Hi guys!
Yesterday we discussed the problem about additional resources. Our decision
was to use the following structure of JobOrigin:
{
"name":"j_o",
"url":"savanna://3bfae02a-87fb-4f5e-bee5-baa135ff4be0"
"libs":[swift://example.savanna/lib1.jar, hdfs://user/hadoop/lib2.jar]
}
But today I realize
Hi Arindam,
Currently Savanna has plugin structure and Vanilla plugin supports only
Hadoop 1.1.2. But this restriction is reflected mostly in configs for
Hadoop. You may take a look to all config-defaults files in
https://github.com/stackforge/savanna/tree/master/savanna/plugins/vanilla/resources.
Hi Jon,
Unfortunately, I'm not able to reproduce this issue with vanilla plugin.
The behavior you described is not correct. Here a json I used to repro an
issue:
{
"add_node_groups": [
{
"name": "worker-tasktracker",
"count":1,
Hi folks,
We have some concerns about Oozie+Hive workflow.
Should we install Hive as a separate service to make it work with Oozie? If
it if the case why we *should* have hive-jars in Oozie's share lib in HDFS?
Should the version of separate-installed Hive be equal to hive-jars from
Oozie's share