My kgdb over bhyve is not catching panics anymore. I swear the same setup worked last year. Does this work in 16-current or 15-whatever for you folks? I have more words here: forums.freebsd.org/threads/kgdb-ove…ch-panic-anymore-anybody-else.99950

Ltning: openzfs/zfs #17898

cool, that makes (some) sense. :)

I just copied ~40 logical TB from one place to another and mv'd back to get around this.

Had to take a node out of the Galera cluster and stop it .. which kinda sucks. Almost as much as my original decision to try 1MB. :D

curious about what issue you encountere with 1MB recordsize?

Just much more write amplification than I expected

I thought having a healthy SLOG would offset that problem, but the writes we're seeing are spread out so much the chance of several being within one 1mb record are minimal.

ah i see. thanks

Hrm, has load/cpu calculation - as seen in top - changed between 14.1 and 14.3?

Hi!

etienne`: Hello

entrop: sup

Sorry, wrong ping. Meant for etienne`.

I decided to update my 2.0 installation to 2.0.5, and well, it blew it

the new bootloader now vaguely complains about the system partition (probably because it extends past the cylinder 1024) and REFUSES to load the kernel

and I cannot use the remapping in bios because the wd driver REFUSES to work with >16 heads

that is some S for stability

who ses cylinders and bios's these days

ses = uses

freebsd 2.x definitely dows

I can work it around by using 2 disklabels, with a / partition under 1024 cylinders but that�s stupid

is there a way to make the kernel use a root partition other than which it was loaded from?

sjjd, with the currenbt loaders, I think there is. With 2.0.5's loader? No idea. If it even had one.

there is a loader which asks for a kernel path and parameters

iḿ going to try to massage the mbr

well it didn�t help

but it�s ironic it worked in 94 and they broke it in 95

in this guide freebsdfoundation.org/wp-content/up…oads/2023/11/reushling_poudrier.pdf ccache setup would not work as there is no parent dataset /var/cache. In default ZFS layout there is already zroot/var dataset with /var mountpoint. And there is already cache dir inside with some content. There is no zfs way of creating zroot/var/cache/ccache by following the guide. Was it a mistake or a some custom layout was used?

i would like to use zfs dataset for ccache

Is anybody running samba as a DC on ZFS?

oh loooook reviews.freebsd.org/D53597

:p

:eyeballs: Neat

What needs to run 24 times?

hopefully nothing

i took that number from kevans

I guess no harm done with choosing 25

it's a hard limit

it runs as long as files appear + 1 more

or i mean then it won't

anyway

now i need to port it over to periodic.conf where i actually needed it

yeah, just chosen arbitrarily

25 rounds to finish resolving rc.conf is a lot and probably a sign of someone having goofed

odds are it usually resolves in under 4

i only need 2

:)

that was missing in periodic

i tried 5 levels

and those are levels eh

who added vendor.conf i should look

and how do i make

git diff -U1000000 rc.conf | arc diff --raw

better

i can't find how to add local-uncommitted diff

there are also untracked files elsewhere in repo

i have left repos untouched and just have changes in them

which i have saved outside of repo and outside of git so i have backup and can forward/reverse patch :p

git: cd4f1705af83 - stable/15 - release: Make fetch happen in GCE images