Hello, We use NiFi cluster (ver. 1.1.0) and now have some stability
issues. After a week of work one
node’s heap becomes full and it is not cleaning by GC. So our NiFi cluster
become very unstable and can’t perform any operations until we restart it. We performed a dump analysis in the Eclipse Memory Analyzer and in Memory leaks report got this: 52,775
instances of "org.apache.hadoop.hive.conf.HiveConf",
loaded by "org.apache.nifi.nar.NarClassLoader @ 0x4c02d7b60" occupy 10,665,704,184
(93.19%) bytes. These instances are referenced from one instance
of "java.util.HashMap$Node[]", loaded by "<system
class loader>" Keywords java.util.HashMap$Node[] org.apache.nifi.nar.NarClassLoader @
0x4c02d7b60 org.apache.hadoop.hive.conf.HiveConf As I see, the problem can be related with the Hive Streaming processor.
Maybe someone encountered the same problem and can give advice on how to solve it?
... View more
