cloudgene server
¶
Starts the Cloudgene web-application which provides an user-interface to submit jobs for all installed workflows.
Command¶
cloudgene server [--conf <hadoop_conf_dir>] [--user <hadoop_username>] [--port <port>]
Parameters¶
Parameter | Required | Description |
---|---|---|
--conf <hadoop_conf_dir> |
no | Path to Hadoop configuration folder (e.g. /etc/hadoop/conf )) |
--user <hadoop_username> |
no | Execute Hadoop steps on behalf of this username (default: cloudgene) |
--port <port> |
no | Start the web-application on this port (default: 8082) |
Examples¶
Start server¶
cloudgene server
The webservice is available on http://localhost:8082. Please use username admin
and password admin1978
to login. You can use the --port
flag to change the port from 8082
to 8085
:
cloudgene server --port 8085
Start server and connect it with a Hadoop Cluster¶
If Cloudgene is installed on the Hadoop Namenode, then you can start the server with the conf files located in $HADOOP/conf
:
cloudgene server --conf /etc/hadoop/conf
However, if Cloudgene is not installed on the Hadoop Namenode, then you have to copy the config files from the Hadoop Namenode to your Cloudgene instance and start the server with this configurations:
cloudgene server --conf $CLOUDGENE_HOME/hadoop-conf
Cloudgene executes your MapReduce steps on the remote cluster. You can use the --user
flag to set the username which should be used to execute your Hadoop jobs (e.g. it uses the HDFS directory of this user for all files).
For production we recommend to add your cluster to the configuration file to avoid to set the parameters on each startup. More about adding the cluster to the configuration file can be found here.