how can I quickly transfer a dataset and its snapshots to another zpool ?

you can zfs-send one snapshot at a time

ok, Inhaven

I have hundreds

time to script it

syncoid would paper over zfs send | zfs recv to make it pretty easy. part of sanoid pkg

last1: for one single dataset to get all snapshots sync the first then use -I, for a dataset and all its children -R is an option but not exactly the same as sending individual snapshot

what is the error "send msg: No such file or directory" saying when it comes back from an ngctl mkpeer? I can't find anything about this error online or in manpages

jmnt: so what I would I do exactly

last1: read man zfs-send and man zfs-recv

hello

howdy. no other chatter, so I thought I would ask if anyone else found that VNET jails start up noticeably slower than non-VNET jails. for me, definitely. watching the system boot up is annoying due to the pauses when VNET jails are starting

I should have said "if you or anyone else has..."

I haven't measured the time it takes vnet jails to start but they do have to set up the network stack first. And then the jail booting up runs through the network startup scripts.

One of the advantages for me of FreeBSD is that it needs to be rebooted less often than systemd/Linux systems which these days have security patches every few days it seems.

i am passing a directory into a jail via nullfs and rw on fstab, but it seems that root on the jail is able to bypass any ownership/rwx permissions set on the folder? Is there a better way to pass a folder into the jail such that root in the jail is still limited in what it could do with the folder?

I am actually a little curious why this works this way.. any user uid and gid on jail that match that of the host systems has the same permissions on the folder set from outside the jail

yes, but they cannot escape the jail

also a reason to drop privileges inside the jail

rtprio: yea I'm aware of that.. but it seems if an intruder gains root on the jail they can definitely do anything root can do to any nullfs folders.

I'm wondering if theres a better way to do it.. even though nullfs is pretty convenient

What files and/or directories are we talking about here? I am left to guess and there are many guessing possibilities.

rwp: so for my use case its a local webserver with access to media files.. given what I mentioned above I mitigated worst case scenario of: jail gets hacked, root on jail deletes all the media filess by having my files snapshotted/backed up but yea it wouldn't be nice if they do get deleted

so yea I'm only passing in via nullfs whatever that jail absolutely needs rw access to.. if not it would be ro.

I assume you have already thought of putting the media files on a dataset read-only in the jail and that doesn't work because you need to add files there from the jail too?

I could give the jail access to the files via some other protocol like smb or something.. but its just a bit weird to add protocol overhead / access latencies to the files when they are essentially right on the same host :)

would be nice to have an option in jails to not allow root to have full access to passed in nullfs dirs. not sure how easy or possible that would be :)

rwp: yea the webserver in question needs to be able to add media files / modify the existing ones (metadata for example)

and so yea you can argue, if anyone gets access to the user the webserver is running as they can delete the files without even needing root and thats true.. but they would also be able to delete / modify some other dirs from nullfs that are rw and meant for another process on the same jail. its a bit specific and probably won't happen often but yea.

rwp: I still remember when the Linux community was laughing at Windows for having to reboot it all the time. Thanks to systemd, the BSD community is now laughing at Linux for having to reboot it all the time.

slackware still runs init

even has run levels!

lw,

Only Linux kernel updates require a reboot, and you can easily just put it off if you don't want to bother yet. It's not forced aggressively like on Windows.

Ah, I see the Linux defender has logged on.

mark your calendards

JustBleedFan, Actually if any of the systemd components are upgraded those also require a reboot. And udev is one of those components. And systemd updates are frequent right along with Linux kernel updates.

Imagine people realizing that systemd is an OS in itself

Also dbus too. Forgot to slip that in there.

systemd is a message passing operating system.

rwp: we don't need to litigate Linuxisms in here.

Relitigate, I should say. They made their bed.

My first comment was that it was an advantage for FreeBSD. I was extolling the virtues of FreeBSD. But I will stop.

Didn't read like it to me, since it mentions nothing about FreeBSD, but I'll let sleeping dogs lie.

don't really see why linux anything is on-topic in here

last I saw, this was #freebsd, not #linuxsucks

If one is directly contrasting things with specific examples of what FreeBSD does better, that's basically the best case scenario, I think?

I said, 10:46 <rwp> One of the advantages for me of FreeBSD is that it needs to be rebooted less often than systemd/Linux systems which these days have security patches every few days it seems.

comes off like a fanboy on ubuntuforums

So (me checks topic) "Have fun, be civilized, respect each other". We are scraping very close to failing here.

rwp: Still looks like it's #freebsd to me.

Pretty sure "respect each other" also encompasses respecting the topic of the channel.

In that case everyone should help out here and change the discussion to something interesting about FreeBSD. Just do it.

Anyone have any recent experience with VM hosting where FreeBSD is a first-class option?

timothias: Because Slackware has always been that distro that everytime something "modern hip shiny thing" comes up, it goes like "meh, fuck it".

CRUX is the same way, and it's even more BSD-like than Slackware.

remiliascarlet: I always liked slackware. Did what I wanted it do do with not much fuss.

It's kind of like a pattern. Came out before the 2000s, so it must be good.

win 98 came out before 2000

and it was balls

Microsoft has its own weird pattern.

I know this is freebsd channel, but anyone get any crowdstrike love?

Going from good to bad to good to bad to good to better to OK-ish to best to worst to a bit better than the worst to privacy nightmare to the privacy fucking hell.

And no, I didn't.

I bought a mac to get away from win 11

and thats saying something cause im not really a mac person

I did see where they were saying convert bitlocker keys to barcodes so entry would be quicker

Oh yeah, don't forget to scan your QR codes like a total slave!

barcodes...

Well, you can do something cool with barcodes too: invidious.jing.rocks/watch?v=bOfpQt4KFCc

sorry for off topic

hi, did anyone compile NCSA-Mosaic-2.7 in FreeBSD x86_64 ?

anyone have a digitalocean droplet with freebsd on it? there's always been this snapshot zroot/ROOT/default@base_installation and @digitalocean_installation ... and they use 192M and 752M, respectively, but now refer to 2.67G and 3.28G, respectively, and I feel like i don't need 4yr old snapshots anymore.. but they almost have sentimental value... should i destroy them?

i destroy most snapshots of zroot/ROOT/default, because i don't think i need to rollback to 12 or 13-RELEASE. but i sort of like having the "infinite" (12.1 maybe?) rollback feature for a VPS where this vps provider doesn't even support freebsd anymore

in a related matter, i deleted over 6G of files in /var/db/freebsd-update/files, around 3G in /var/spool/clientmqueue and around 1G in /var/cache/pkg, and it did not free up any space as far as ZFS was concerned, lol

i think it's safe to destroy them

and it didn't free it up because they're probably held up in snapshots, yes

scoobybejesus: zfs list -o name,used,usedbysnapshots