Reader aborts reading complete data set, but workbench continues to run. Can I have the workbench fail when this occurs?

Hmm, not a great solution but you could run a pre-query to fetch the total number of features in the table. Then after the read you can compare the number of the last feature read vs the result from the pre-query. In the case where the numbers are different you can run it into a terminator.

Changed from a Reader to a FeatureReader which gives the <Rejected> output port and then set 'Rejected Feature Handling' to Terminate Translation.

As the error is intermittent, and can't duplicate on demand, will have wait and see if this solved the problem. But I think it should.

There is a parameter called Ignore Failed Readers. If you had this set to No, then that might resolve the problem. Having said that, it defaults to No anyway, so it's unlikely to help, but I thought I'd mention it.

I am having the exact same issue, reading 5 feature classes, and writing them to another location. Occasionally, one or two of the feature classes will be skipped, but workbench will continue to run/complete.

@mark2atsafe​ -​ I confirmed that I do have the ignore failed readers set to "No". I submitted a case yesterday - https://community.safe.com/s/case/5004Q00002UIXZiQAP/published-job-does-not-fail-when-it-should

I have not had any issues since switching to a Feature Reader Transformer. In fact it hasn't failed yet. Definitely seem differences between it and the Reader.

My impression from the documentation is that 'Ignored Failed Readers' is for when the complete reader fails, say like in an inability to connected. In my case the reader was failing half way through reading the rows and proceeding with only a subset of the data.

Was there any update to the case? I’m getting the same error. 

I tried adding the featureReader Transformer and it fails less often. It seems to run fine off of FME desktop, but fails when published to the server. 

In my case it does appear to be a memory issue. Is there any tips to optimize memory when reading in features so it doesn’t skip them?

Hey @arutherford 

If your issue is related to memory, you can try manually removing unnecessary attributes in the Reader Feature Type dialog or use the AttributeManager or AttributeRemover transformers to eliminate them.

Additionally, we have some helpful knowledge articles on enhancing performance - https://support.safe.com/hc/en-us/sections/25407340008973-Performance - that might assist with this issue.

In the future, feel free to create a new thread for any issues you encounter to get better visibility.

Hope this helps!

Emma