Closed kush99993s closed 7 years ago
R is single threaded, so Plumber is as well. For future searchers, this type of information will eventually be documented here: https://www.rplumber.io/docs/runtime.html (but it's mostly empty now).
Plumber currently can only handle one request at a time. Meaning that if you have an endpoint or filter that takes 5 seconds to return, all other requests are blocked and inactive for those 5 seconds. This is why it's important to load balance across multiple R processes if performance or availability is important to your API. (Discussed here).
There's talk about incorporating Promise-based handling into Plumber now that these asynchronous features are being developed in the underlying httpuv
package. But there are no definite plans there yet.
Hi,
I was wondering if you can provide little more information about plumber infrastructure ?
Is it single threading or multithreading ?