vokkim / crudivore Goto Github PK

The tests always close the port 5000 in express, but all the express routes remain and calling start in server.js again does not overwrite them.

The result is that all the tests use the same, first configuration. I didn't find a good way to fix this, it may require us to rethink how the tests are initiated.

Add support for proper logging of events with different levels such as error, info and debug.

Possible libraries:

• Winston
• Bunyan

Crudivore should survive whenever PhantomJS crashes:

• Log the incident
• Restart the PhantomJS process (?)
• Serve the request with new worker
• Detect possible infinite loops for PhantomJS startup

Google, Facebook, Bing etc. can cause plenty of concurrent requests on the site, and Crudivore should be able to serve the requests within a reasonably low response time.

• Return 503 if PhantomJS workers are unavailable
• Write a stress test case with Gatling or similar to test the response times
• Document the concurrency related problems and suggestions for cache usage with e.g. Varnish

Additionally, Crudivore could offer a configurable option to return some 50x status code in case of a timeout for waiting window.crudivore.pageReady, therefore allowing sites to respond properly when pageReady is not set for example due to javascript failures.

• Try the new PhantomJS 2 for more stabile service