with Boot Environments being cloned during a freebsd-update, am I right I should want the last created clone to be Active? Because right now my Active (default) is from mid-2020 whlie the newest is of today

Using zfs destroy -nrv on each clone shows there is nothing in them, even though there is space alocated, up to 962M

weust, you realize -n is a no-op though right ?

CmdLnKid: yes, I was following this: dan.langille.org/2021/04/15/deleting-old-freebsd-boot-environments which mentions a dry-run using -n.

Title: Deleting old FreeBSD boot environments – Dan Langille's Other Diary

weust: I'm pretty sure it makes changes to the current dataset/BE and creates those snapshots as a means to get back before the changes

having to call bectl activate after every call to freebsd-update would be annoying

still on root with ufs here. commonly writable areas for me are mounted via a dataset from a seperate disk

no bectl here. no need for it tho it would be nice for critical servers

i find ufs easier to recover

tho i can't think of one time in the last 10 years ive had to do so

OK, but is there some kind of rotation where it cleans up after x amount or something? Because only adding and taking up space doesn't sound too good either.

I feel a guide on this and how to deal with it is missing, at least I can't find it. Only on how to remove them.

that space can free when its needed. if you need confirmation of space removal "reboot" itll come up clean after

a couple sync calls or a zpool scrup may do it as well

if it doesn't then i would say there still has to be something there somewhere

weust: do you have old snapshots taking up space?

Let me paste that on pastebin. one sec

inspect zfs list -t snap

Title: dpaste/BZLg (Plain Text)

Title: dpaste/iWs1 (Plain Text)

same output, kinda

nvm de icinga setup of course

so you do have snapshots sitting around

that would be your goal

just remove em and youll see a diff results

they look like automated snapshots

They are automated. Created during a freebsd-update

962M was approximately what you were seeing un'freed

So that is wasted space right now, or not?

the BE was claiming zero because it was removing exactly what it was in control of

not if you zfs destroy zroot/ROOT/default@2022-10-29-17:42:28-0

etc... etc...

but without using -R it will shows that it cannot destroy the snapshot: snapshot has dependent clones

is it listed in bectl list -a ?

if it is destroy it there

weust: its only "wasted" if you never intend on going back

@channel isn't there a bectl maintenance facility in ports ? maybe included ?

In my first pastebin link I would say so? zfs list -t snapshot -o name,clones -r zroot/ROOT shows this too

bectl destroy can remove the environments and the snapshot

Title: dpaste/zdt6 (Plain Text)

ah ok

yeah

So, to destroy the oldest I would do: bectl destroy -o 13.1-RELEASE-p2_2023-03-16_201934

personally i prefer creating snapshots in unix time and calculating ... "is older than" and automatically removing them

if a systems been running for 14 days then i don't need any older snapshots

blame that on the one that build that into freebsd-update ;-)

i'm not even sure if you need the -o but sure

the -o comes from the website I mentioned earlier

weust, pretty sure that behavior cam from opensolaris bectl

we followed

you mean beadm

aah

yes

jwmaag: dan.langille.org/2021/04/15/deleting-old-freebsd-boot-environments shows bectl destroy -o. that's all I've got

Title: Deleting old FreeBSD boot environments – Dan Langille's Other Diary

But, I know my server boots fine, so I can destroy all except default, right?

right

or perhaps leave the last one from today

you can do that too

unless you are dying for space ...

that in itself is really uncommon these days

Home server. zroot has 22G in use of 185G. space it not an issue.

Like I said, I am missing som documentation on how to handle this. It's cool that it's done, but only taking up space isn't imo.

assuming freeb

assuming freebsd-update doesn't handle it itself, which I doubt it does atm.

it doesn't and it shouldn't

no thats not up to that utility

in a bsd way of thinking that would be hazordous

unless you feel like creating something that creates a record of previous recorded snapshots

but specifically thats for backup tools

OK. Then I am missing a part in the documentation that mentions what to do after an upgrade. maintenance thing. at least to check it out, etc.

that might be a worthwhile contribution to the documentation

anyway, I will write it down in my own notes

when you ... freebsd-update --- keep in mind this is what happens and you may want to remove them

awareness would be the better term, perhaps. know it's being done, know space is being used.

yes

now it just tells you to do a install after a upgrade/fetch, reboot and run install again. that's it.

haven't read the man pages in depth so those should definately be considered

see also sections may be needed

or examples

yeah. right now man freebsd-update doesn't mention it at all.

pin the contributors with committers lines in git/svn [blame]

I did read in forums.freebsd.org/threads/zfs-how-…snapshots-and-not-damage-data.85436 that you can set an option in freebsd-update.conf. CreateBootEnv no" to precent automatic creation of boot environments.

Title: ZFS: How to properly remove unnecessary snapshots and not damage data? | The FreeBSD Forums

"with suggestions"

will do

ooooh that thread

sheeesh

hehe

no dougbt you are using zfs on root FFS

best laugh ive had today

well, I did mention that several times? my pastebins showed it.

lol yeah assumptions... seem to be a problem there

gotta verify

Only my OPNsense on my thin client is running UFS atm. The new OPNsense server is running ZFS as well. should swap them this weekend.

rather than pfsense

Been using that since very early 2015

yeah, was fed up with the forums there. very toxic because of a certain individual. went to OPNsense it was nice and relaxed. happy ever since.

yeah used to be involved with a pro php dev, he at one point pointed out all the problems with pfsense and said he'd rather be running opnsense

that was my changing point

i use both now as pfsense being my backup firewall in the most basic sense if there is a problem not fixed as fast as it probably should be ill fall back on pfsense until its fixed

eggs ==> baskets and all that BS

both work fine for what they do. but knowing the history on why opnsense was made and the crap they got, which was insane, I stay away from pfsense for that reason.

yep

just shows what a corp influence can have on a project

yeah, but also especially that one person. and some fanboy on twitter at the time. that guy was nuts.

his bio mentioned he was there to disprove anything opnsense did. anything.

ignorance is bliss

yup

need to head out. thanks for the help once again, everyone.

have a good one man

are jails only for daemons or can we put a single command in a jail that we want to run more locked down?

yes

?

you can do both

ty!

they sometimes get called service jails if they're just for a single command

typically you'd want to build something statically, so that there's only a single file in the jails filesystem that needs to be run without any sort of dynamic libraries

why call it a service jail when it's not persistent like a daemon is? isn't daemon synonymous with service on other OS's?

it's as persistent as the command (perhaps a daemon) it's running

fair amount of stuff doesn't work well with static linking

kevans: who can I interest in my automount bug, which I would argue has at least mild security implications (but not enough that I made it private)

#272446 / D40961

(the security impact is that it arbitrarily turns off important flags like "readonly", "nosuid", "noexec")

hmm

trasz would also have been a good candidate, but I think he's been largely not around as of late-ish

yeah, almost all commits were by trasz, one by rew

polyex: no idea of the etymology, just giving something to use as a keyword

kevans: yeah the last time i heard from trasz was when he released me from my mentorship

Linux uses /proc/<pid>/environ to get the a process Environment. How can we do that on FreeBSD?

procstat has an option for it, but iirc it relies on the process not having clobbered it

Mem: 2502M Active, 53M Inact, 1491M Wired, 955M Buf, 277G Free

what a satisfying upgrade

RhodiumToad: cool, thank you

meena: i even introduced "penv", "pargs" and "pwdx" shortcuts

all in /usr/bin

otis: how long have we had --libxo in procstat?

meena: since 2015

cool

libxo is so cool

to run a command in a service jail we use jexec right?

you could also have it defined in jails.conf

the base jails that reuse userland, are they reusing the host userland or a jail's userland?