Hello, Swap files are written under the flow file repo, in a sub directory called swap. I don't think it is configurable anymore.
When a queue has flow files in it, the content of those flow file is not in memory in the queue, its just java objects representing the pointers to the flow files. So swapping is to help with memory when you have too many java objects, it writes out the information about what was in the queue to a swap file, but again its not the content of those flow files which is why it is under flow file repo. Thanks, Bryan On Fri, Dec 6, 2019 at 5:10 AM sanjeet rath <rath.sanj...@gmail.com> wrote: > > Hi Team, > > Any help on my trailed mail query will be very helpful.One more thing if i > add a new directory for swap storage will it improve any performance in > processing. > > Thanks & Regards, > Sanjeet > > On Thu, 5 Dec, 2019, 12:06 PM sanjeet rath, <rath.sanj...@gmail.com> wrote: >> >> Hi Team, >> >> As per the nifi document, once the number of flow file in a connection >> queue will reach the threshold value, then it removed from hash map(Java >> heap space) and moved to disc. >> could u please confirm the deault which repository it moves to , it's >> content repository or to flowfile repository. >> >> could i configure the swap directory using nifi.swap.storage.directory >> parameter ? >> >> Thanks, >> sanjeet
