Model Testing Job Failed

Whenever I want to test my model on all the test data (1761 samples), the job starts but gets a job failed message. It states that no data was received for 20 seconds. It does work when I manually click several samples to test.

@wouter When looking at the logs I see that the job does succeed, but there’s 35 seconds between the start of the classification and the next message. We think this is because we’re copying some data between containers, which goes through a (pretty slow) network share, and then times out (timeout is 20 seconds for this job). We’ll add some extra logging here to inform what is happening, and thus also won’t time out.

This is now resolved on production!

1 Like