Hacker News new | comments | show | ask | jobs | submit login

seems like laurenzos bulletproof node.js code just got caught in the eventloop

The site is running on Wordpress. Incidentally, one of the major reasons for Wordpress blogs to go down under heavy traffic is excessive keep-alive times causing too many Apache threads to be open. It's exactly this lots-of-connections problem that node specializes in handling gracefully.

A postscript to this, now that the blog is back up and I can do some measurements: the server is Apache 2.2.14, and it's using the default keep-alive timeout value of 15 seconds. Looks like a typical case of Wordpress and Apache with the default configuration suddenly getting slashdotted. More on this from patio11:


That was embarrassing. I do know how to craft a good apache config, and am just going to claim a case of being stupid.

It is ironic that node is in fact designed to handle this type of memory/connection scalability problem reliably, but in this case it was pure administrative (me) error.

omg excessive keep-alive times

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact