@s198 In your use case you could probably handle this without needing to have a sub process group usng flowfile concurrency since the GetHDFSFileInfo processor sets a number of attributes in the FlowFiles it produces that you can use to correlate the various FlowFiles to one another. Since you have already written out all your individual HDFS files to your SFTP processor, you could remove all content on the FlowFiles using ModifyBytes (no sense in waiting CPU to merge content if you don't need to keep it anymore), so all you have are zero byte FlowFiles with attribute data. Then feed that stream of zero byte FlowFiles to a MergeContent processor. I would configure your Merge Content as below: Assume defaults for any property not mentioned here Correlation Attribute Name = hdfs.path min num entries = <set to a value higher then you would expect to listed in any HDFS directory> max num entries = <set to value larger the above> max bin age = <set to value high enough to allow all files to reach this processor before this time expires> What above will do is start placing all FlowFiles that have the same value in the FlowFile's hdfs.path flowfile attribute in the same MergeContent bin. The min num entries will prevent the MergeContent processor from merging the bin until this value is reached or until the max bin age expires. The bin age starts as soon as first FlowFile is allocated to the bin. So basically since we don't know how many files might be in any given HDFS directory, we are controlling merge via bin age instead of number of FlowFiles. This builds some delay in your dataflow, but results in 1 FlowFile output for each HDFS directory listed out and fetched. You can then take that 1 zero byte Merged FlowFile and use it to complete your single FlowFile downstream processing of job completion. While above would work in ideal conditions, you should always design with failure possibility in mind. I would still recommend placing every component from "GetHDFSFileInfo --> RouteOnAttribute --> UpdateAttribute --> FetchHDFS --> PutSFTP -->ModifyBytes --> UpdateAttribute" inside a process group configured with "FlowFile Concurrency = Single FlowFile Per Node" and "Outbound Policy = Batch Output". This would allow you to make sure that all Fetched FlowFiles are successfully processed (written to SFTP server) before any were out put form the process group to the ModifyBytes and MergeContent processors. You never know when a issue may prevent of slow writing to SFTP server. This allows you to more easily handle those failures and assure any retries or error handled before exiting that PG and completing you job. This also allows you to set a much shorter max bin age in your MergeContent processor since before any FlowFile in that batch are released they will all have been processed, so the will all reach MergeContent at same time. Please help our community thrive. If you found any of the suggestions/solutions provided helped you with solving your issue or answering your question, please take a moment to login and click "Accept as Solution" on one or more of them that helped. Thank you, Matt
... View more
