Re: Cannot resolve Datonode address in slave file
Hi, Sim, I've two suggessions, if you haven't done yet: 1. Check if your other hosts can ssh to master. 2. Take a look at logs of other hosts. 2009/4/2 Puri, Aseem aseem.p...@honeywell.com Hi I have a small Hadoop cluster with 3 machines. One is my NameNode/JobTracker + DataNode/TaskTracker and other 2 are DataNode/TaskTracker. So I have made all 3 as slave. In slave file I have put names of all there machines as: master slave slave1 When I start Hadoop cluster it always start DataNode/TaskTracker on last slave in the list and do not start DataNode/TaskTracker on other two machines. Also I got the message as: slave1: : no address associated with name : no address associated with name slave1: starting datanode, logging to /home/HadoopAdmin/hadoop/bin/../logs/hadoo p-HadoopAdmin-datanode-ie11dtxpficbfise.out If I change the order in slave file like this: slave slave1 master then DataNode/TaskTracker on master m/c starts and not on other two. Please tell how I should solve this problem. Sim -- M. Raşit ÖZDAŞ
RE: Cannot resolve Datonode address in slave file
Hi Rasit, Now I got a different problem when I start my Hadoop server the slave datanode do not accept password. It gives message permission denied. I have also use the commands on all m/c $ ssh-keygen -t dsa -P '' -f ~/.ssh/id_dsa $ cat ~/.ssh/id_dsa.pub ~/.ssh/authorized_keys But my problem is not solved. Any suggestion? -Original Message- From: Rasit OZDAS [mailto:rasitoz...@gmail.com] Sent: Thursday, April 02, 2009 6:49 PM To: core-user@hadoop.apache.org Subject: Re: Cannot resolve Datonode address in slave file Hi, Sim, I've two suggessions, if you haven't done yet: 1. Check if your other hosts can ssh to master. 2. Take a look at logs of other hosts. 2009/4/2 Puri, Aseem aseem.p...@honeywell.com Hi I have a small Hadoop cluster with 3 machines. One is my NameNode/JobTracker + DataNode/TaskTracker and other 2 are DataNode/TaskTracker. So I have made all 3 as slave. In slave file I have put names of all there machines as: master slave slave1 When I start Hadoop cluster it always start DataNode/TaskTracker on last slave in the list and do not start DataNode/TaskTracker on other two machines. Also I got the message as: slave1: : no address associated with name : no address associated with name slave1: starting datanode, logging to /home/HadoopAdmin/hadoop/bin/../logs/hadoo p-HadoopAdmin-datanode-ie11dtxpficbfise.out If I change the order in slave file like this: slave slave1 master then DataNode/TaskTracker on master m/c starts and not on other two. Please tell how I should solve this problem. Sim -- M. Raşit ÖZDAŞ
Re: Cannot resolve Datonode address in slave file
you should append id_dsa.pub to ~/.ssh/authorized_keys on the other computers from the cluster. if your home directory is shared by all of them (e.g., you're mounting /home/$user using NFS), cat ~/.ssh/id_dsa.pub ~/.ssh/authorized_keys might work. however, if it isn't shared, you might use 'ssh-copy-id' to all your nodes (or append id_dsa.pub manually). 2009/4/2 Puri, Aseem aseem.p...@honeywell.com Hi Rasit, Now I got a different problem when I start my Hadoop server the slave datanode do not accept password. It gives message permission denied. I have also use the commands on all m/c $ ssh-keygen -t dsa -P '' -f ~/.ssh/id_dsa $ cat ~/.ssh/id_dsa.pub ~/.ssh/authorized_keys But my problem is not solved. Any suggestion? -Original Message- From: Rasit OZDAS [mailto:rasitoz...@gmail.com] Sent: Thursday, April 02, 2009 6:49 PM To: core-user@hadoop.apache.org Subject: Re: Cannot resolve Datonode address in slave file Hi, Sim, I've two suggessions, if you haven't done yet: 1. Check if your other hosts can ssh to master. 2. Take a look at logs of other hosts. 2009/4/2 Puri, Aseem aseem.p...@honeywell.com Hi I have a small Hadoop cluster with 3 machines. One is my NameNode/JobTracker + DataNode/TaskTracker and other 2 are DataNode/TaskTracker. So I have made all 3 as slave. In slave file I have put names of all there machines as: master slave slave1 When I start Hadoop cluster it always start DataNode/TaskTracker on last slave in the list and do not start DataNode/TaskTracker on other two machines. Also I got the message as: slave1: : no address associated with name : no address associated with name slave1: starting datanode, logging to /home/HadoopAdmin/hadoop/bin/../logs/hadoo p-HadoopAdmin-datanode-ie11dtxpficbfise.out If I change the order in slave file like this: slave slave1 master then DataNode/TaskTracker on master m/c starts and not on other two. Please tell how I should solve this problem. Sim -- M. Raşit ÖZDAŞ -- Guilherme msn: guigermog...@hotmail.com homepage: http://germoglio.googlepages.com
Cannot resolve Datonode address in slave file
Hi I have a small Hadoop cluster with 3 machines. One is my NameNode/JobTracker + DataNode/TaskTracker and other 2 are DataNode/TaskTracker. So I have made all 3 as slave. In slave file I have put names of all there machines as: master slave slave1 When I start Hadoop cluster it always start DataNode/TaskTracker on last slave in the list and do not start DataNode/TaskTracker on other two machines. Also I got the message as: slave1: : no address associated with name : no address associated with name slave1: starting datanode, logging to /home/HadoopAdmin/hadoop/bin/../logs/hadoo p-HadoopAdmin-datanode-ie11dtxpficbfise.out If I change the order in slave file like this: slave slave1 master then DataNode/TaskTracker on master m/c starts and not on other two. Please tell how I should solve this problem. Sim