, fun!

:-)

lol yea.. I remember using that tool for the flash itself

booted up in FreeDOS

if I understand what is going on truenas/freenas has a sas2flash tool natively, ported from an older version that can do read accesses fine it simply cannot write

so ... using that logic if you are going to boot to another os to piss about with tool may as well just make a freedos boot

maybe you still have the usb key handy? :D

nope lol

that’s optimistic thinking haha

:)

I can send you the binary, but you would be crazy to run something from IRC ;-)

name it something you would trust from irc like 'medical notes.exe' ;D

lol

that's ok. thanks. I can grab the flash tool bootable usb again.

so I think my issue is the ipv6_defaultrouter

there is a sysctl somewhere to control what ip version is tried by default

does /etc/rc.d/routing restart put into effect a change on that command?

on that arg sorry.

brb

It's not _really_ a Broadcom tool, it's just on their site because they got bought by Avago who also bought up LSI.

sas2flash, that is

yummy yummy conglomerate munchies

Well, yeah

Copyright (c) 2008-2013 LSI Corporation. All rights reserved

ATTO SCSI initiators, lol

As in fibre channel HBAs?

I got two in 1998 I think and they just worked forever until PCI died out and then I lost them during some move or whatnot

never had much luck with LSI Logic and Adaptec

I always had great luck with Adaptec.

'Course. . . I was also good about firmware updates and being really anal about how things were cabled & terminated.

I will explain a thing I have explained a few times before, as I have been involved. The difference between AVX and AVX2 is looping over an array, such that the compiler gets a list of addresses of valid data. Then you accelerate the logic in the pipeline for what is called gather-scatter.

what is the appropriate way of retriving the size of memory currently used by a specific process?

root@gateway:/usr/ports/sysutils/sas2ircu # top -b | grep dnsmasq | perl -e 'while(<STDIN>) { my @cols=split(/\s+/); print "$cols[7]\n" }'

4052K

:-)

:D

wouldn't that be $cols[6] tho?

1024 nobody 1 20 0 16M 4052K select 2 4:11 0.00% dnsmasq

ah its split on space

\s+

so nope 7

beacuse of the space at the start

but surely there is some other base supplied utility to retrieve this number from a PID, no?

I mean you're literally using perl

(nothing against perl)

I mean hack it with sed/awk from base

alright, so that is the way to do it properly?

there is no memof or similar?

I think its the only way to do it that I am aware of

not even sure procfs can do it if mounted

that being said I mean you could look at the code for top

there must be some native C code to actually determine it

certainly. I was looking for a shell-only solution so far. I did do the awk thing but felt like there this might be not the most efficient way (from shell-only perspective)

its so little overhead on todays ott hardware ~ well yeah I mean that is likely what happened

top -b exists for interaction with shell scripts etc

didn't know about -b so that's good, thanks :)

and its so little processing no one cared enough to make 'howmuchres' as a binary

np

no point in building such a binary then I guess

you could but I kind of doubt anyone would ever use it other than yourself, the internet is glued together in perl and all ;p

jbo: ps -p 34 -o rss

substitute 34 with desired PID

and rss with desired memory statistic, there's a few

Between 2012-2013 I was at CERN OpenLab for GEANT(5), also referred to as Intel Lab with access to engineering samples, with weekly meeting engineering debug meetings where I met several Intel research lab staff asking me weird questions. For all intents and purposes given diplomatic attache status, the work is open and free.

If you have /proc mounted you should be able to use that from scripts, but that's no longer the default I believe.

-o rss= # to omit printing the header

What gather-scatter actually involved, was discussing software and physics theory with legends as the code was awful. Involving tracing particles in suimulated through a detector where polygons are not used, where a screw or a plate is an accurate geomatreic object with different math, and how to distribute and parallize this in a way compatbile with AVX.

heh, i have a customer sample board that was at sandia national labs before my workplace ended up buying it

(the board, not the labs)

lol

I still have it and the engineering sample chips it came with

(e5-2680 chips in the end, but they didnt have that designation at that point)

And for what it matters lattera, why didnt i publish? I was mentall exhausted, and Sverre Jarp was retiring, and my advisor too busy complaining about float precision from MPI results - needing more time than the average person.

zpiro: I have a sneaky suspicion that you're a bot...

lattera: maybe not, twitter.com/BalderOddson/status/1582086301007826944

Title: Olav Smorholm on Twitter: "For who's whom in Eastern Asia, trust in China would involve orientation of whom everyone is from ancient times? Who's landlocked? Now or usually, and never by necessity. Fixing North and South, no matter how attractive it is to look elsewhere." / Twitter

what's the policy on annoying random irc bots in this channel?

you dont like VimDiesel ?

Title: IRC/Bots - FreeBSD Wiki

I'm not a fan of being tagged in random bot spam

lattera: I could ask you to ask Ben Segal, or the fortran guy behind original GEANT that knows the danish guy behind c++ and adviced Berners Lee, but wouldn't be NexGen.

zpiro: your comments seemingly came out of nowhere and had little to do with FreeBSD

-sigh- markov bot?

or just really tired college professor?

HARD TO TELL!

ISTR there were spam bots that replayed old channel logs, moslty aimed at defeating content-based spam detection.

thumbs: the previous ones were, and are true.

thumbs: on topic that is.

thumbs: and generally, lives necessities and coincidence have their issues.

how is it directly freebsd related?

it should probably go in #freebsd-social

(if what you're saying is legit)

z_pir0 seems pretty spammy.

zpiro: what is freebsd?

Seems he was turned off.

Besides, everyone knows it's "a kind of linux."

I hear its better for running ftp sites with ;-)

I know, it's part of the base OS: /usr/libexec/ftpd

I'm sure I'll receive much hate for the sentiment, but I miss having bind in the base OS, since I've used FreeBSD for so many DNS servers.

ISC is deprecating bind anyways

(or rather, has)

nacelle: it is an evolution of the bsd kernel, that respect the intent of orgiginal purposes and function of the sub function of this classical unix offering, and want it to be free, as in truly free for praid, non-paid, gratis and all eyes around.

CrtxReavr, if it makes you feel better I think perl should still be in base :P

you know what we should do make a 'prebsd-pack' pkg that has bind, sendmail, perl etc... in it

"Alex, what is Dissociated Press?"

a one shot installer to put in the missing gpl stuff most of us know

zpiro: what is a bsd kernel?

daemon, I gotta disagree there, especially since replacing perl in the build system with awk was such a boss move.

I mean. . . awk was always there, but replacing perl's role in the build system, was a master class change.

nacelle: curious question, assuming the question of a kernel is not involved, it involves a respect of unix philospophy and clearly controlled sub systems.

crinstall anything, apache nginx postfix exim or any DE and what gets dragged in

its not that I have a big problem with things like bind or perl not being there, but the moment you install almost anything they get sucked in anyway

daemon, be sure to tell me which hosts you install it on so I can sic the Morris worm on them. :-)

:-)

zpiro: why do you speak like no one else I've ever spoken with does?

zprio is the AI that got that guy at Google fired.

nacelle: say no to .no is a slogan i still struggle with, almost like its racist.

the troll comes from here, apprantly.

zpiro: since you can't demonstrate to several channel regulars that you aren't some variant of a Markov bot, I'm going to quiet you if you don't stop this right now.

debdrup: you are welcome to that.

o_O

you know what

Am i know? Oh well.

nacelle: am I a markov bot?

probably?

thank you

CrtxReavr: have you had a chance to play with kea?

ok so I'm having some issues with ipv6 on my fbsd box. I have the ipv6_defaultrouter set and I can ping it. But doing ping google.com for example doesn't go through.

my interface ix0 has a public ipv6 and link local one. ping google.com I can see trying to ping the resolved ipv6 address for that domain name but just doesn't go through.

Which FreeBSD version?

13.1-p2

I have rtsold_enable="YES"

and ifconfig_ix0_ipv6="inet6 accept_rtadv"

Do you have a firewall on that computer? ipfw2, pf, ipf, or whatever?

the fbsd box has no firewall but the router I specified in ipv6_defaultrouter does

they are both on the same subnet though and firewall rules allow comms between both boxes

OK. Try traceroute6. That should tell you something about who drops the packets.

packets do make it to the router as per traceroute6

And not beyond that?

ok.. well I did traceroute6 directly to the router, not that helpful I guess. Doing traceroute6 to google.com for example shows the first hop not going through

so 1 *

nacelle, I have not. . . though I have written copious shell scripts to do automation with dhcpd.

adonis, can you pastebin the output of ifconfig & netstat -rn

ok

Um. . now would be preferable.

ifconfig pastebin.com/uPbadFyS

Title: bruce-ifconfig - Pastebin.com

netstat: pastebin.com/M1b8Vh6s

Title: bruce-netstat - Pastebin.com

ix0 is your egress interface?

on the fbsd box yea. its pretty much connected to my switch which connects to my router

so ipv4 works for example

Why do you have v6 IPs from two different prefixes?

good question lol

On ix0?

I can actually ping 2600:4040:a689:8301:ec4:7aff:fe1f:7724

not sure tbh, let me restart the machine. It may have to do with some config changes I made and routing restarts..

So the routing is working.

by it might be defaulting to the other IP as your source IP.

fbsd restart in progress :), let's see if I still get two prefixes, if so maybe a bad config on my router..

What happens when you: ping6 ff02::2%ix0

going to have to give me a few minutes there while I wait for that box to restart, it's slow to restart.

Why restart it?

because I had made some ipv6_defaultrouter changes + service routing restart's prior and maybe that's why I have the two prefixes?

just want to get back to state 0

ok yup i see only 1 prefix on it now

and there you go ping google.com goes out via ipv6 and works

hmmm.. now wondering why the two prefixes and why two prefixes causes an issue with routing..

adonis: I understand the instinct to restart if you're used to other platforms, but with FreeBSD the only useful thing you can do with a reboot that can't be accomplished any other way is to update the kernel.

So far as I remember, nothing in the base system has in-memory state that can be different from what's on-disk, unless it's not been SIGHUP'd

..since the last recent change.

And clearing out all your nuetrino bit flips... ;-)

debdrup: yea I understand that. service routing restart wasn't enough though, most likely because I need to restart the iface itself..

Not sure a reboot accomplishes that, maybe a cold start if you pull the plug from the machine and hold down the power button to discharge all the capasitors?

Reboot is a CPU reset, but it doesn't affect hardware state beyond that.

still felt a restart was faster then trying to guess if that iface restart + routing restart was going to be enough.

A reboot fixes bit flips because it assumes nothing about eh state of RAM when it boots.

adonis: there's also the netif service that can use a restart

I htink I had to both netif and routing restart, and there is a specific order I don't recall.

I found it easier to reboot since I was dealing with VMs at the time

But if you are running enterprise gear that takes five mins to reboot that is a pain

A few minutes? I'd love that.

Initiating Lifecycler Controller...

My bigiron is approximately 10 minutes to a boot.

Yeah, close to that for the Dell junk

Well, "bigiron" compared with every other box I own.

Anywho, bedtime.

Have a good one.

Erhard: 5+min restarts are definitely cool :)

definitely not*

ask a 1970's IBM tech about IPL time

I like manually setting bits with physcial switches to load code for booting.

Speed reading …

adonis: did you restart the OS to work around a networking issue?