Tasks Continually Retried because of 429 "App Error"

[Bill Speirs]

I have a Mapreduce job that is obviously run using the pipeline API. I have a number of tasks in my queue that have been retried 20 to 40+ times. When you examine the task under the "Previous Run" tab, the HTTP response is always 429 and the reason is "App Error". Can anyone help me track down what is really going on here? I don't see any exceptions thrown in my logs, so I don't think it's a real error, but rather some quota/threshold I'm hitting.

The other weird thing is that it is taking 25s to 35s to process this task, yet when I look in the logs it is saying only 5,000ms to execute the request that is returning with a 429.

How can I get better debug information to know what quota/threshold I'm hitting? What knob should I turn to make this run faster... ie, not have to be retried that many times?

Thanks!

Bill-

[Tom Kaitchuck]

This is normal behaviour for a MapReduce that has more shards than workers, It is simply the library pushing back on the queue.

--
You received this message because you are subscribed to the Google Groups "Google App Engine Pipeline API" group.
To unsubscribe from this group and stop receiving emails from it, send an email to app-engine-pipeli...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.