Thanks Olga. But what it running in mapreduce mode? Once the distributed cache is enable in this mode, there should still be some way to read these cached files. Actually, searching all the source files in pig-0.7, I can't find 'DistributedCache.getLocalCacheFiles' anywhere and I soppose there is no other way to read cached files. This is what confuse me. Any other ideas?
-Gang ----- 原始邮件 ---- 发件人： Olga Natkovich <ol...@yahoo-inc.com> 收件人： pig-dev@hadoop.apache.org 发送日期： 2010/6/7 (周一) 6:50:01 下午 主 题： RE: distributed cache in pig This is because Hadoop 20 does not support distributed cache in local mode. My understanding is that it would be part of Hadoop 22. Olga -----Original Message----- From: Gang Luo [mailto:lgpub...@yahoo.com.cn] Sent: Monday, June 07, 2010 3:40 PM To: pig-dev@hadoop.apache.org Subject: distributed cache in pig HI all, I notice that whether pig use distributed cache depends on the context (local or mapreduce). When running in mapreduce mode, the distributed cache is always enable (e.g. replicated join). However, I never find such method, DistributedCache.getLocalCacheFiles(job), which get the cached file from the local disk. So, how does pig read these files from local disk? I am looking at the pig 0.7 source code. Thanks, -Gang
