UnsatisfiedLinkError on Lib rocks DB dll when developing with Kafka Streams

后端 未结 6 1123
夕颜
夕颜 2020-12-18 05:34

I\'m writing a Kafka Streams application on my development Windows machine. If I try to use the leftJoin and branch features of Kafka Streams I get

相关标签:
6条回答
  • 2020-12-18 05:56

    You are missing some native libraries that the rocksdb dll depends on. See https://github.com/facebook/rocksdb/issues/1302

    0 讨论(0)
  • 2020-12-18 05:58

    Faced similar issue in Mac. As per this link, https://github.com/facebook/rocksdb/issues/5064 issue is related to older libc installed in my version of Mac OS (10.11.6).

    0 讨论(0)
  • 2020-12-18 06:00

    I had same issue while using jdk 1.8. It got resolved when I changed it to jre.

    0 讨论(0)
  • 2020-12-18 06:05

    My problem was permissions in /tmp/ directory (CentOS)

    rockdb uses

    java.io.tmpdir 
    

    system property internally to decide where to place librocksdbjnifile, usually something like this /tmp/librocksdbjni2925599838907625983.so

    Solved by setting different tempdir property with appropriate permissions in kafka-streams app.

    System.setProperty("java.io.tmpdir", "/opt/kafka-streams/tmp");
    
    0 讨论(0)
  • 2020-12-18 06:18

    Recently I came through this problem too. I managed to solve this in two steps:

    1. Delete all librocksdbjni[...].dll files from C:\Users\[your_user]\AppData\Local\Temp folder.
    2. Add maven dependency for rocksdb in your project, this works for me: https://mvnrepository.com/artifact/org.rocksdb/rocksdbjni/5.0.1

    Compile your Kafka Stream App and run it. It should work!

    0 讨论(0)
  • 2020-12-18 06:19

    I updated my kafka-streams project to the latest released version 1.0.0.

    This version suffers of this bug but after patching it and uploading this patched version on the internal Artifactory server we were able to execute our kafka-streams agent both on Windows and on Linux. The next versions 1.0.1 and 1.1.0 will have this bug fix so as soon as one of these versions will be released we will switch to them instead of the patched version.

    To sum up the Kafka guys solved this bug with the 1.0.0 release.

    0 讨论(0)
提交回复
热议问题