zjffdu commented on a change in pull request #3411: ZEPPELIN-4272: Zeppelin
fails to use s3a configured for zeppelin.note…
URL: https://github.com/apache/zeppelin/pull/3411#discussion_r306138301
##########
File path:
zeppelin-zengine/src/main/java/org/apache/zeppelin/notebook/FileSystemStorage.java
##########
@@ -60,16 +63,42 @@
public FileSystemStorage(ZeppelinConfiguration zConf, String path) throws
IOException {
this.zConf = zConf;
this.hadoopConf = new Configuration();
+ URI zepConfigURI;
+ URI defaultFSURI;
+
+ zepConfigURI = getURI(path, "Failed to get URI path");
// disable checksum for local file system. because interpreter.json may be
updated by
// non-hadoop filesystem api
// disable caching for file:// scheme to avoid getting LocalFS which does
CRC checks
+
this.hadoopConf.setBoolean("fs.file.impl.disable.cache", true);
- this.hadoopConf.set("fs.file.impl", RawLocalFileSystem.class.getName());
+ String defaultFS = this.hadoopConf.get(FS_DEFAULTFS);
+ defaultFSURI = getURI(defaultFS, "Failed to get defaultFS URI");
+
+ if (isS3AFileSystem(defaultFSURI, zepConfigURI)) {
+ this.hadoopConf.set("fs.s3a.impl", S3AFileSystem.class.getName());
Review comment:
I doubt why hadoop doesn't support it, ideally hadoop should support this
otherwise any downstream project has to set s3 filesystem explicitly which is
annoying.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
