View Source

Here is a sample process of how to install and configure a 2DC (2 data-centers, each with 2 masters and 2 workers) and nagios for monitoring.

h2. Launch EC2 Instances

# Choose "onto-docker-X" instances (Tested with onto-docker-9).
!EC2 Management Console - Chromium_005.png|thumbnail,border=1,width=100!
# Choose the appropriate instance type (e.g. m3.medium).
!EC2 Management Console - Chromium_007.png|thumbnail,border=1,width=100!
# Open the instances for all communication.
!EC2 Management Console - Chromium_008.png|thumbnail,border=1,width=100!
# Select Number of instances: 8 in Configure Instance Details step.
!EC2 Management Console - Chromium_009.png|thumbnail,border=1,width=100!
# Launch\!

When the instances start, they pull the latest GraphDB docker image, along with LDBC 50 and start it (takes a few minutes).

h2. Configuration

Each instance can act either as a master or as a worker. For this scenario, configure them by first editing the following configuration file: [Test install of GraphDB on EC2^test.config].

Save the file either in the current folder or in the user home folder as "test.config". Just change the IP addresses for each box to one of the public IPs of the newly created instances:

{code}
# name/IP or hostname/OS user/contol JMX PORT

box1 54.72.215.230 - - 9999
box2 54.72.236.150 - - 9999
box3 54.76.125.3   - - 9999
box4 54.76.154.205 - - 9999
box5 54.76.156.52  - - 9999
box6 54.76.156.53  - - 9999
box7 54.76.156.72  - - 9999
box8 54.76.156.78  - - 9999
{code}

Be sure that the Amazon "Security groups" and instances firewalls have proper open ports.
The following assumes that you have installed (unzipped) the new-cluster-test project, provided by Ontotext as a separate ZIP file:
{code}
$ cd ~/new-cluster-test
$ ant amazon-configure
{code}
This process takes awhile, so you can now start deploying the nagios instance.

h2. Launch nagios instance

You need one more EC2 instance. This time choose 64-bit Amazon Linux AMI (tested again with t3.medium, though t1.micro may be enough). You just need SSH (22) and HTTP (80) ports open for this one.

After the instance is up, you need to SSH to it and run the following commands there:

{code}
$ sudo yum install docker
$ sudo service docker start
$ sudo docker login -u ontotext -p ontodocker -e zdravko.tashev@ontotext.com
$ sudo docker pull ontotext/owlim-nagios
{code}
Now you are ready to configure and run nagios. First make sure that the cluster is configured and running. You also need one of the masters' URLs. E.g. the master in box1 above: [http://54.72.215.230:8080/openrdf-workbench/repositories/master].



{code}
$ sudo docker run -d -p 80:80 ontotext/owlim-nagios conf-and-run http://54.72.215.230:8080/openrdf-workbench/repositories/master
{code}That's it\! The start script will read the cluster's configuration, generate one for nagios and run it. To view the results point your browser to [http://54.76.155.44/nagios3/] or [http://54.76.155.44/pnp4nagios] for graphs (you'll have to put the external IP of the nagios machine here). The credentials are nagiosadmin/nagio.


h2. Run sample LDBC-based load test
This is run from the same project new-cluster-test:
- check the settings in your test.config, the following parameters at the end are related to the load test:
{code}
ldbc_aggregation_agents 16
ldbc_editorial_agents 2
ldbc_run_time 3600
{code}

- then start the tests
{code}
ant ldbc-run
{code}


* Note: ldbc_run_time:
** LDBC benchmark required java 7;
** The ldbc-run ant target depends on *Unix.