[ https://issues.apache.org/jira/browse/PIG-3653?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Travis Woodruff updated PIG-3653: --------------------------------- Attachment: PIG-3653-3.patch Good idea. New patch attached that reads a pig.predeployed.jars property. > Add support for pre-deployed jars > --------------------------------- > > Key: PIG-3653 > URL: https://issues.apache.org/jira/browse/PIG-3653 > Project: Pig > Issue Type: Improvement > Reporter: Travis Woodruff > Assignee: Travis Woodruff > Attachments: PIG-3653-2.patch, PIG-3653-3.patch, PIG-3653.patch > > > This may be a niche use case, but this is a simple change that may be useful > to others. > When jobs are started from a network not local to the the Hadoop cluster, the > upload of the job jars can take a significant amount of time. This can be > ameliorated by preinstalling large libraries on the Hadoop slaves. However, > there are certain jars (or parts of jars, e.g., the pig classes) that cannot > be skipped using the existing PigContext.skipJars functionality, since when > specific packages are specified, skipJars are not skipped. -- This message was sent by Atlassian JIRA (v6.1.5#6160)