HN Explorer

13 comments

[ 4.4 ms ] story [ 39.5 ms ] thread
tiffanyh
Off topic question: what's the current process to make extremely small "minimal" freebsd server install? Is it NanoBSD[1], because the unfortunate downside is that the base can't change (I realize that's a feature) and for a web server or database server that might not be ideal.

[1] https://docs.freebsd.org/en/articles/nanobsd/

drewg123
If you're building from source, then I'd just disable what you don't need via src.conf(5)'s various WITHOUT_FOO settings. They are described in the man page itself.
sebcat
I did this a while ago. the .iso.xz came in at 22.1 MB and the .iso at ~80MB.

That's for a ro image with tmpmfs and varmfs[0], a minimal kernel, base and the packages nginx-lite, jansson, re2, curl, acme.sh and some services.

The system boots from an ISO into running nginx and hosting a basic network scanner[1] at a VPS provider.

I basically built from source with a custom SRCCONF[2], KERNCONF[3] and some other cleanup[4].

There may be easier ways to do it but it worked out pretty well for me.

[0] https://www.freebsd.org/cgi/man.cgi?rc.conf(5) [1] https://disco1.sajber.se/ [2] https://github.com/sebcat/yans/blob/master/tools/freebsd/vm/... [3] https://github.com/sebcat/yans/blob/master/tools/freebsd/vm/... [4] https://github.com/sebcat/yans/blob/master/tools/freebsd/vm/...

raj2569
> The topic of load average is generally very broad and it is calculated differently on various UNIX systems, so a separate article may be dedicated for just that.

I have never seen an article explaining how the load average in Linux is calculated.

5e92cb50239222b
Tangentially related: modern kernels have a much better metric (PSI, or pressure stall information)

https://lwn.net/Articles/759781/

https://lwn.net/Articles/775971/

da_chicken
Well, that's still an aggregated value that someone decided to calculate based on what they thought would work best for what they're looking for. It's not necessarily more a correct definition of load, just one that is more appropriately actionable for the problems it's designed for.

In other words, PSI is "better" for some definition of "better".