Hi Lior,

I encountered the same problem in my test environment and at first I was not sure if I destroyed the LuceneSearch with the Lucene update during the following commit. (https://github.com/apache/zeppelin/commit/7288702a53d9c246bd6e7194e03e00f9b64fe61a)

While debugging the code, I found the configuration option ZEPPELIN_SEARCH_INDEX_REBUILD. The default value is false, so I changed it to true. Since I don't persist the LuceneIndex in my K8s cluster in the Zeppelin Server Pod, this option rebuilds the index on every startup.

Of course, rebuilding can take a long time and consumes resources during rebuilding, depending on how many notes you have.

I hope you can solve the problem like me.

Best regards
Philipp


Am 02.09.21 um 13:31 schrieb Lior Chaga:
Hi,
I'm facing the same issue as Axel reported in https://issues.apache.org/jira/browse/ZEPPELIN-5437 <https://issues.apache.org/jira/browse/ZEPPELIN-5437> (but without the preceding NullPointerExceptions)

To be honest, when I first tried remote debug to understand why I don't see results, I managed to debug into the SearchService, to make sure it's not the NoSearchService implementation.
After restarting the pod, I got the exception:

    /LuceneSearch.java[query]:135) - Failed to open index dir
    MMapDirectory@/tmp/zeppelin-index
    lockFactory=org.apache.lucene.store.NativeFSLockFactory@5270/
    /8e8d, make sure indexing finished OK/
    /org.apache.lucene.index.IndexNotFoundException: no segments* file
    found in MMapDirectory@/tmp/zeppelin-index
    lockFactory=org.apache.lucene.store.NativeFSLockFactory@52708e8d:
    files: [write.lock]/


I'm running 0.10, using FileSystemNotebookRepo, and notes are stored on HDFS.

Working perfectly with zeppelin 0.8.2...

Any clue why?
Thanks

Reply via email to