Greetings All, I’m configuring a fresh Apache setup with Hadoop 2.2, Pig 0.12 and Hive 0.12. Things started off well; HDFS and YARN worked without a problem, then Hive and Pig. Next I configured Hive to use a remote metastore and at that point Pig stopped working with the error in the subject line.
Now it doesn’t look like these two things should be at all related, but given that Pig was working before I configured the metastore to be remote (that all works fine), and I’ve verified that pig works properly if I use PigStorage rather than HCatLoader, it seems to point to some kind of Pig/HCatalog interaction. Pig is being started with –useHCatalog; It can also ‘load’ and ‘describe’ tables accessed via HCatLoader. It’s when a m/r job is submitted (dump) that the error is thrown. So, the problem seems to be that m/r jobs using tables loaded from PigStorage work fine; those using HCatLoader fail. Hive is working fine with HCatalog. Except for the same few messages copied 12 times by various websites, there’s not a lot out there on this error, and most of the time the posts are asking about Pig 0.12 and Hadoop 2.2. I did find, and followed, a Pig/Hcatalog configuration ‘blog post: http://www.thecloudavenue.com/2013/11/InstallingAndConfiguringHCatalogAndIntegratingWithPig.html but that didn’t change anything. I have confirmed that pig is using the new recompiled for Hadoop 23 JAR file. Any suggestions on how to debug this? Regards, - Steve
