D4Science Infrastructure

I'll provide configuration information in the next days

There are 3 file to be configured through ansible scripts:

• ~/orientdb-community-2.1.13/config/orientdb-server-config.xml
• ~/orientdb-community-2.1.13/config/hazelcast.xml
• ~/orientdb-community-2.1.13/config/default-distributed-db-config.json

I propose to start in the following way:
I would like test the cluster so that adding a new server instance must not imply to restart the others (especially on production). To accomplish this we can configure hazelcast to use:

• multicast
• encryption on server synchronizations.

Due to actual sharding limitations (http://orientdb.com/docs/last/Distributed-Sharding.html#limitation) I would like to avoid it (seems that they will improve this part in the next version 2.2).

In particular:

Regarding orientdb-server-config.xml:

• it is important to force the distributed mode as explained here http://orientdb.com/docs/last/Distributed-Configuration.html#orientdbserverconfigxml

<parameter name="enabled" value="true" /> 

• You can decide if you want to store the admin password as plain text or as hash:

<user resources="*" password="7EA0EAF7FCCECA5C5B261B69CD8EC0F725FB8B3102AF51A4033CC3DED22EBD8A" name="root"/>
<user resources="*" password="PlainPassword" name="root"/>

• Please note that during the development I will have the need to modify this configuration by adding tag. An example can be:

<hooks>
   <hook class="org.gcube.informationsystem.hooks.DocumentHook" position="REGULAR"/>
   <hook class="org.gcube.informationsystem.hooks.RecordHook" position="REGULAR"/>
</hooks>

Regarding hazelcast.xml:

• please use multicast configuration http://orientdb.com/docs/last/Distributed-Architecture-Lifecycle.html#auto-discovering
• Use security (you can select PBEWithMD5AndDES or what you think is more secure) can find detailed information at the following links: http://orientdb.com/docs/last/Distributed-Architecture-Lifecycle.html#security

Regarding default-distributed-db-config.json please remove clusters property.

More details can be found here:
http://orientdb.com/docs/last/Distributed-Configuration.html

Feel free to contact me for any doubt.

orientdb01-d-d4s.d4science.org 146.48.122.23
orientdb02-d-d4s.d4science.org 146.48.122.14
orientdb03-d-d4s.d4science.org 146.48.122.148

No ansible at the moment for the orientdb/hazelcast configuration, it will be done later, please configure the cluster by hand if you want to test it.

You can connect via ssh as orientdb user and manage orientdb service with
/home/orientdb/orientdb start/stop/status

Ok. I'll take care of configuring the cluster.

@tommaso.piccioli@isti.cnr.it I need your support to correctly configure hazelcast.

Can you configure hazelcast with static hostnames? We will try multicast in the future.

We need to setup OrientDB cluster with the 3 VMs available

I'm assembling a oriendb ansible role to provision all the pieces.

The playbook is ready and the configuration seems corret. We have errors on the hooks, because we didn't create the rules that install the needed jars. I need directions for those.

Done. The hook has been installed, the configuration updated, the services restarted.