Pipeline failures

[marcominerva]

Motivation and Context (Why the change? What's the scenario?)

See #432.

[dluc]

I would avoid the "string" type which limits to a single unstructured entry, and use List<PipelineLogEntry> LogEntries if this is really needed. Thinking about the scenarios, I still don't see a use case though.

[dluc]

how do we ensure this is consistent?

for instance, if ContentStorage becomes unavailable and the pipeline job expires, the value will remain false and never change to true because there's nothing left running to retry.

[dluc]

ditto about consistency. I think a better approach would be assigning an approximate completion time and using that to establish if the pipeline failed (one scenario would be "it failed because it timed out")

[dluc]

which code is throwing this exception? why stop retrying?

[dluc]

ditto, where is the exception coming from? e.g. if UpdatePipelineStatusAsync is failing in the try block it will likely fail also in the catch block

[marcominerva]

The Logs property was just an idea, but we can remove it for now, the important thing is how to understand if the pipeline actually failed. I have added the catch blocks in the Distributed and InProcess Orchestrators to handle all unexpected errors like an attempt to decode an invalid PDF file (that will cause an Exception in the stepHandler.InvokeAsync method).

@dluc Your idea about assigning an approximate completion time and use it to determine the status of the pipeline is quite interesting and will allow to add the failure management without touching the Orchestrators at all, but how to determine a suitable value for the completion time? Maybe it could be a setting in configuration, that is saved in the DataPipeline object when building it?