On the first run you want namenode to initialize its directories (where it
store VERSION file, fsimage and edits).
On the subsequent formats - you are making sure you have a new EMPTY file
system. If you don't do format NameNode will load up fsimage and edits.
There is also matter of generating
NameNode metadata - is in memory data structures describing file system
structure (directories, files, and file's blocks).
NameNode image is on-disk representation of the metadata. (fsimage file).
Edit logs - is a journal that keeps all the events that cause changes in the
metadata and allows
You can run multiple data nodes on the same machine.
You should create a separate config directory for each dn.
The following stuff needs to be created
hdfs-site.xml
Pid/log/data/tmp dirs
log4.properties
Master/slaves
And then start these data nodes
( something like this: bin/hdfs --config
The JIRA is committed already. So loosing a directory will not stop the
datanode anymore. If directory shows up later again it will not be added
automatically.
On 9/11/09 10:56 AM, Edward Capriolo edlinuxg...@gmail.com wrote:
If you lose a directory the datanode stops. See
For JMX you can also look at JMXGet.java class. You can use this object to
get the data thru JMX.
Boris
On 8/24/09 1:27 AM, Stas Oskin stas.os...@gmail.com wrote:
Hi.
One way you can do this is thought JMX.
http://www.jointhegrid.com/svn/hadoop-cacti-jtg/trunk/src/com/jointhegrid/had