GPS Example Not Working

29 views
Skip to first unread message

Simone Centellegher

unread,
Oct 31, 2014, 6:40:07 AM10/31/14
to stanford...@googlegroups.com
Hi!
I'm trying to run the gps example I found in the documentation
I have followed all the documentation instructions but I can't get gps work on a small amazon ec2 cluster (1 master, 2 slaves)

when I run the example i get this!

ubuntu@ip-172-31-20-7:~/trunk/master-scripts$ ./start_gps_nodes.sh 2 PROVA1 \
> "-ifs /data/soc-Epinions1-d-n.txt \
> -hcf /usr/local/hadoop-1.2.1/conf/core-site.xml \
> -jc gps.examples.pagerank.PageRankVertex###JobConfiguration \
> -mcfg /machine-configs/test_machine_config.cfg \
> -log4jconfig ~/trunk/conf/log4j.config \
> -other -max###40"
master GPS_DIR=/home/ubuntu/trunk
starting gps master locally
1: /home/ubuntu/trunk/scripts
2: -1
3: 2
4: PROVA1
ubuntu
starting gps worker /home/ubuntu/trunk/scripts
-Xincgc -Xms50M -Xmx50M
1 172.31.20.9 2
ssh -i ~/hadoopkey.pem -o StrictHostKeyChecking=no 172.31.20.9 "/home/ubuntu/trunk/scripts/start_gps_node.sh  2 PROVA1 \"-ifs /data/soc-Epinions1-d-n.txt -hcf /usr/local/hadoop-1.2.1/conf/core-site.xml -jc gps.examples.pagerank.PageRankVertex###JobConfiguration -mcfg /machine-configs/test_machine_config.cfg -log4jconfig ~/trunk/conf/log4j.config -other -max###40\""running command: ssh -i ~/hadoopkey.pem -o StrictHostKeyChecking=no 172.31.20.9 /home/ubuntu/trunk/scripts/start_gps_node.sh /home/ubuntu/trunk/scripts 0 2 PROVA1 "-ifs /data/soc-Epinions1-d-n.txt -hcf /usr/local/hadoop-1.2.1/conf/core-site.xml -jc gps.examples.pagerank.PageRankVertex###JobConfiguration -mcfg /machine-configs/test_machine_config.cfg -log4jconfig /home/ubuntu/trunk/conf/log4j.config -other -max###40"
./start_gps_node.sh: line 30: /home/ubuntu/var/tmp/PROVA1-machine-1-output.txt: No such file or directory
1: /home/ubuntu/trunk/scripts
2: 0
3: 2
4: PROVA1
ubuntu
starting gps worker /home/ubuntu/trunk/scripts
-Xincgc -Xms256M -Xmx256M
2 172.31.20.8 2
ssh -i ~/hadoopkey.pem -o StrictHostKeyChecking=no 172.31.20.8 "/home/ubuntu/trunk/scripts/start_gps_node.sh  2 PROVA1 \"-ifs /data/soc-Epinions1-d-n.txt -hcf /usr/local/hadoop-1.2.1/conf/core-site.xml -jc gps.examples.pagerank.PageRankVertex###JobConfiguration -mcfg /machine-configs/test_machine_config.cfg -log4jconfig ~/trunk/conf/log4j.config -other -max###40\""running command: ssh -i ~/hadoopkey.pem -o StrictHostKeyChecking=no 172.31.20.8 /home/ubuntu/trunk/scripts/start_gps_node.sh /home/ubuntu/trunk/scripts 1 2 PROVA1 "-ifs /data/soc-Epinions1-d-n.txt -hcf /usr/local/hadoop-1.2.1/conf/core-site.xml -jc gps.examples.pagerank.PageRankVertex###JobConfiguration -mcfg /machine-configs/test_machine_config.cfg -log4jconfig /home/ubuntu/trunk/conf/log4j.config -other -max###40"
1: /home/ubuntu/trunk/scripts
2: 1
3: 2
4: PROVA1
ubuntu
starting gps worker /home/ubuntu/trunk/scripts
-Xincgc -Xms256M -Xmx256M

The documentation says that the results should be in the hadoop fs in /user/$USER/gps/output/ but I cannot find it! So i'm assuming that there are some error in my gps configuration.. 
can somebody halp me?
Thanks,

Simone

Semih Salihoglu

unread,
Oct 31, 2014, 4:20:31 PM10/31/14
to Simone Centellegher, stanford...@googlegroups.com
This output looks fine but is not very helpful for finding the problem I think. Can you send the output from the workers and the master? By default the log files are under /var/tmp/{execution-name}-machine*-output.txt.

Best,

semih

--
You received this message because you are subscribed to the Google Groups "stanfordgpsusers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stanfordgpsuse...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Simone Centellegher

unread,
Nov 2, 2014, 11:28:14 AM11/2/14
to stanford...@googlegroups.com, centelleg...@gmail.com, se...@stanford.edu

In the attachments there are the log files of the workers! 
I cannot find any output file neither in the master nor the workers..
it seems that there is the file ~/trunk/conf/log4j.config missing, but I have checked in both the workers and the file is exactly there!
And there are some connection problem..

This is the content of the configuration files I have edited:
in the master, the slaves file ~/trunk/master-scripts/slaves contains the ip addresses of the workers:
172.31.20.9
172.31.20.8

and the machine config file ~/trunk/test-data/test_machine_config.cfg contains
-1 172.31.20.7 1111
0 172.31.20.9 2222
1 172.31.20.8 3333
PROVA1-machine1-output.txt
PROVA1-machine1-output.txt
Reply all
Reply to author
Forward
0 new messages