Change the blocksize in 2.5.1

Hello everyone, I've just installed Hadoop 2.5.1 from source code, and I have problems changing the default block size. My hdfs-site.xml file I've set the property property namedfs.blocksize/name value67108864/value /property to have blocks of 64 MB, but it seems that the system

Re: Change the blocksize in 2.5.1

Make it final and bounce namenode On Nov 20, 2014 3:42 PM, Tomás Fernández Pena tf.p...@usc.es wrote: Hello everyone, I've just installed Hadoop 2.5.1 from source code, and I have problems changing the default block size. My hdfs-site.xml file I've set the property property

RE: Change the blocksize in 2.5.1

It seems HADOOP_CONF_DIR is poiniting different location!!? May be you can check hdfs-site.xml is in classpath when you execute hdfs command. Thanks Regards Rohith Sharma K S -Original Message- From: Tomás Fernández Pena [mailto:tf.p...@gmail.com] On Behalf Of Tomás Fernández Pena

Re: Change the blocksize in 2.5.1

Hi Thanks for your kind answers. I've found the problem. The point is that I'd only specified the dfs.blocksize parameter in the hdfs-site.xml of the NameNode and datanodes, but no in the client. My question now is, how can I avoid that the client change the value of blocksize? I've tried to

Re: MR job fails with too many mappers

Hi, as I said before, I wrote TableInputFormat and RecordReader extension that reads input data from an Hbase table, in my case every single row is associated with a single InputSplit. For example if I have 30 rows to process, my custom TableInputFormat will generate 30 input splits

Fwd: Upgrade from 4.6 to 5.2

-- Forwarded message -- From: Mike Rapuano mrapu...@choicestream.com Date: Thu, Nov 20, 2014 at 10:05 AM Subject: Upgrade from 4.6 to 5.2 To: user-h...@hadoop.apache.org Trying to upgrade a cluster running cdh4.6 to cdh5.2 it hangs upgrading hdfs metadata with Namenode RPC Wait.

Question zkfc monitor rpc timeout

hi all, We are testing the zkfc for name node HA. And I see the design of the zkfc, there is a monitor thread which will monitor the healthy of name node and the design mention that the RPC timeout option, i want to know how to configure this option and if the option is used just used when the

Hadoop Streaming Use Non Printing Character as Key/Value separator

Hi there, I am writing a hadoop streaming job where certain columns contain natural languages. In that case, use '\t' as the default delimiter is not a choice for me. Does anything know how to pass a non printing character, like SOH 'start of header' as the key/value separator? I tried to pass

RE: About HDFS-RAID

We just start doing the work, so it’s not available even in trunk. Regards, Yi Liu From: Vincent,Wei [mailto:weikun0...@gmail.com] Sent: Wednesday, November 19, 2014 3:08 PM To: user@hadoop.apache.org Subject: Re: About HDFS-RAID Thanks Liu But I found that the HDFS-7285 still open, so it

What the option ha.health-monitor.rpc-timeout.ms really means?

hi, all I am using the name node HA feature(zkfc), and there is a configuration: ha.health-monitor.rpc-timeout.ms, There are two situation: 1. Active name node is down if the zkfc will wait ha.health-monitor.rpc-timeout.ms, then call the failover ? Or not wait? 2. Active name node is too busy