incubator-chukwa-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From DKN <devapra...@in.ibm.com>
Subject Re: A demo setup on a single linux server
Date Thu, 02 Jun 2011 17:36:50 GMT
Thanks Eric. I can now see all the pieces tie together .. (at least to the
extent that I thought in 3 days .. ). I have raised the bug regarding hicc
binding to localhost.

A few more questions to wrap this up.

1. Nothing is written to Hadoop table in HBase. Is there something that I am
missing ? I understand that ClusterSummary is populated ONLY after the pig
job is successfully run (so, not seeing anything in ClusterSummary table is
expected.

2. If I have only HbaseWriter for the collector, and if I have configured an
agent to monitor a file, what will be written to HBase ? Is it possible to
make use of HBase for custom data types like monitoring a custom log file
(like /var/log/messages ) and provide visualization for such a monitoring ?

3. What is the difference between what is written to HBase versus what is
written to HDFS ? I was wondering if I can have just one write to HBase
(which intern can be configured to use HDFS as the persistent store .. ) and
omit the SeqFileWriter ?

4. I observed a strange behavior in agents. I modified the initial_adaptors
file to have only one agent to monitor a file. When I restarted the agent
process and did a "list" [using telnet localhost 9093], I could see all the
older agents running. Could this be a bug.

Thanks and regards, DKN

--
View this message in context: http://apache-chukwa.679492.n3.nabble.com/A-demo-setup-on-a-single-linux-server-tp3001627p3016124.html
Sent from the Chukwa - Users mailing list archive at Nabble.com.

Mime
View raw message