Any idea to speed up the open_mfdataset for reading many many big netCDF files?

[wqshen]

I have several WRFout files for 20-year climate simulations, when i use open_mfdataset to read them, it takes me 10 - 20 minutes to finish on my server.

Is there a way to speed up this process?
Multiprocessing ??

[shoyer]

Try changing the coords argument to list the explicit coordinates that should be concatenated together. (We should probably change this argument name: it makes more sense on concat than open_mfdataset.)

[jhamman]

@wqshen - can you report the versions of xarray/dask that you are using (the original issue template had examples of how to do this)?

[jhamman]

Closed via #1981