Phil Zampino created KNOX-1518:
----------------------------------

             Summary: Large HDFS file downloads are incomplete when content is 
gzipped
                 Key: KNOX-1518
                 URL: https://issues.apache.org/jira/browse/KNOX-1518
             Project: Apache Knox
          Issue Type: Improvement
          Components: Server
    Affects Versions: 1.1.0
            Reporter: Phil Zampino
            Assignee: Phil Zampino
             Fix For: 1.2.0


org.apache.knox.gateway.filter.rewrite.impl.UrlRewriteResponse employs  
java.util.zip.GZIPInputStream for gzipped content streams.

There appears to be an expectation in the GZIPInputStream of the 
InputStream#available() method, for which the behavior is varied across 
InputStream implementations. InputStream implementations that do not satisfy 
this expectation cause the GZIPInputStream to terminate prematurely, resulting 
in only partial reads.

There is an OpenJDK bug ([https://bugs.openjdk.java.net/browse/JDK-8081450]) 
for this, and the Oracle JDK suffers from the same.

This can be overcome in Knox with code.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to