i got 2 nvme drives in external usb enclosures. if i plug 1 in and run gpart show i see it, but if i plug the other in gpart show doesn't show it. any idea how i can debug that and see why not?

it does show up in /var/log/messages when i connect/disconnect it fwiw

demido: Is the second one formatted at all? You say you see in /var/log/messages when you plug it in, does it show up a device just list the first one?

If so, you likely just need to wipe that second one's formatting out and reformat.

ya it's probably that 1 is formated ty

and hey ek you ever figure out that bhyve issue from 6 months ago?

demido: That would be my first guess. Yep.

And, to answer your question, no. :( I was never able to reproduce that same error again. I thought I was close just recently, but no dice.

So, I guess the "issue" is fixed since I cannot for the life of me figure out how I produced it in the first place.

ah dang. runner up to fixing shit is upgrade to new OS version lol. i'm on 14.1 still so when i reinstall come 14.3 i hope it has some nice improvements, especially to bhyve, zfs, and the mem mgmt system

well.. thank god for me using the rails on that isilon. i was actually able to slide it out far enough to put a new battery in.

It was eat or be eaten.

polarian: I have used Solaris and Illumos before, but apart from ZFS and being Unix-like, both have nothing in common with FreeBSD.

tinkering with cbsd and wondering if jail on local is something i was meant to set somewhere or if cbsd simply says local because the jails are local

like you don't name your local node?

Anyone know why this might be happening? dpaste.org/sVAOC - it says address is already in use but I've confirmed there is nothing listening on that port...

Never mind, my IP was assigned to lo0 for some reason... that was odd, removing it fixed it.

for a freebsd14 node where is the correct place to place SSL certificates/keys for use, eg: on a RHEL box it would be /etc/pki/tls/[certs,private] on an Debian box it would be /etc/ssl/[certs,private] what's the BSD location for these functions ?

Don't quote me on this but I think local additions would be in the /usr/local/etc/ssl/ directory.

quick question: As far as I know, bhyve doesn't support nested virtualization: this is, run for example KVM inside a Linux guest in Bhyve. I don'¡t know if I can enable this somehow, but i want to use this to run a kubernetes testing thing in a Linux VM

rwp: gives me something to research and short cut to the right query, thanks

psycorama weirdly enough echo "thecorrectpassword" |geli attach -C -j /dev/stdin /dev/ada0p3 still results in a "wrong key for ada0p3", but when I tested it with a zfs-encrypted disk (via the setup; on a vm) - it doesn't say wrong key. The UFS FDE was done from

forums.freebsd.org/threads/an-encry…titions-using-ufs-filesystems.96248  -- the example of "VladiBG"

¯\_(ツ)_/¯

yeah, weird..

weird, indeed

appjail has awesome documentation

ok. i had to turn this thing off for a while. cbsd isn't half bad. clonos wasn't that good of an interface and things were broken in it.

i'm trying appjail now but that learning curve is like pulling teeth

remiliascarlet: twas a joke :P

Guys, does 13.5 support KTLS with tls 1.3 and ChaCha20 or that's reserved for 14.x

Onepamopa: please check, issue command "sysctl kern.ipc.tls.stats.ocf"

I see tls13_chacha20_encrypts / decrypts (so far 0's, haven't yet installed nginx)

can we bring native xfs support to freebsd or there's some licensing stuff? :O

unwrapped_monad: very old topic, there was one approach in the past but it didn't become mature

there is no licensing issue, you can start right now if you want :)

ooh noice (⁠・⁠o⁠・⁠)

would be really cool to have root on xfs

why? :)

XFS is roughly 30y old nowadays, though the Linux impl was advancing a lot

though I doubt it's that interesting today

megaTherion: i find it vewy simple and has been my default since my first gentoo install ^_^

well its not optimized for flash either, spinning rust is the past ;)

I dont even have HDDs anymore

megaTherion: There's still valid uses for mechanical hard drives, particularly for applications with heavy I/O that can really burn through an SSD's write cycles

Archival is another thing worth considering, since the data in SSDs can fade if they're unpowered for a long time

DaliborFox: why would data fade away in an SSD?

dont have such use cases

megaTherion: Simply due to their mode of operation, the data is stored as a charge stuffed into a place on a special transistor that is hard for it to escape, but it does eventually escape. The controller on an SSD periodically refreshes old data so that it remains valid, but if you leave an SSD drive unplugged for a year or two, you can expect some data loss

megaTherion: well its not optimized for flash either | it feels kinda fast tho (⁠´⁠-⁠﹏⁠-⁠`⁠；⁠)

DaliborFox: well there is bitrot in HDDs.. do you have any technical papers for evidence?

megaTherion: The manufacturers themselves warn of it, e.g. here's IBM telling customers to not keep their drives unpowered for more than 3 months: ibm.com/support/pages/potential-ssd-data-loss-after-extended-shutdown

unwrapped_monad: feel free to start porting XFS :) maybe one can get the original r/o impl running again

megaTherion: aye, i'll give it a shot soon after my gsoc ^⁠_⁠^

gsoc?

google summer of code :)

i'll be working on building a wifi utility like iwctl for freebsd ^_^

oh

SSDs and all NAND Flash storage such as SD cards are not rated for long term un-powered storage. Those require periodic reading and writing of the data to keep it refreshed.

Half life depends upon the implementation but probably 3 years I would guess.

but as long as they are powered they are fine?

The first symptom being that old laptops slow down and boot slowly. The SSD firmware is using more and more internal ECC and retries to read the data.

Powered up depends upon the use and the firmware because lots of cheap crap will never rewrite cells that are already written. Those will still slow down even if powered. Those will still require a full rewrite at some point and then they will go fast again.

btw is it possible to recover data from a failed ssd? (⁠・⁠o⁠・⁠;)

Depends upon how it is failed. And there are a lot of possible ways for something to fail.

unwrapped_monad: sure, depends on the failure... there are people replacing failed controllers etc.

ooh

like ssds that reach end of life

I have had many cheap SSDs (OCZ, I am looking at you) but I have never been able to wear out a good quality Intel or Samsung SSD due to wear. I am sure it can happen. But the failures that have gotten to me have been the controller dying or other on cheaper Kingston and similar SSDs where the controller is no longer talking to the SATA.

all SSDs I have are running and in use (I dont archive on SSDs), the only SSDs which failed to me were early OCZ ones :D

I have a failed stack of those really cheap OCZ that failed from a school that bought a hundred of them because they were cheap and within the first year about 10% of the OCZ drives had failed.

(they were not cheap, back then OCZ was doing enterprise stuff)

In that particular school environment all of the machines were effectively medium weight thin clients, no local data, we had a PXE boot and install process where in about 10 minutes we could reset any client workstation. But for most of us here it will either be servers or personal systems with important local data and configuration. Everyone please use RAID and the Backup Rule of 3-2-1 to protect it.

sure SSD without Backup would be pretty stupid... but same with HDDs

however you can get really good monitoring out of SMART

I don't see spinning drives as villains like many people do today. They are well understood and work well. But for a mobile laptop I only ever use SSDs. SSDs and spinning are both good at different things and bad at different things.

rwp: power consumption is a big issue

power consumption and shock durability.

Have you seen how much power a data center consumes and where it is consumed? The difference between spinning drives and SSDs is not that large of a difference in the total number.

the difference is HUGE, believe me :)

Also SMART has not lived up to the promise of being a predictor of future failure. It is good for diagnosing actual failure though and that is useful.

Data center power is one of the reasons those are most often located next to hydro electric where electricity is least expensive.

SMART is a totally fine standard, which is working and being fine for years

spinning drives in constant use versus ssds in constant use -- ssds end up using much less power.

hydro is also good for open-flow water cooling. 2-in-1

wavefunction: hence I migrated for my small home office setup... but I live in germany and we've to pay *a lot* for electricity

megaTherion: Made me sad to see y'all weren't following up with nuclear.

megaTherion: What's your rate-per-kwhr?

I'm at $0.10263 USD/kwhr

I think its 0.36 EUR / kWh

Wow.

some pay less, others pay more

and this is not all we pay sadly

I'm sure there are base-rate charges, etc.

ya I'd like to go nuclear but we're ruled by complete morons

ya we have some CO2 taxes and other crap we have to pay on top of that

bruh the installer TUI went unresponsive at the select a country step :')

unwrapped_monad: you're on current?

nyo im on release 14.2

eh try current, you're gonna be developing the system

ooh ohkie

maybe i messed up with the manual partitioning (⁠╥⁠﹏⁠╥⁠)

you chose zfs right? boot environments are gonna be really useful for you

are they like btrfs snapshots? (⁠・⁠o⁠・⁠)

any chance of getting ufs encryption via the installer? ;)

Sure. Choose GELI encryption. I am pretty sure it Just Works. But I always choose ZFS instead of UFS these days.

unwrapped_monad, You should read this: wiki.freebsd.org/MasonLoringBliss/ZFSandGELIbyHAND

rwp: thx <3, i'll go through

anyone recommend a strategy to kee

woops

Does anyone recommend a strategy to keep changes to base src tree across multiple git branches?

I messed with the Makefiles and wa t to

Wow, something is wrong with this keyboard...

I messed with the Makefiles and want to preserve them, even against base, but don't want to commit them (ideally)

farhan: stash the changes, or better - commit to main branch and cherry-pick to other branches

Definitely commit your changes. Commit them to a local branch. Then you can fetch the upstream branch routinely. Rebase your commits against the upstream.

There are lots of articles written describing how to work with upstreams.