No problem at all, glad you got it sorted out! The delimiter functions more like a filter, where all subkeys below the prefix with the delimiter are ignored.
A good explanation is found here: http://docs.aws.amazon.com/AmazonS3/latest/dev/ListingKeysHierarchy.html And a quote from that link: ``` To list only the root level objects in the bucket you send a GET request on the bucket with "/" delimiter character. ``` So that makes sense why you could list the contents of the leaf "directory" but nothing above it. Adam On Fri, Jul 21, 2017 at 3:52 PM, Laurens Vets <laur...@daemon.be> wrote: > Sounds like you had some success using the prefix > "AWSLogs/<aws_id>/CloudTrail/ap-northeast-1/2017/07/03/", is that right? > If so, when you right-click and choose "View State", is the timestamp and > latest key name stored? > > Try changing the prefix to > "AWSLogs/<aws_id>/CloudTrail/ap-northeast-1/2017/07/", > clearing the existing state (right-click => "View State" => "Clear State"), > and start the process again. Does it continue to produce files? What > happens as you continue to go up the directory structure (adjusting the > prefix and clearing the state)? Does it stop producing files at some point? > > Keep in mind that for this use case the prefix should always end in a > forward slash ('/'). > > In our case, we've been using ListS3 to ingest CloudTrail logs on NiFi > 1.1.1 for some time, and it has always recursed into S3 "subdirectories" as > expected. > > > I found it, I had Delimiter set to "/". Apparently, this does not work. > After removing this, everything seems to work. > I'm sorry I kept the list busy with this stupid mistake... > > >