Can you run normal MR jobs, like the example Pi calculation? Sometimes a no-reducer problem stems from DNS issues— reducers use node names, not IP addresses, so you need to have each machine knows how to resolve the names of all the other machines in the cluster.
If it's a new cluster, you may want to look at CDH3, as it is a pretty big leap forward. On Aug 11, 2011, at 9:35 AM, Travis Powell wrote: > Have you checked your logs? These are often the best places to start. > Look at the running job and click on the running count, the current task, > then the task logs. > > Sometimes they’re helpful, sometimes they’re not. > > http://hadoop-master:50030/jobtracker.jsp > > Travis Powell / tpow...@tealeaf.com > > From: hadoop n00b [mailto:new2h...@gmail.com] > Sent: Thursday, August 11, 2011 9:32 AM > To: user@hive.apache.org > Subject: Reducer Issue in New Setup > > Hello, > > We have just setup Hive on a new Hadoop cluster. > > When I run a select * on a table, it works fine but when I run any query > which needs a reducer, like count(1) or a where condition, the query just > sits there doing nothing (map 0%). I see some message like no reducers to > run. How do I fix this? The configuration et al is exactly like another > environment that is working fine. > > We are running CDH2 with Hadoop 0.20.1 and Hive 0.4.1 > > Cheers!
