> There are some FreeBSD machines in our infrastructure which run NGINX. After the recent announcement of the F5 purchase of NGINX we decided to move back to Lighttpd.
Right, but that doesn't really explain anything. Maybe the author is concerned about some future changes in the nginx project and they're preemptively moving away from NGINX, but even if that's the reason, it's not much of one. If they're using the open source version of NGINX, then it seems highly unlikely anything will change that will impact them.
I don't find that satisfactory. They don't have to explain themselves at all, but it would be nice know why the F5 purchase would charter that decision. Also if the F5 purchase makes you want to leave nginx, why lighttpd? They don't have to answer either question, but we can still say it would be nice to have it explained.
It's not even an article really. It's a blog post and it seems targeted as a reference for other admins with chargen.one. Not sure why it got posted to HN.
Just FYI: You are shadow banned... I was for a while too without realizing it. The really nasty thing about HN is they can't even claim that bots are accidentally censoring people. They do it manually!
Depends what you're doing. If you just want "http server" of some kind, then either is fine. Sure, features vary, but how much does it actually matter? Yeah, http2 is nice, newer compression is good, ... features matter, but for a lot of uses it really doesn't make a noticeable impact.
I would argue the opposite: Anything but nginx is probably overkill unless you have specific requirements. I mean, sure, if you just want to serve some files from a development machine into the intranet right now, `python -m http.server` will do the trick. But for anything that is either a productive system, or one serving content to the public internet, nginx is a sane default choice.
For a production system facing the public, lightttpd strikes me as inappropriate.
Any of the “I don’t care how poorly it supports modern browsers, I just need something barebones that speaks http” usecases the person I was responding to seemed to be suggesting would typically be something like a local dev http server, or a barebones http server sitting behind a more robust proxy.
Which brings us back to, for those usecases, nginx is probably overkill.
You mean that unnecessarily complex successor of http1.1 that brings nothing to the table you'd ever need if your website wasn't a bloated piece of </censored>?
The article explicitly says that they had been using lighttpd, started switching to nginx because lighttpd appeared to go dead, and were now switching back because lighttpd had resuscitated while they were worried about F5 taking over nginx.
I have no idea what their use case was, other than the lack of HTTP/2 support, if all you are doing is just serving lots of static files Lighttpd is still a very good choice.
> We haven’t seen a lot of open source projects doing well after the parent company had been acquired.
Seems premature. Often the project gets forked when the parent project goes downhill. Right now NGINX is an excellent, stable, performant, flexible web server with lots of community support and documentation.
28 comments
[ 2.1 ms ] story [ 59.6 ms ] thread> There are some FreeBSD machines in our infrastructure which run NGINX. After the recent announcement of the F5 purchase of NGINX we decided to move back to Lighttpd.
They were in the middle of a lighthttpd -> nginx transition and noticed that lighthttpd has now fixed whatever caused the transition to start.
Like all acquisitions, this one creates uncertainty, so they backed out the nginx deployment.
This seems totally rational on their end. Why fix what is not broken, and why depend on two things when you can depend on one?
Doesn't make any sense, technical or other. Clickbait article.
The acquisition led them to abandon their plan to migrate to nginx.
Also, I thought running through the config for their production site was useful.
Any of the “I don’t care how poorly it supports modern browsers, I just need something barebones that speaks http” usecases the person I was responding to seemed to be suggesting would typically be something like a local dev http server, or a barebones http server sitting behind a more robust proxy.
Which brings us back to, for those usecases, nginx is probably overkill.
[1] https://h2o.examp1e.net
Haven't used Lighttpd in 15 years but I wonder how well it compares to nginx and Apache.
Seems premature. Often the project gets forked when the parent project goes downhill. Right now NGINX is an excellent, stable, performant, flexible web server with lots of community support and documentation.