Platform Team/Sugar Network/Implementation

From Sugar Labs
Jump to navigation Jump to search

Sugar Network version cycle: | 1.0 |

Summary

The requirements:

  • Implement the initial server in a short period of time;
  • Server should work on XO laptops;
  • System should support synchronisation between distributed servers and provide full featured text search.

That's why using Active Document to:

  • Server should be as simple/lightweight as possible;
  • No SQL, just NoSQL;
  • Since server should support full featured text search, use Xapian and keep it as a "NoSQL replacement" with storing data directly in files system;
  • If server will start serving users from the Internet and current implementation won't manage to handle multiple requests, another implementation might be created.

All resources are being represented as Active Document's classes, resource objects as Active Document's objects.

Usage

Public instance

Testing instance is api.network.sugarlabs.org. It support HTTPS connections using StartSSL certificate.

Add the 18.85.44.120 api.network.sugarlabs.org IP to /etc/hosts file and use http://api.network.sugarlabs.org:8000 as connection url.

Local instance

Attach Server:Factory repository. For Ubuntu repositories, download GPG key:

wget -qO- <REPO-URL>/Release.key | sudo apt-key add -

And install sugar-network-server package.

To see current configuration:

sugar-network-server config

Default configuration is (tweak /etc/sugar-network-server.conf file to change it):

  • server listens 0.0.0.0:8000 address;
  • database in /var/lib/sugar-network directory;
  • log directory in /var/log/sugar-network.

To start server:

/etc/init.d/sugar-network-server start

To simplify testing, untar example db data to /var/lib/sugar-network and restart server (it will take some time to create indexes).

Features

Collecting usage statics

Server will accept usage statistics from clients (and, for master server, from nodes) and will keep them in --stats-root directory.

Todo

  • Network notifications framework to, e.g., let client cache resources and refetch them only on updates. Reuse existing or implement something similar to Facebook's real-time updates.
  • Support collaborative work on Wiki pages for different resources.

Getting involved