Schemas, schemas everywhere

iRODS 4.x development is moving forward quickly.

One of the major things we’ll be adding soon is better introspection… iRODS needs much better capabilities with regard to reporting on what is installed, what is configured, and its current status.

A bit of this functionality will be made possible by a standardization of what an iRODS Zone is defined to be [http://schemas.irods.org/]. We have developed a first draft of a version 1 schema that defines an iRODS Zone, and all its constituent parts.

This schema serves two main purposes in the short term:

  1. We can use it internally to deploy and test topologically (meaning, multiple hosts). Our Continuous Integration server will be able to generate and dynamically deploy an iRODS zone_bundle that can be put through the full test suite.
  2. We can use it externally by having Zones deployed by others issue a “Zone Report” in the form of a JSON file. This file can be sent to us for debugging, analysis, and/or redeployment for reproducibility.

UPDATE 2014/12/04: This has been renamed to zone_bundle – The links below have been updated, but the displayed code remains the original (and now outdated).

The top level (zone_bundle.json) that defines a set of iRODS Zones:

A Zone contains an iCAT Server and zero or more Resource Servers:

  • icat_server.json – Defines a Resource Server with additional database connection information
  • resource_server.json – Defines an iRODS Server, configuration, and plugins

An iCAT Server (icat_server.json):

A Resource Server (resource_server.json):

The database information (host, user, and password) will be stored in database_config.json:

iRODS configuration files will (soon) themselves be JSON and will be included directly in the zone_bundle schema:

  • server_config.json
  • host_access_control_config.json
  • hosts_config.json
  • service_account_environment.json

The rest of the files in the /etc/irods configuration directory will be represented in configuration_directory.json:

A plugin will be represented in plugin.json:

And every resource in the Zone will be listed as an instance of resource.json:

A sample zone_bundle file can be found in the draft v1 directory listing.

We are using nodejs to validate zone_bundle files against this schema. This will allow us to make more assumptions in our code as we know that a set of inputs are good before we start to process them (deploy a cloud infrastructure, debug a tricky edge case, etc.).

We are actively developing this schema and welcome any input.

Next:

Previous: