Hi Sam2020,   Just wanna add,I encounter this issue as well. Got the exact same error while downloading the parcel from Cloudera Manager. Im using local/internal repository.    My permission on /opt/cloudera/parcel-repo is already cloudera-scm without any modification needed. And the sha file also does not have 1 or 256 at the end.   What I did was: 1.Remove old unused 7.1.7 p0 parcels. 2.Restart CM server 3.Redownload 7.1.7 p74 from CM UI.   *Im on cdh 6 planning to upgrade to 7.1.7 p74 without doing the 7.1.7 p0 *I do realize that all other parcels including 7.1.7 p0 have .torrent file except this 7.1.7 p74. Not sure if that have anything to do with it ... View more

Hi, @rahulsingh710 ,   Sorry, but I've run out of ideas. This is really weird and I'm not sure what the problem is. If you have a Cloudera support subscription, I'd suggest you open a case with them, if you already haven't done so.   Cheers, André ... View more

Hi @Aditya-Moghe ,   Could you please share the exact steps you took to install and start NiFi?   André   ... View more

Great .. This works 🕺Thank you @araujo    ... View more

Hi @araujo, Thank you for your answer.  Our Kafka servers are always designed to consist of 100 partitions in my company and each partition belongs to a data of an id. I explicitly generate the partition value attribute according to an id before going into Publish Kafka processor. So, there is no chance that some values could not be valid partitions. Also, I dont know why but every flow file with same partition number does not result in same partition. This one confused me. Maybe, hash function returns a value more than 100 and that value may be processed randomly in the kafka servers since we have only 100 partitions. As a last ditch solution and if it makes sense for the community, I may open a pr to add another config in publish kafka processor whether partition value should be hashed or not.  ... View more

@georg    Did the above answer your question?   Cheers, André ... View more

Hi, @RonMilne ,   A claim can contain one or more flowfiles and NiFi uses it to optimize storage for the content repository. It's totally possible that input and output flowfiles of a processor belong in the same claim.   Please see this article for an explanation of what claims are and how they are managed.   Cheers, André   ... View more

THANK YOU SO MUCH. It makes total sense.     ... View more

@Mun ,   Instead of using the GetSFTP processor, you should use the pattern ListSFTP -> FetchSFTP. The ListSFTP processor is stateful and will remember which files were already fetched before.   Cheers, André ... View more