Hm, alepzi needs to stick around longer.

my jail's /usr/local/etc/rc.d/quickwit is termbin.com/1dvy and /etc/rc.conf has quickwit_enable="YES". on jail startup it fails: "root: /etc/rc warning failed to start quickwit". in the logs it says "cant assign requested address os error 49". but after jail startup, sudo service quickwit start works. so is there a bug in jail or vnet that when it's starting up it fails to let networking work?

alepzi: stick around after asking

alepzi: So, if you can start it manually once the jail is up, but you can't start it automatically, that indicates a race condition, and the service almost certainly needs to start later.

This could indicate a bug in the specified dependencies.

Hm, but it does want networking there.

digging for a sec

ty. and sorry bout network flaps, i finally figured that out

Well. That looks like it should work. So that is strange.

alepzi: Certainly worth a bug report.

alepzi: How are you setting up networking in that jail?

the standard hooks from the handbook page sec

docs.freebsd.org/en/books/handbook/jails/#creating-vnet-jail the sample config in that section

Title: Chapter 17. Jails and Containers | FreeBSD Documentation Portal

alepzi: Can you use a pastebin for the example? Or specifically what I want to see, is your exec.start including /bin/sh /etc/rc after the network config?

My theory is that if you're calling "/bin/sh /etc/rc" first then you've got a race.

bbiab, got to start cooking, but I'll check in in a bit. But it's a race if you can start it manually after the jail boots, and I'd be inclined to look at your exec.start ordering.

mason: termbin.com/ai4x

alepzi: Yep.

alepzi: Put 'exec.start = "/bin/sh /etc/rc";

'

as the very last thing you do in exec.start

alepzi: I gave an example here, using either static config or DHCP: wiki.freebsd.org/MasonLoringBliss/JailsEpair Note that "/bin/sh /etc/rc" is the very last thing you want to add to exec.start.

Title: MasonLoringBliss/JailsEpair - FreeBSD Wiki

Now going to eat the stuff I cooked, but this should get you going.

omg that works ty!!

any other obvious probs with my jail.conf?

alepzi: Nothing jumps out.

bbiab

mason: if this is such a footgun that you knew it before i pasted the link, maybe the jail service could do a config check when starting a jail and emit a warn on that?

found some more fun with this

igb0 comes up as 100BASE-TX

no clue why, have other hosts on same switch at 1000. trying to force to 1000BASE-TX hangs the NIC

controller is I210

mewt: did you mean 1000baseT instead?

though, i don't know why it would come up with the wrong media type

tried a different cable?

yeah, sorry

but no, it's not a cable issue

I've swapped to new cable. sorry for the spurious x

is there a nice way to run a script upon quitting xorg started with xinit?

currently checking if it forces to 1000baseT, if it doesn't, I guess that was not my mistake

what im hoping to do with it is switch my tmux colors when going back to vt

concussious: i think you could replace your last 'exec twm' with 'twm' followed by another command? as the window manager won't return until you quit

lw: thanks once again! i'll play with that a little.

yep, this looks like a hang...

this is never good.. dpaste.org/9hXRH

Title: dpaste/9hXRH (Bash)

for some reason, my zpool is totally hosed.. but when i try to get a "new" one zpool create storage raidz da0 da1 da2

i can not "destroy" it because i can not import it but when i try to create it, it tells me the drives are to a storage.. in a loop..

i did do a zdb -l on the drives and this is what i get.. dpaste.org/eLSv2

Title: dpaste/eLSv2 (Bash)

well zpool import -nF storage seems to get me further

now it just says recreate the dang thing :(

yeah man, with two faulted disks, that's the end

zpool destroy -f storage it is

or it is dead dead?

dang nab it, looks like 1 of the 3 drives is good.. the other ones will NOT allow me to create a pool

cannot zero first 4096 bytes of '/dev/da0': Device not configured

but yet they show up in dmesg..

restore from backups,

might pick a more robust pool configuration

but don't i need to get the zpool operational first?

i can have data repopulated, right now it is complainining i can not even make the pool

if anyone else has any clue what I can try with that igb0 that won't negotiate above 100BASE-TX, please let me know and I'll try tomorrow morn

i probably could send all the data as server is only using 1.27TB and i have 2.47TB with the one drive..

not ideal

you'll have to destroy the pool and clear out the disks

but run smartmon on them to see how bad they are

if you get 4096 bytes of '/dev/da0': Device not configured... looks like you'll be shopping for some new disks

dang nab it

And alepzi has gone away again, but FWIW "I can do something manually once the system is up but not during boot" is always a race condition, regardless of what system it is. Nothing unique to jails.

And in this case, "the service can't grab a socket" means "the network isn't up yet" and so we winnow down the prospects based on that, and there we are.

rtprio: thanks for confirming

i can at least use this one drive fro a little while and hopefully my sanoid/syncoid thing will populate

if you rebuild it as a single drive zpool, you can add a mirror later

i would suggest using the same make and model of drive if you do that

i first have to remember how i setup this backup system

it has been months

well when you figure it out, document it so you don't have to rough it next time

yeah, i have enough notes to be a pain in the butt for me now

i lost a zpool when the external chassis i used lost power

that is what happened with me

don't use external chassises

1 drive made it ther 2 did not

hehe

yes, we had these talking points in the past

thankfully this is the remote backup server and not the main server (which does have internal drives)

i upgraded my VMs on oracle cloud with freebsd-update and now they're not responding and the commands to get to the console don't work

🤷‍♀️

a high quality external chassis is fine (unless there's a serious bug in zfs or cam) but cheap enclosures using sata/usb i would be suspicious of

mine was sata. it might be handy in the future to do a backup and store it in a safety deposit box but yeah

well at least the syncoid script is TRYING to work

need some datasets

well look at that.. bhyve images are going ..woohoo

oh hey, mpd has a web interface, that's handy

hrm.. i do not have compression on for this

probably should fix that

well lets see zfs set compression=zstd storage/backups does

if you're using syncoid, probably nothing, as it'll just copy the already-compressed blocks

i do not recall setting syncoid to do that.. is that a default behavior?

actually maybe that's not the default

syncoid --recursive --no-privilege-elevation --mbuffer-size=128M --sshkey=/home/backups/.ssh/id_rsa storage/music "${REMOTE_HOST}:storage/backups/music"

that is the command i am running.. to backup music

makes me wonder if i can make this script "better" with just doing that

it definitely is not GREAT 1.32x

blah

we shall see in about 7 minutes when it does the other backups

hi, I am using /bin/sh normally. Is there any way with /bin/sh to go to the first character of the line or move from onw word to another (i.e. the equivalent to 0-$ and b-w in vi)

?

onw=one*

As a side question, does that depend on the terminal program you are using?

sh

sh -V

pastebin.mozilla.org/AXSyYRmk. why is that ?

why does sh not have precmd ?

I am using mc for like 20 years, never had a problem with it ... now I have to run it with -u, even mcedit with -u

so don't run sh. it appears to be an easy fix

I don't trust other shell ...

it literally tells you what else to do to fix it, i'm not sure what you're expecting

for workstation, ok, for servers, no

I want it like it was before ... when it worked with sh ... (if possible)

maybe run mc with --nosubshell like it says?

yeah, for now I will do that ...

or use a different shell

rtprio I did not know there was a VI compatible mode

is there any drawback for using that? -given that you use VI normally-

can I add -V to my user in /etc/password ?

Hello, since a few days I get the following error when running pkg upgrade, could anyone help me out?

Checking for upgrades (1 candidates): 100%

Processing candidates (1 candidates): 100%

Cannot solve problem using SAT solver, trying another plan

I'm using pkgbase by the way

CamilleScholtz: and it's not telling you what problem the SAT solver ran into?

Nope, I also ran it with the debug flag and it's not giving any additional info. I'm now running upgrade -f, seeing if that might do something

Hmm reinstalling everything seemed to have fixed it, strange

Hi folks. I was wondering, what FS would you use on a freebsd vm guest? On linux it's the consensus not to use ZFS there, but what's the way to go on Freebsd?

Devil_Kin: Even when running on VMware as a VM (where we have VM-level snapshots), I still use ZFS on FreeBSD - because we can.

dvl: reason for the question was that it's said often that zfs works best when dealing with the hw directly, not through a virtualisation layer

Devil_Kin: Yes, in general that is true, and I think they are referring to raid. In VMware, we give ZFS a single drive, and let it go.

dvl: ok, thx!

Devil_Kin: It is more convenient to have ZFS on the VM - quotas, snapshots on file systems, etc. Why should we go without just because it's a virtualized host? etc.

yeah, makes sense

well looks like bhyve images, can compress quite well.. mp3/mp4 not so much :) makes sense

ZFS on host is much better than on every vm

yeah, zfs inside a zfs is like the pandora's box of chaos

How so? I've been doing just that.

subjective, personal experience, when i tried to "restore" a zfs inside a zfs. it would NOT work

not sure what exact root cause is.. but when i made the zfs and ufs inside.. the same problem never occurred

so purely a subjective (probalby something i did wrong) reason for not preferring to do that

so i just avoid having pandora's box

Fair enough.

/25

well, i don't have much say over the host in this case

so i don't know what's running there

i made a rc.d script for quickwit which doesn't ship with 1 but it hangs the server because when the server's starting up it outputs to stdout but never goes into the bg. any way to fix that in the rc.d? termbin.com/1dvy

or maybe redirect the output to a log file?

quickwit does not have a way to run in the background itself? If so that's where things will be hanging. Waiting for it to background itself.

You could use the "man 8 daemon" program to do it for quickwit.

ya guess i'll have to tyvm

going to come back once more to ask: I've got an igb0 on a supermicro X11SCH-F that refuses to go above 100BASE-TX while other machines do from same switch/router. How can I diagnose this? Already tried swapping cable

I210 is the controller, specifically

can't find any relevant error output with basic stuff like dmesg. When I try to force 1000BASE-T the interface just dies. I can try locally to see if I can figure out what is happening after that point I guess

Is this at home or in a datacenter? Is this plugging into a managed switch port or a small consumer switch?

home

tried TL-SG108E and my router (wrt1900ac v2) directly

Managed switches can be configured to be anything and I have had managed ports misconfigured before.

let me take same cable and shove it in something else with gbe really quick

If it is a consumer switch then I would power cycle the switch and I would swap ports with another cable.

Yes. Swap cables. Swap ports. Swap devices. Something frustrating is happening.

right now I think I've eliminated the switch already, because I'm connecting directly to a router port

Does that motherboard have 2+1 network ports on the back? Two for the system and one for the IPMI port? Make sure you are plugged into the port on the outside closest to the PCI-e slots.

Yeah, on my T430, the same cable pulled from the X11SCH-F comes up as 1000Mbps full duplex

The port closer to the ram side will be either the dedicated IPMI or the shared IPMI port.

I have IPMI access

That is not a guaranteed test that it will work with other ports though. I have seen that before.

The IPMI gets separate DHCP reservation and I SSH into the machine. I haven't confused it with the IPMI

hm

So igb1/igb0 cannot be guaranteed to be the same even reported that way?

Get a light and inspect the port itself looking for bent pins in it, looking for gunk that may have collected in it.

I *can* try and see how igb1 comes up I suppose, sec

Wait... You said the IPMI gets separate DHCP reservation so is that on a separate second cable? Or is that on this same cable?

there are 3 ports

1 is IPMI, oterh 2 are igb0/1

each has own cable. igb1 comes up as gbe as far as I can see, let's confirm

guess it can be an issue with the jack then. awful

Good. Thanks for the clarification. That's what I was hoping. But it was not clear. That's all good.

I don't like using only the shared IPMI port because there is the danger of misconfiguration such that the IPMI could possibly be accessible to the hostile Internet. And of course that is always to be avoided.

This is on your LAN so it is all fine. But in a datacenter that can put the IPMI on the public net and that would not be good.

Anyway... As to your port not coming up with GigE speed... I don't know. It's almost purely hardware link at that point. Swap everything. Inspect everything. It should work. If it does not then I don't know what to suggest.

Yes, I have blocked the IPMI from WAN outgoing actually

I'll try one more cable swap, then it's almost definitely the jack x_x

I don't recall there being any BIOS configuration that would affect this but there might be. I would boot to the firmware configuration screen and walk through every configuration dialog available and review it. Maybe there is something in there. Reseting to factory firmware config and starting again might be something to try if you suspect it.

As another A-B comparison thing I would download get.refracta.org/files/daedalus/ref…ta_12.0_nox_amd64-20230829_1853.iso and boot it from USB and see if that live boot does anything different. I would use "ethtool" there to dump the configuration of the ethernet.

I think ethtool is preinstalled in that image. If not then just install it "apt-get install ethtool".

...apt-get?

well, i got igb0 to come up as 1000BASE-T playing with cables

so this looks to be a combo of hw and case. Gross

At least you know that it can work if everything is jiggled correctly.

ethernet link is almost entirely encompassed into the chips on both ends of the cable. Almost all of them are MIDI-X these days meaning no crossover needed and can connect direct between two ports. That's another good test if you run into this again. You can avoid a switch and see if the link comes up directly connected to another port.

I take it the case is wedging against the connector? Honestly that sounds like either the wrong backplate or the motherboard was mounted slightly off. I obviously can't see it.

no, just pressing down the clip puts your finger between case + connector

If it is a backplate for a different motherboard then I would remove the backplate. It's mostly there only for EMI reasons. It's not really needed. If it is getting in the way and bending the cable causing trouble.

It came with the thing

Okay. I am just brainstorming ideas with you.

(the motherboard)

thanks for the pointer on direct connect, I have had to do that before and should've here

There is almost no clearance between the case and the clip. I always use a flat screwdriver or my pocketknife or a pci cover plate to unclip there.

I can't quite envision exactly what the problem is that you are having but I am glad that you have an understanding of it now and have seen it work. I am sure you will get it sorted out into a reliable connection now that you are at this point. Good luck! :-)

yeah, I think I'm good now. Thanks. also is clearly my NAS throughput problem, pushing 200M which is not amazing but good enough for me over sshfs now

Please watch the movie "Room (2015)"

it's a really powerful film

I am going to show it every single parent female I know of

it is just -that- powerful

"Held captive for years in an enclosed space, a woman (Brie Larson) and her young son (Jacob Tremblay) finally gain their freedom, allowing the boy to experience FreeBSD for the first time."

ah, that one

entikan: We have #freebsd-social for at

ok

Hello everyone. I am trying to pkg remove plasma5-discover and it seems it wants to:Installed packages to be REMOVED:

kde5: 5.27.11.23.08.5_2

plasma5-discover: 5.27.11_1

plasma5-plasma: 5.27.11

it will break the system, correct?

It will remove your kde, if this is what you mean by breaking the system.

yeah i wouldn't do that if you're using kde as your DE

spmzt, yes. dstolfa, indeed! Thanks. Any tips on how to remove it without breaking the system?

Carbon666: i'm not sure. i think kde assumes quite a deep level of integration, so it might not be possible without breaking the DE

yeah, thanks, mate!

Ok, can FreeBSD use linux swap partitions?

i'm thinking about going back to Linux on my laptop. I can't handle the constant slew of bugs and things taking almost a year before their handled in the bugtracker

not being able to suspend my laptop

it constantly freezing and crashing every two days

wifi glitches

isn't it your hardware?

wacom drawing tablet just doesn't work no matter what

Carbon666: no. Linux had none of these problems. It's also pretty standard hardware

a Thinkpad T430

Isn't that supposed to be the most compatible laptop out there?

gotcha

I am not sure, mine is pretty damn uncommon and runs well. I am trying to add swap to it right now

only reason I don't is because most linux distros are systemd now, or have a gnome team that sabotages things

Carbon666: does suspend work?

hahaha

so i'm stuck between a rock and a hard place

or.. system-breaking bugs, and sabotage

Hum, maybe it complain about acpi or something, sfox. I have this installed for 3 days and had time to use it for just between sleeps

he bios is coreboot-edk2

any upside to having daemons log to their own file or is it always better to log right to syslogd?

literally reference implementation stuff

I am enjoying the system

i enjoy freebsd too

i no longer have any problems with it on the server

I am still finding ways to make it runs smoother

i'm a freebsd enjoyer

oh another problem nobody seems to know how to fix

sometimes the system is laggy

to the point even audio glitches out

especially when downloading faster then 500kbms

thats RAM

600kbp/s

check if you have swap on

or if I compile software using all of the cores

I don't. I am trying to asdd one, ca I use a linux swap partition for that?

i do have a swap but it's only 2gb

no you have to create a bsd swap

you and overwrite linux swap partition with a bsd one

alright, I have seen some tutorials on that, but for lower versions, so I decided to ask in here

it is like windows

you can open and close them?

haha

I've quitted Windows since 8

never used a WIndows 11

have you checked the man page?

Windows*

I am looking up the handbook

but as of now, it only mentioned the auto partition way and didn't mention any manual swap configurations

I need a partition manager to split my linux swap into 2 partitions

one for freebsd-swap

if I open firefox, bye bye all RAM and the system starts to act very hard to try to not "hang" things.

try falkon

it uses less resources

yeah, just transitioning backups through it. and I like firefox :(

ok, so. I am not sure, maybe I will go to my Void Linux to edit my partitions, because I am really unsure about doing it in here

everything looks "the same", but I think linux, and GNU, ofc... it is nice to be around something else...

I am trying to read about the GEOM

aight, gotta go! Keep on rockink!

Rocking*7

there's more showing up in a jail's /var/log/messages than the jail host's /var/log/jail_testjail.log console log. why's that?

.

o/

\o

alepzi: Hey, meant to say last night, what you described is a general race condition - nothing strictly related to jails.

ya i gotta figure out rc stricter

As for the logs, look at the in-jail syslog.conf vs the host config.

if a dir is 755 any way to force files made in it be 644 instead of 600?

i think that's what umask is about

Do you mean files already in it, or also new files?

If the former, use "find /path/to/dir -type f -print0 | xargs -0 chmod go+r" (untested)