You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@nutch.apache.org by Ken Krugler <kk...@krugle.net> on 2006/01/14 18:33:28 UTC

Error running MapReduce - Host key verification failed

Hi all,

We've got the nutch-2006-01-12.tar version of Nutch, and are trying 
to run it on three machines.

192.168.0.100 is the "master" machine, where we run the JobTracker 
and NameNode processes.

192.168.0.101 and 192.168.0.103 are the "slave" machines, where we 
rung the TaskTracker and DataNode processes.

When we fire off the the Nuch daemons with ./bin/start-all.sh, we get 
the following error right away:

192.168.0.103: rsync from 192.168.0.100:/home/crawler/nutch
192.168.0.103: Host key verification failed.
192.168.0.103: rsync: connection unexpectedly closed (0 bytes 
received so far) [receiver]
192.168.0.103: rsync error: error in rsync protocol data stream (code 
12) at io.c(420)
192.168.0.103: starting datanode, logging to 
/home/crawler/tmp/logs/nutch-crawler-datanode-crawlerw3.log

When I dump the datanode logfile from this 192.168.0.103 machine, I get:

060114 082814 10 parsing file:/home/crawler/nutch/conf/nutch-default.xml
060114 082814 10 parsing file:/home/crawler/nutch/conf/nutch-site.xml
060114 082814 10 Opened server at 50010
060114 082814 11 Starting DataNode in: /home/crawler/tmp/ndfs/data
060114 082814 11 using BLOCKREPORT_INTERVAL of 3314538msec
060114 082814 11 Exception: java.net.ConnectException: Connection refused
060114 082814 11 Lost connection to namenode.  Retrying...
060114 082819 11 using BLOCKREPORT_INTERVAL of 3314538msec
060114 082819 11 Exception: java.net.ConnectException: Connection refused
060114 082819 11 Lost connection to namenode.  Retrying...
060114 082824 11 using BLOCKREPORT_INTERVAL of 3314538msec
060114 082824 11 Exception: java.net.ConnectException: Connection refused
060114 082824 11 Lost connection to namenode.  Retrying...
060114 082829 11 using BLOCKREPORT_INTERVAL of 3314538msec
060114 082829 12 Client connection to 192.168.0.100:8009: starting

When I dump the tasktracker logfile from 192.168.0.103, I get:

060114 082832 parsing file:/home/crawler/nutch/conf/nutch-default.xml
060114 082832 parsing file:/home/crawler/nutch/conf/nutch-site.xml
060114 082832 Server listener on port 50050: starting
060114 082832 Server handler 0 on 50050: starting
060114 082832 Server handler 1 on 50050: starting
060114 082832 Server listener on port 50040: starting
060114 082832 Server handler 0 on 50040: starting
060114 082832 Server handler 1 on 50040: starting
060114 082832 Lost connection to JobTracker 
[main1/192.168.0.100:8010]. ex=java.net.ConnectException: Connection 
refused  Retrying...

It seems like the 192.168.0.103 machine doesn't have the right 
settings for connecting to the 192.168.0.100 machine. Is there a way 
to check this outside of running Nutch?

Thanks,

-- Ken
-- 
Ken Krugler
Krugle, Inc.
+1 530-470-9200

Re: Error running MapReduce - Host key verification failed

Posted by Gal Nitzan <gn...@usa.net>.
Hi Ken,

I am not such an expert on linux, however I have solved this problem
with ssh key based authentication.

http://www.cyberciti.biz/nixcraft/vivek/blogger/2004/05/ssh-public-key-based-authentication.php

after this operation you should be able to connect to all your nodes
from your namenode/jobtracker

Hope it helps.

G.

On Sat, 2006-01-14 at 09:33 -0800, Ken Krugler wrote:
> Hi all,
> 
> We've got the nutch-2006-01-12.tar version of Nutch, and are trying 
> to run it on three machines.
> 
> 192.168.0.100 is the "master" machine, where we run the JobTracker 
> and NameNode processes.
> 
> 192.168.0.101 and 192.168.0.103 are the "slave" machines, where we 
> rung the TaskTracker and DataNode processes.
> 
> When we fire off the the Nuch daemons with ./bin/start-all.sh, we get 
> the following error right away:
> 
> 192.168.0.103: rsync from 192.168.0.100:/home/crawler/nutch
> 192.168.0.103: Host key verification failed.
> 192.168.0.103: rsync: connection unexpectedly closed (0 bytes 
> received so far) [receiver]
> 192.168.0.103: rsync error: error in rsync protocol data stream (code 
> 12) at io.c(420)
> 192.168.0.103: starting datanode, logging to 
> /home/crawler/tmp/logs/nutch-crawler-datanode-crawlerw3.log
> 
> When I dump the datanode logfile from this 192.168.0.103 machine, I get:
> 
> 060114 082814 10 parsing file:/home/crawler/nutch/conf/nutch-default.xml
> 060114 082814 10 parsing file:/home/crawler/nutch/conf/nutch-site.xml
> 060114 082814 10 Opened server at 50010
> 060114 082814 11 Starting DataNode in: /home/crawler/tmp/ndfs/data
> 060114 082814 11 using BLOCKREPORT_INTERVAL of 3314538msec
> 060114 082814 11 Exception: java.net.ConnectException: Connection refused
> 060114 082814 11 Lost connection to namenode.  Retrying...
> 060114 082819 11 using BLOCKREPORT_INTERVAL of 3314538msec
> 060114 082819 11 Exception: java.net.ConnectException: Connection refused
> 060114 082819 11 Lost connection to namenode.  Retrying...
> 060114 082824 11 using BLOCKREPORT_INTERVAL of 3314538msec
> 060114 082824 11 Exception: java.net.ConnectException: Connection refused
> 060114 082824 11 Lost connection to namenode.  Retrying...
> 060114 082829 11 using BLOCKREPORT_INTERVAL of 3314538msec
> 060114 082829 12 Client connection to 192.168.0.100:8009: starting
> 
> When I dump the tasktracker logfile from 192.168.0.103, I get:
> 
> 060114 082832 parsing file:/home/crawler/nutch/conf/nutch-default.xml
> 060114 082832 parsing file:/home/crawler/nutch/conf/nutch-site.xml
> 060114 082832 Server listener on port 50050: starting
> 060114 082832 Server handler 0 on 50050: starting
> 060114 082832 Server handler 1 on 50050: starting
> 060114 082832 Server listener on port 50040: starting
> 060114 082832 Server handler 0 on 50040: starting
> 060114 082832 Server handler 1 on 50040: starting
> 060114 082832 Lost connection to JobTracker 
> [main1/192.168.0.100:8010]. ex=java.net.ConnectException: Connection 
> refused  Retrying...
> 
> It seems like the 192.168.0.103 machine doesn't have the right 
> settings for connecting to the 192.168.0.100 machine. Is there a way 
> to check this outside of running Nutch?
> 
> Thanks,
> 
> -- Ken