[
https://issues.apache.org/jira/browse/OOZIE-171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13101772#comment-13101772
]
Hadoop QA commented on OOZIE-171:
---------------------------------
bwhite remarked:
I made a work around for Streaming users that changes to a directory "a" and
then runs the program with whatever args you have. You put all of your files
into an archive and use <archive>myarc.tar.gz#a</archive>, now with this
wrapper you are in the proper directory with all of the correct names. Use it
like ./wrapper.py my_streaming_mapper <my_mapper_args>
http://pastebin.com/hPnHjybX
Of course this is just a temporary fix.
> GH-166: Modify the logic of adding .so and .so.1 files into cache
> -----------------------------------------------------------------
>
> Key: OOZIE-171
> URL: https://issues.apache.org/jira/browse/OOZIE-171
> Project: Oozie
> Issue Type: Bug
> Reporter: Hadoop QA
>
> It is related to code at JavaActionExecutor:addToCache() method
> String fileName = filePath.substring(filePath.lastIndexOf("/") + 1);
> if (fileName.endsWith(".so") || fileName.contains(".so.")) {
> // .so files
> if (!fileName.endsWith(".so")) {
> int extAt = fileName.indexOf(".so.");
> fileName = fileName.substring(0, extAt + 3);
> }
> uri = new Path(path.toString() + "#" + fileName).toUri();
> uri = new URI(uri.getPath());
> }
> The following things should work in <file> tag.
> 1. a.so.1 to be linked as a.so.1
> 2. a.so.1#b.so to be linked as b.so
> 3. a.so#a.so should be linked as a.so
> 4. a.so should be linked as a.so.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
