rtyler: something like 15-20 hours, iirc

er, sorry

rtprio: ^

(and still nothing available during that time; there's a symlink that's flipped once a repo set syncs to all mirrors to quickly update the published set, atomically)

@cfb ill

oops

kevans_: thanks

hm, what component would be a linuxemu PR? I want to report that problem with the elf interpreter being a symlink I helped to debug some days ago

Worth asking in #freebsd-bugs if no one has an answer here, maybe.

nimaje: is it with the kernel side, or linux userland construction?

did freebsd develop something like zram?

kevans_: pr #289739 path translation doesn't get applied a second time if the ELF interpreter is an absolute symlink like it is the case with userlands produced by debootstrap

nimaje: is this a regression or has it always been this week?

s/week/way/

there's a change that first appeared in 14.0 that makes me wonder

kevans_: no idea, I only tried that because someone here had troubles some days ago and I currently don't have any system that isn't 14.3

Is there a way to figure out why a ZFS dataset is busy?

souji: perhaps with lsof, that applies to every filesystem

or you mean you can't do operations like unmount it?

(If you see slow i/o and lot of activities, you can do `zpool status` to ensure it's not doing a scrubbing)

`iostat` also gives you the general disk activity

Dereckson: I want to destroy a dataset. There isn't much load on the system right now, so I don't think thats the issue. It probably is some file that is still open or .... I'll try lsof for now. Thanks.

Interesting; lsof does not show any open files in that directory. To be more precise, this happens when I want to destroy a thin jail created with bastille. I had this kind of issue for some time now that somtines the datasets connected to the jail cannot be deleted or destroyed. When I have some time I'll look into it a bit more.

If you don't care about the opened file content you can try `zfs unmount -f`

bastille don't use zfs hold by the way?

(zfs hold allows to protect a dataset from deletion)

Found a way to check that easily: zfs list -t snapshot | awk 'NR>1 {print $1}' | xargs zfs holds

zfs holds <all your snapshots>

Yeah, already checked if they are hold or so. Since I wnated to destroy it anyway, zfs umount -f did the trick.

Not sure why this is happening now more than some month ago... But thank you for helping :)

hm, there are enough ports packages for stable/15 now that i can switch my test system to use pkg.f.o, right?

ivy: i would think so

i'm twiddling my thumbs waiting... patiently... for this amd64/main build to finish

at the current rate it's looking like 3-4 days

(for finished + sync, and assuming the last 2100 ports don't have any surprises)

the browsers are already finished and i think it's past all of the gcc/llvm flavors, so fingers crossed

did python get fixed? there might be a lot missing without that

we have python311 and a number of py311- ports, at least

net-snmp failed which caused 68 others to be skipped, and that's the highest-impact failure in this run

hmm. have noticed that FreeBSD seems to consistently be snappier than Linux, at least at the commandline.

have never been sure why that is

freebsd is probably cheating somehow to give that appearance to increase its appeal :-)

Could be /bin/sh. It's not as featureful as bash but lacking those features does mean less parsing and therefore it can be a tad bit faster.

you know, that's possibly true and very easy to test

i've noticed taht zsh is quite sluggish, especially when its churning though oh-my-zsh and all those neat little plugins

its a cool shell, but you can't load it down, it will get nearly unusable