sort -o file_3 file_1 file_2

there a way to cat >> ~/path/file but only if the line isn't already in the file? getting duplicate lines injected

Would be messy+expensive. What is the source of input that is being appended?

You could run "uniq file >| x && mv x file" post-append

Or, open the file in read+append mode; keep the memory last N lines; write the input only if not already present in those remembered lines.

grep -Fx <line> <file> || echo <line> >> <file>

parv: gist.github.com/kenrap/560386e9e53389751c640a8d413442a0

Title: gist:560386e9e53389751c640a8d413442a0 · GitHub

this has a link to the RC3 ISO images that results in a 404: freebsd.org/news/newsflash/#2023-10-27:1

Title: FreeBSD News Flash | The FreeBSD Project

namely the link behind the word "available" which points to: lists.freebsd.org/archives/freebsd-current/2023-October/001595.html

maybe someone wants to fix this. ;-)

ziv, fix yer shit.

babz, Using grep -q there is good. Means grep will stop reading once it encounters the line. Maybe faster if the file is large. Also it emits no output.

o_O

Makes sense though, since there's no output.

Since the only purpose of grep -q is to act as a boolean with the return code after the exit status becomes known then it can short-circuit doing more work in that case.

Before -q was added we had to simply redirect the output >/dev/null but otherwise it is the same concept.

I believe grep -l does the same optimization too.

good to know

because sharing is caring, i just implemented this setup on my raspberry pis and it has had a nice increase in "performance" edersoncorbari.github.io/tutorials/freebsd-performance-tuning

Title: FreeBSD Performance Tunning Ederson Corbari

dude. a lot of that stuff has been sniffed from teh #deeppast. but nonetheless thanks for bringinng it all together

mention of hardware, purpose etc... are in non-relation

you talking 1bhps on a web server or 3 logins on a shell server ?

the distribution ... are you talking tuning for inter nodes/proxies, whats your traffic shaping look like ? etc etc...

does anyone run openbsd as a guest in bhyve?

with *UEFI* ?

i can't seem to get it to boot

well i want to clarify, i did not write this article, i found it and wanted to share in my ever enduring journey of trying ot increase performance on this raspberry pi

next up.. kernel compiling.. fun!

that won't improve performance, but might free up some memory for you

There's a mail on the announce list about the FreeBSD vendor summit.

voy4g3r2: that article contains at least some amount of cargo-culting, because loading device drivers that aren't needed for booting is slower than loading them via rc.conf(5) via kld_list and/or devmatch.

Also, note that HTCP isn't exclusively "best", and there's a reason why FreeBSD has a pluggable TCP stack to switch congestion control mechanisms.

debdrup: understandable, just figured would share.. as the congestion stuff was fascinating.. your mileage may vary of course.

debdrup: the "speed" of raspberry pi images just "feels slow" in comparision when i load the raspbian distro.. but it is probably expected as they have people who explicitly work on it

voy4g3r2: one problem I had was that the mmc stack for some reason wouldn't raise the sdhci speed making disk access operations ungodly slow

i started rolling with GENERIC-MMCCAM because of it

kevans: on a raspberry pi?

armin: i got 2 of them, both 4b w /4 gig.. great little play machines

this could be a bad rabbit hole: ftp3.au.freebsd.org/pub/ldp/HOWTO/NFS-HOWTO/performance.html

Title: Optimizing NFS Performance

on tcp, what's the difference between function stacks and congestion control modules, and why does bbr have both?