The npm blog has been discontinued.
Adding redundancy to npm-www
We talked recently about our new registry architecture. Tonight we had 13 minutes of downtime on npmjs.org, the website (which we call www), without any accompanying downtime in the registry.
What happened: the website node app stopped responding to connections (but didn’t actually crash). We’ve not fully dug into why that happened yet, but it’s possibly related to a soft restart we did earlier today.
What we’re doing: we’re going to change the way we restart www after changes. We’re also going to add a second, redundant www server to avoid having the site completely taken down if one box malfunctions. Part of being npm, Inc. means we now have the resources to add this kind of redundancy, which we didn’t previously.