eyJhb changed the topic of #nixos-on-your-router to: NixOS on your Router || https://logs.nix.samueldr.com/nixos-on-your-router
<gchristensen> machine # netboot.ipxe : 265 bytes [script]
<gchristensen> machine # bzImage... ok
<gchristensen> machine # initrd... ok
<gchristensen> machine # manifest... ok
<gchristensen> machine # nix-store... ok
<gchristensen> machine # initrd... ok
<hexa-> very ok
<gchristensen> 35M ./result/manifest
<gchristensen> 4.4M ./result/bzImage
<gchristensen> 9.3M ./result/initrd
<gchristensen> 1.0K ./result/netboot.ipxe
<gchristensen> 330M ./result/nix-store
<gchristensen> manifest is much bigger than needed but will be easy to shrink
<gchristensen> I can test a change to this system in qemu in about 15s as opposed to about 10 the old way
<gchristensen> 10m*
<gchristensen> the only problem is it can't shut down :(
<clever> gchristensen: you must have seen my netboot_server.nix ?
<gchristensen> yeao
<gchristensen> :)
<gchristensen> ok bed time. g'night
<clever> nn
Shados has quit [Quit: Shados]
Shados has joined #nixos-on-your-router
lc has quit [Ping timeout: 260 seconds]
locallycompact has joined #nixos-on-your-router
teto has joined #nixos-on-your-router
<gchristensen> (it works now, and has a test!)
<hexa-> sweet
asymmetric has quit [Ping timeout: 260 seconds]
asymmetric has joined #nixos-on-your-router
<mdlayher> question: the existing Prometheus WireGuard exporter packaged in NixOS does some weird and unidiomatic things with metrics output, and also relies on shelling out to the wg(8) CLI tool to do its work. i have authored a more idiomatic one that uses the native interfaces for all implementations via the official Go library (which I authored).
<mdlayher> at the moment i have packaged my own version of the exporter locally for my use, but would it be appropriate to propose adding it upstream in some way?
<mdlayher> since there is already services.prometheus.exporters.wireguard though, i'm not sure what my options are.
<gchristensen> how incompatible is the output?
<mdlayher> very. the existing one exposes lots of labels on each metric (resulting in high cardinality) and mine goes more along with the usual best practices by requiring a bit of PromQL to piece things together as needed and cause less time series churn
<gchristensen> I think it is okay to just replace it and have release notes saying the output changed, and maybe explain how to derive the current metrics from the new metrics (and the thinsg you can't do anymore)
<mdlayher> gotcha! i will do a bit of prep work and hopefully have something posted in a day or two. I'll ping you on the PR
asymmetric has quit [Quit: Peace.]
asymmetric has joined #nixos-on-your-router
asymmetric has quit [Changing host]
asymmetric has joined #nixos-on-your-router
asymmetric has quit [Client Quit]
asymmetric has joined #nixos-on-your-router
sshow has joined #nixos-on-your-router
genesis has quit [Ping timeout: 272 seconds]
genesis has joined #nixos-on-your-router
fooker has quit [Ping timeout: 246 seconds]
fooker has joined #nixos-on-your-router
fooker has quit [Ping timeout: 260 seconds]
fooker has joined #nixos-on-your-router
teto has quit [Quit: WeeChat 2.8]
genesis has quit [Ping timeout: 272 seconds]
genesis has joined #nixos-on-your-router
genesis has quit [Ping timeout: 272 seconds]
genesis has joined #nixos-on-your-router