Are you waiting for the data set to fully load before allowing the process to respond to requests?

Are you running just a single node.js process or more than one?

Does the node.js process restart if it crashes?

Could you run in a debugger with a breakpoint inside a check for records being empty?

> it stores 4 gigs of data in memory

I'm not sure how a single node process was able to load 4 gbs worth of data into memory? I thought there was a hard limit of 1.5GB per process.

It's 1.7GB (for 64bit node) unless you change the limit with the `--max-old-space-size` flag.

Link to source broken.

What do you mean? If I click the headline I go to the article.

The article has a link to "source" at the bottom that 404s.

That article is the source, but yes, the source link should not be there.