there any way to get the bit rot detection & fixing benefits of zfs on shit like usb flash drives?

I submitted a bug and a fix via bugzilla almost a month ago (documentation error). How long do they take to be acted upon usually?

RoyalYork: depends on time of year, work schedules, phase of moon, severity of the issue.. lots of things. Thank you for your submission. It's not wasted, even if it takes a while to process or even acknowledge.

RoyalYork_: which Bug id?

RhodiumToad: That was strange. After making those changes to /boot/loader.conf and disabling/moving things into /etc/rc.conf, I rebooted. My keyboard stopped behaving normally. I have a QMK-firmware based keyboard (Moonlander), which took a bit more setup on FreeBSD when I initially got it. After the reboot, it reverted to behaving as it did when I

first plugged it in, as if the tweaks to fix it were gone.

I ended up putting usbhid_load="YES" and hcons_load="YES" back into /boot/loader.conf and it returned to the expected state. For reference, part of the initial configuration I did to get the keyboard working was building a kernel with "nodevice ukbd".

Makes me want to figure out what the root cause was.

meena, it was for bug id 270804

RoyalYork_: posted it in the docs channel

Thanks, whats the IRC channel?

RoyalYork_: #bsddocs on EFnet

Thanks

The undocumented documentation channel

vkarlsen: it is documented, tho: wiki.freebsd.org/IRC/Channels

Title: IRC/Channels - FreeBSD Wiki

The undocumented documentation channel that is documented decently.

meena: That ruined the joke :)

vkarlsen: i'm sorry ;P

vkarlsen: I tried fixing the joke with alliteration.

debdrup :)

RoyalYork_: I've responded to it; once you get me the updated patch, it shouldn't take too long (what'll probably take longest is that I need to build the tree, and my laptop isn't that fast at that).

The documentation team does try to pay attention to reports, but it helps a lot if a patch is included when you file the bug.

Even then, stuff does fall through the cracks.

debdrup, I saw your response. Thanks. Wasn't aware of git-format-patch(1). I had looked at Chapter 3 of the Doc Project Handbook and must have misread something

I'll go through your notes and respond back to the bug

RoyalYork_: I don't know that git-format-patch is recommended, but it makes it easier. :)

got it

is there a way to only get the output of the last line of route (-n) -v get 0.0.0.0/0 ?

debdrup, does this look right? (bsd.to/gFDR) Wanted to know before I submit the patch. If not, are you able to point me to a resource on updating handbooks

I'm using default ftpd server from FreeBSD 13.2, and my client is Windows ftp.exe, everything works well, however, cosmetically I'd like to remove "501 OPTS: UTF8 not understood" output within the ftp.exe during the connection to FreeBSD's FTP Server..

My question is; is there any switch at the server side to eliminate such warning/output to clients?

we could implement UTF8 in ftpd… but that would go against the goal of FOREVER KILLING FTP

its not going to join gopher quietly

I see..

telnet died, ftp will survive!!!111!!shift1

telnet isn't dead for some purposes

MUCKs, MUDs, MUSHes, etc

you can use netcat but I've seen some server devs develop features that do use telnet client features

tercaL: that sounds like client is trying to negotiate UTF-8 support and the server says "nah man, i dont speak that gibberish"

xtile: Well, culturally, in terms of Windows.. C:\ telnet - 'telnet' is not recognized as an internal or external command, operable program or batch file. *sighs.

Totally harmless warning

ftp really does need to die tho

Haven't used Windows in many months.

rustyaxe: Thanks a lot.

Pretty sure you can still enable it via settings, the telnet command, though I'd rather ssh into a Unix system and telnet from there.

I'm excited about Haiku-OS. Will move to that *for desktop purposes* whenever it's released.

I use FreeBSD as my desktop :P

You're super cool :)

Who plays muds anymore? They were big back in the 90s, but if any are around I would be surprised if there was more than 20 concurrent players. Also, the youngest person playing would be 40

I could see some benefit with programming and learning to make a MUD, but as far as playability goes, that ship has sailed

There was a Discworld MUD I fired up a few years ago

I think that's still going

RoyalYork_: that's not made from git-format-patch

s/$/(1)/

RoyalYork_: I play MUCKs, though I wouldn't mind trying a MUD sometime.

I'm in my 20s.

and I've seen people in their teens on MUCKs and MUSHes before.

MUCKs and MUSHes are a form of virtual reality, even if text-only. It's easy to explain them to modern people as being like Second Life or VRChat (but more like SL)

I think they're a useful technology.

Not sure as for MUDs, but as for MUCKs... "503 players are connected. (Max was 779)" and "62 players are connected. (Max was 185)" for two popular ones at this moment.

I don't disagree. I grew up playing them. I just think the goldenage of text based games is long gone and the vast majority of people wouldnt tolerate playing such a game

I don't disagree with that. It's true.

Most people barely tolerate reading, in general

y read wen u can tik tok ???

:B

My own mother is growing more functionally illiterate over time as she uses her cell phone more and more.

She doesn't read labels to food she buys at the store, then gets surprised.

debdrup, I'll figure it out... sorry to bother

xtile, thats normal. When i was a kid my mother would buy RC Cola thinking it was Coca Cola. She never read the label. I would always ask her what was wrong with her and why cant she read labels. Now in my 40s I do the same thing and my kids complain. The circle of life

ah x3

You'll get there too

I promise you

I hope not.

Reading is a useful skill. uwu

I hoped not either, but such is life. Good luck

Thank you.

When I use pf and adaptive.begin/end, will the 'interval' setting also scale accordingly? So when the state count is high, will it purge states more frequently?

And is there a way to inspect per-rule timeouts and state limits?

do cronjobs get checked even when the system is in sleep, the lid is closed, ect, is there a way to make them? I am writing a simple battery protection script to gracefully powerdown.

on low battery. I am really confused my laptop went to 0% battery with the lid closed.

Are you sure your laptop is sleeping when the lid closes? I specifically ensure that mine does not because I often need to carry it and I don't want it suspending unless I specifically request it.

Even when suspended the machine's DRAM must remain actively refreshed. Suspend to RAM is a low power state, not a zero power state.

Depending upon the machine a suspend-to-ram sleep might drain a battery between 24 hours and 7 days or longer all things depending upon the machine.

Meanwhile... If the machine is suspended then no cronjobs are not checked. No operating system processes are running during that time.

DRAM refresh is a hardware feature not a software feature. Hardware runs at low power. Software is suspended.

+1

If you want to use lower power when the lid closes, shutting off the display's backlight will likely do the most good. If you like, you could make an alternative to zzz(1) that also sends a SIGSTOP to chrome and firefox and anything else you run that's particularly heavy.

thankyou

the lid closing causes the backlight to turn off

however when I returned to it, the battery was at 0% and needed plugging in to powerup

I just don't want that to ever happen again, and get to a graceful shutdown beforehand

I don't know if there is a FreeBSD equiv to the Linux sleepd which will monitor battery and automatically invoke sleep when battery reaches a threshold.

But implementing something like that seems fairly straight forward, but of course I haven't done it so don't know.

I gather freeBSD stays running when I shut the lid. The screen goes black but IRC has not disconnected

i like that, I hope that means cron is also running?

yes

cron will be running as long as the system isn't actually suspended

where would be a logical place to put my little script for cron

either /root/bin or /usr/local/libexec would be my first choices, depending

:) hopefully that works

is there a way to check the output of any echo's in the cron ran scripts?

You should get a mail with the output

as long as you have working mail and you didn't disable mail from cron

it runs every minute, can I redirect it to a temporary plaintext file?

you can use exec >>myfile 2>&1 in the script to make output go to a file

or use >>myfile 2>&1 in the command given to cron

"* * * * * battery-checker >>/path/to/log 2>&1" in cron work?

is that for /etc/crontab or for a per-user (including the root user) crontab?

(the format between the two differs; /etc/crontab has an extra "user" field before the command)

I su'd into root, then crontab -e

ok, so that's root's per-user crontab, in which case the line should be correct

but you'll need a full path to the script, most likely

(unless you put it in /root/bin and that's on root's path as controlled by login.conf

(which I think it is by default; you can see with env -i -U root | grep PATH )

sweet

I believe on some systems, you may get battery status notifications via devd as ACPI events

I am using apm -t, which gives me the remaining time in seconds

what do you mean? I need to learn more about battery power usage

and laptop acpi things...

does acpiconf show you anything about batteries?

yes

i was using awk to parse it

what does it show exactly?

what should i look for

not sure. I don't have anything with a battery to check against

Hi. I updated my zpool by adding vdev_zaps_v2 feature to it. After rebooting, my efi partition can't find a bootable zfs partition

Please tell me what to do.

uh.

what is the exact error you get? is it from loader.efi?

Yes

From bootx64.efi

yeah so boot off a thumbdrive or something and update your bootx64.efi

what version of loader.efi is it actually running?

morpho: it might be interesting to put an entry in /usr/local/etc/devd/ to log any notify messages related to ACPI or the ACPI battery subsystem specifically

The most recent loader.efi in the main branch (which is copied to the esp partition as bootx64.efi) definitely supports booting a zpool that has vdev_zaps_v2 enabled?

no idea

debdrup, I figured out how to use the git format-patch command properly. I have updated the PR

RhodiumToad: the last time anything in the stand directory was updated was before the latest merge of openzfs, so I don't think there is efi support for booting the zpool with that feature enabled. Maybe in a few hours, they will update the stand directory of the main branch to resolve the issue

I'm not sure why that would be necessary?

What do you mean?.. how can I boot my system unless there is efi support for that particular zfs feature?

there is very little zfs code in stand/, most of it is imported from contrib/

unless the loader needs to use that feature to find a kernel, i don't see why it would be necessary to update

ok. I found the specific list of features that loader allows for zfs

I am ready for this

Title: zfsimpl.c « zfs « libsa « stand - src - FreeBSD source tree

Can you explain to me why after updating the zpool by enabling vdev_zaps_v2, my efi installation can no longer find a bootable partition? This obviously means efi doesn't support this feature yet

vdev_zaps_v2 is not currently in it, so you're a bit screwed

did you get any "ZFS: unsupported feature:" errors from loader?

No. It simply says "Failed to find bootable partition"

So, it looks like I have to wait for the stand directory to be updated in the main branch. Then I'll use my other computer to compile the new loader.efi

why is my clock an hour behind?

or, you could add that feature yourself, recompile stand, and see if that helps

morpho: ntpd slacking?

What timezone are you in? Is the timezone correct? Are you running ntpd?

Oleg_: is there nothing at all output before that message?

Oleg_: please show _everything_ that was output

morpho: incorrect DST configuration? mismatch with other OS as to whether the cmos clock is set to utc or wall time?

it says BST 21:36 but it is GMT 22:36 right now

date, that is

we could come up with reasons for… at least an hour

it being exactly an hour makes the most likely explanation a conflict over the cmos clock

have you booted into any other OS recently?

a linux usb

the time is wrong

If ntpd is configured to run and to set the clock at boot then booting to FreeBSD should correct the problem. It didn't. Ergo ntpd is not configured correctly.

but the timezone is also wrong, it is not currently british summer time

do you have ntpd_enable="YES" in rc.conf? and ntpd_sync_on_start="YES" ?

yes, i use ntp

huh? of course it is BST

Wed 3 May 2023 22:39:40 BST

Is this a case of "UTC or GTFO?"

morpho: do you have ntpd_sync_on_start="YES" in rc.conf ?

Iceland does that

rwp: without ntpd_sync_on_start which is not the default, ntpd will not correct a 1-hour error

RhodiumToad, Correct.

no RhodiumToad

morpho: then add it and do service ntpd restart

I thought ntpd_sync_on_start was becoming deprecated

RhodiumToad: ibb.co/WngX0vL

according to whom?

Title: 16831500238297180895919653543287 hosted at ImgBB — ImgBB

i read that somewhere during installation

no, that's the other one

nvm

morpho, It's was, but then it has more recently been un-deprecated.

oh?

Some 20 years ago someone made a comment commit to that effect. But then about six months ago(?) there was discussion that it never happened. And peopled decided to keep it.

It would take some digging but I am sure I could dig up the timeline and email postings. Pretty sure I could find it. But it would take some digging.

same.

In the meantime... "ps aux | grep ntpd" should show -g as a command line option to ntpd.

morpho: also, does the /etc/wall_cmos_clock file exist?

meena: I have zero coding skills, so, I can't add that feature to stand myself.

i have added sync on boot

it exists, empty

Oleg_: what does the lsdev command (at the OK prompt) show?

morpho: that file tells freebsd to believe that the cmos clock is set to wallclock time rather than UTC; if some other OS sets it to a UTC time, then it'll be wrong at the next boot and ntpd will have to correct it

morpho: as long as ntpd_sync_on_start is enabled, this should be no problem other than causing a 1-hour time jump shortly after boot

RhodiumToad: it confirms that FreeBSD ZFS is on disk3p3

thankyou

Re ntpd there is this: reviews.freebsd.org/D16519

Title: ⚙ D16519 Add a note to the rc.conf(5) manual that ntpd_sync_on_start is a preferred alternative to ntpdate_enable

Oleg_: what exactly does it say

RhodiumToad: a totally standard answer: "disk3p1: EFI, disk3p2: FreeBSD swap, disk3p3: FreeBSD ZFS". I am not sure why you need this information, as we determined EFI can't boot vdev_zaps_v2 pools. I emailed the EFI developer for FreeBSD; he will update the stand directory

I'm trying to match up what you're seeing with the error messages in the code in order to confirm that my understanding is correct.

RoyalYork: just to check, you only want to be credited as with a mononym (ie. not your full name)?

debdrup, is thats acceptable, then yes. If you would prefer my last name I can reamend it

RoyalYork: Full names are preferred, feel free to PM it to me, I'll just edit the author field manually.

Its public anyways

Anthony Bravolisimo

I'll change my bugzilla profile while im at it

Blargh, I forgot t o mention the PR. >:(

Anyway, it landed as e7be269af4 and I'll add that when closing the PR.

Great. Thanks. I know it's a inconsequential change I made, but it feels good to contribute **something** to this project. My skills are limited.

is acpi part of the kernel or a seperate daemon?

do I need to enable it?

ACPI itself is part of the system bios; the kernel has an acpi module, built in (generally it can't function without it) to interface to it

the fact that acpiconf worked for you suggests that everything there is enabled