@wolf480pl better lb than nginx (i am not talking about nginx pro). Works pretty great with containers and orchestrators.
@quad@weeaboo.space @wolf480pl@mstdn.io pretty graphs is pretty good tbh. nginx doesn't have any proper metrics, because it's gated on the closed source version, which ain't great.
Personally I dump the logs through goaccess if I need an overview
@quad@weeaboo.space @wolf480pl@mstdn.io eh, at least you can instrument the closed source one. The open source one barely has even that. Just look at what kinds of metrics the prometheus exporter gives with open source, and closed source versions. https://github.com/nginxinc/nginx-prometheus-exporter
@ignaloidas @quad yeah access log parsing is cool until you have hanging connections that didn't close yet, so they don't appear in access logs yet. Metrics aren't just about pretty graphs, they're let you make alers and help diagnose problems when SHTF
@quad @ignaloidas it does expose per-entrypoint and per-upstream metrics in prometheus format
regardless, I'm happy as long as we have more web servers competing. i don't want to end up with another apache conquering the whole world. currently stuff is decent, with a good web server like nginx holding majority, but not monopoly
@wolf480pl It is more a matter of features than performances (look at nginx plus to see differences with nginx : more lb algortihms, cookie stickyness, ...).
traefik can handle most of these advanced features and has a really nice way to manage certificates.
@Fritange like, it doesn't matter that it has all these nice features if an 8-core instance of it falls over under mere 2k req/s
@Fritange so my question, from a person running traefik in prod toba person running traefik in prod, is: what kinda req/s are you seeing, and what should I expect my traefik to handle?
server_name example.com;
location / { proxy_pass http://10.0.0.1; }
(Plus path to ssl cert and key if ssl)
Though? But built in let’s encrypt sure is a QoL improvement I guess
@quad @wolf480pl @piggo Yeah I am a big dummy but most #nginx config files explain them self. Unless you are doing super fancy stuff.
That, and major servers like Apache and IIS has conditioned people into thinking that every web server is torture to set up.
@wolf480pl @piggo
- replies
- 0
- announces
- 0
- likes
- 1
@wolf480pl imo you have a different use case. Maybe you should go to a basic haproxy config. I never told than traefik is the best solution for every infrastructure. In fact, I have many configurations with differents lb/reverse proxy (mostly nginx, haproxy, traefik).
Here is a post comparing different benchmark on lb
https://www.loggly.com/blog/benchmarking-5-popular-load-balancers-nginx-haproxy-envoy-traefik-and-alb/
There is no absolute answer to what is the better lb. It depends on your needs.
@Fritange hmm so you have multiple ingress controllers in a single cluster?