Hi,

I have an oversized queue due to some unknown problem.  That's not the
issue at the moment.  I am restarting the cluster of 3 nodes (all data is
stuck on 1 node) each time the queues stop processing data.  What I observe
is that from a queue of about 2 million flow files it will proceed through
the first few processors and then eventually no changes will happen.  The
queues show as full, ultimately 0 in and 0 out, the debug log shows that
the processors have stopped from back pressure but the queues which are
causing the back pressure are empty when you list them.  If some queues
have 5000 files in the them they will show as empty and if the processor
actually processes any data it will show say 15000 then drop back to 5000.

Is this a known issue? Can I adjust something to make it able to process
all the backlogged data?  Been restarting this for a week and got it from 9
million down to 2 million but it seems to be getting slower.  Drive and
folder space is no issue.  Heap space is around 24GB/node and only 1 node
with the data is about 50% heap.

Regards,

Jeremy

Reply via email to