Load models automatically on server startup?

[reynoldsnlp]

What will it take to daemonize this?

One approach to this sort of thing is Apache UIMA, but this would take a huge refactor.

[reynoldsnlp]

We don't need to daemonize this. The only performance hit comes the first time that a query is made. Thereafter, everything is loaded and it works fine.

The only possible optimization might be to automatically load the model before any requests are made (on server startup), but I'm not convinced that this is necessary or helpful.

Instead, perhaps on the front end we could have a different indicator for "model loading" vs "performing query".

[mjbriggs]

I have not done any work on this issue. I am somewhat sure that we talked about loading the servers on startup not being incredibly useful since a user could attempt to make a request while a server is loading and the model loading would take precedence over the search request.