i set up headscale in a VPS jail. all working great. my freebsd box at home runs tailscale (on the host), along with my phone and another couple computers. they all connect. i have the freebsd box at home as an exit node. can i have pf take packets on tailscale0 interface forward http/s packet to my caddy (thin/loopback) jail on the same box? i put an rdr rule, and it doesn't seem to work.

hello. after updating to 13.4 in the userland update step ran out of space in /usr. so far googled 2 options: 1) "freebsd-update IDS" and compare checksums or 2) "freebsd-update rollback"

ids shows quite a few files in lib & lib32 with different checksum

will the rollback send system back to pre-13.4 or just before userland update?

or better try to fetch base.txz and try to fix relevant files?

or is there third option how to proceed

\x0304red\x03 yo

has anyone had good/bad experience with having 1 central /usr/ports tree and using nullfs to "share" the direcotry with multiple jails?

i am building a few systems out of /usr/ports and i am making multiple copies of the ports.git and it just seems to be a waste of space and a burdne on management.. where something like a nullfs to the server to each jail would be "easier"

space is not a constraint, it is more my sanity of management

voy4g3r2: cant you mount the /usr/ports dir in the jail ?

mount $host:/user/ports into each jail

yes, that is what i was thinking.. i can setup a fstab in my jails to mount the /usr/ports from host

i was just wondering, if ther eis something i may be missing

just setup the fstab and mount it

if i build one package in one jail and do another in another jail. there is an impact.. or it is isolated enough to not be a concern

i may just end up making a poudriere and have it just build my packages for each jail

voy4g3r2: whatever

but if you are unsure, try building a simple port and see if anything remains inside /usr/ports

doing it now..

i was just thinking while i was waiting for the tree to build

voy4g3r2: you are always a step ahead

i have been known to do that and was like "what happens if i just use one central /usr/ports instead of having now 6.. one for each jail"

just seemed real stupid

if jails are similar versions, and ports are built with similar parameters .. then yes... nonsense

and if you are just building with defaults.. also nonsense.. just use pkg

or build the poudriere repo

which i take it is where poudriere ports -c -p default comes in

becasue the jellyfin, zoneminder, grafana packages is setup with options, specific to my build

and the whole pkg vs ports builds.. mix-up blah

but thanks for the confirmation.. too many /usr/ports is a bad thing

+1 poudriere

seems poudriere and git are the things I need to be learning about. All the cool kids are using them

SponiX: what source control do you know?

rtprio: none, I've avoided learning them over the years lol

okay, that is one approach. depending on your line of work.

all I really know about git is clone and pull. and I also know next to nothing about github

What's going on when, from time to time, I have to restart X because any attempt to launch an application returns "Invalid MIT-MAGIC-COOKIE-1 key" ?

sounds like one of the tmp files in /tmp/.X11-unix/ was deleted

or i would start by checking there

Ltning: ^

Nope, they're there

and xauth extract - $DISPLAY gives the same value as exists in .Xauthority and /var/run/xdg/<user>/lyxauth (which is the global XAUTHORITY path)