Hi John, > Could an excessive number/size of znodes be a factor? I don't think this would be a likely case... more like a client error I think.
The stack-trace suggests that the client closed the TCP session unexpectedly. Maybe the clients are missing the zookeeper.close() calls? Also I wonder if it is caused by https://issues.apache.org/jira/browse/ZOOKEEPER-1105 - are you using C (or python) client? (there was this bug, that the C client didn't wait for the proper session close to finish) Btw: please consider upgrading your cluster, 3.4 is end-of-life now and it is not supported by the community anymore. Cheers, Mate On Thu, Jan 14, 2021 at 7:55 PM John Lindwall <jlindw...@yahoo.com.invalid> wrote: > We're seeing thousands of these a day in our zookeeper logs (zookeeper > 3.4.6): > WARN [NIOServerCxn.Factory:X.X.X.X/X.X.X.X:Y:NIOServerCnxn@357] - caught > end of stream exceptionEndOfStreamException: Unable to read additional data > from client sessionid 0xABC, likely client has closed socket > at > org.apache.zookeeper.server.NIOServerCnxn.doIO(NIOServerCnxn.java:228) > at > org.apache.zookeeper.server.NIOServerCnxnFactory.run(NIOServerCnxnFactory.java:208) > at java.lang.Thread.run(Thread.java:745) > > Any ideas of the cause of this? Could an excessive number/size of znodes > be a factor? We're not seeing any obvious client-side issues. We're not > sure but we believe that these were not happening earlier. > > -- John