Known Issues

Description of the problem

JDK 8u242 and newer versions cause Site-to-Site failure

2018-12-19 16:19:26,811 ERROR [Site-to-Site Listener] org.apache.nifi.NiFi
 java.nio.channels.IllegalBlockingModeException: null
 at java.base/sun.nio.ch.ServerSocketAdaptor.accept(ServerSocketAdaptor.java:121)
 at org.apache.nifi.remote.SocketRemoteSiteListener$1.run(SocketRemoteSiteListener.java:125)
 at java.base/java.lang.Thread.run(Thread.java:834)

If neither of these options work for you, contact Cloudera Support for other options.

Description of the problem:

JDK 8u271, JDK 8u281, and JDK 8u291 may cause socket leak issue in NiFi due to JDK-8245417 and JDK-8256818. Pay attention to the build version of your JDK because some later builds are fixed as described in JDK-8256818.

Workaround:

Consider using a more recent version of the JDK like 8u282, or builds of the JDK where the issue is fixed.

Description of the problem:

You may experience a performance deterioration after upgrade to HDF 3.4.x with Kafka version 2.1.0.

Workaround:

To work around this issue, upgrade to HDF 3.5.x or see your support representative for an available hotfix.

When you configure a NiFi Connection by changing the Load Balance Strategy, NiFi automatically redistributes the data as needed. If this occurs while data is swapped out from the NiFi process, however, this can result in the data not being swapped back in. When this occurs, the data remains in the queue and not be processed.

Associated error message:

You may see an error message in the logs, similar to: Encountered unknown Swap File ./flowfile_repository/swap/1557762343026-9e6b1ea4-016a-1000-ffff-ffff9a287a94-0ef48376-b919-4aae-9b72-0e7acfd1bfb9.rebalance.swap; will ignore this Swap File. This file should be cleaned up manually.

To avoid this issue:

To avoid this, ensure that you stop the Processor upstream of any connection and allow the queue to empty before changing the Load Balance Strategy.

To work around this issue:

To work around this issue if you do encounter it, shut down NiFi and rename the files in the FlowFile Repository's "swap" directory such that it matches the pattern <timestamp>-<connection identifier>-<uuid>.local.swap instead of <timestamp>-<connection identifier>-<uuid>.<other-partition-name>.swap and then restart NiFi. Be sure only to rename those Swap Files whose name contains the relevant connection's identifier.

Description of the problem:

BUG-108848

Description of the problem:

Kafka 1.1.1 producer cannot produce messages with snappy compression on IBM Power.

Description of the problem: You may experience a performance deterioration after upgrade to HDF 3.4.x with Kafka version 2.1.0.

Workaround: To work around this issue, upgrade to HDF 3.5.x or see your support representative for an available hotfix.

HDF 3.4.1.1 contains Apache Kafka version 2.1.0, while HDP 3.1.0 contains Kafka 2.0.0. If you are using HDF with any HDP components, be aware of this interoperability issue. Review the Notable changes in Apache Kafka 2.1.0 to understand any interoperability issues that may arise.