Erik Bergenholtz created AMBARI-11210:
-----------------------------------------

             Summary: Topology information in Flume configuration should be 
processed automatically
                 Key: AMBARI-11210
                 URL: https://issues.apache.org/jira/browse/AMBARI-11210
             Project: Ambari
          Issue Type: Bug
          Components: ambari-server
    Affects Versions: 2.1.0
            Reporter: Erik Bergenholtz
            Assignee: Emil Anca
             Fix For: 2.1.0


In the flume-conf ambari configuration I see that the "content" property 
contains topology related information but isn't being handled automatically as 
part of the topology fix-up phase of blueprint cluster creation.  Currently, 
this is accomplished by registering the appropriate updater for the 
configuration property in BlueprintConfigurationProcessor.  As part of the next 
wave of stack enhancements, this would be moved out of the ambari code and into 
the stack definition.

Below is an example value for flume-conf/content.  Notice the references to the 
concrete host name "john.novalocal" in the property value.  The user shouldn't 
have to specify this information manually in a config override.  In 2.1 it will 
be possible to provision a cluster via the api by specifying host counts 
instead of host names in the cluster creation template so it won't even be 
possible to provide host names in a config override.
{code}
"content" : "# Name the components on this agent\na1.sources = src1\na1.sinks = 
sink1\na1.channels = c1\n\n# Use a channel which buffers events in 
memory\na1.channels.c1.type = memory\na1.channels.c1.capacity = 
10000\na1.channels.c1.transactionCapacity = 100\n\n# Configure the source\n# 
a1.sources.src1.type = exec\n# a1.sources.src1.command =  tail -f 
/var/log/ambari-a1/ambari-a1.log\na1.sources.src1.channels = 
c1\na1.sources.src1.type = multiport_syslogtcp\na1.sources.src1.host = 
0.0.0.0\na1.sources.src1.ports = 8004\n\n# Configure the 
sink\na1.sinks.sink1.type = hdfs\na1.sinks.sink1.hdfs.path = 
hdfs://john.novalocal:8020/tmp/logs/\na1.sinks.sink1.hdfs.filePrefix = 
events-\na1.sinks.sink1.hdfs.round = true\na1.sinks.sink1.hdfs.roundValue = 
10\na1.sinks.sink1.hdfs.roundUnit = minute\n\n#a1.sinks.sink1.type = 
logger\na1.sinks.sink1.channel = c1\n\n# Name the components on this 
agent\na2.sources = src2\na2.sinks = sink2\na2.channels = c2\n\n# Use a channel 
which buffers events in memory\na2.channels.c2.type = 
memory\na2.channels.c2.capacity = 10000\na2.channels.c2.transactionCapacity = 
100\n\n# Configure the source\n# a2.sources.src2.type = exec\n# 
a2.sources.src2.command =  tail -f 
/var/log/ambari-a2/ambari-a2.log\na2.sources.src2.channels = 
c2\na2.sources.src2.type = multiport_syslogtcp\na2.sources.src2.host = 
0.0.0.0\na2.sources.src2.ports = 8003\n\n# Configure the 
sink\na2.sinks.sink2.type = hdfs\na2.sinks.sink2.hdfs.path = 
hdfs://john.novalocal:8020/tmp/logs/\na2.sinks.sink2.hdfs.filePrefix = 
events-\na2.sinks.sink2.hdfs.round = true\na2.sinks.sink2.hdfs.roundValue = 
10\na2.sinks.sink2.hdfs.roundUnit = minute\n\n#a2.sinks.sink2.type = 
logger\na2.sinks.sink2.channel = c2"
{code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to