arpa2.reservoir 0.1.0

ARPA2 Reservoir -- object store with metadata in LDAP

ARPA2 Reservoir: API

This is the central logic for manipulating the stored information of ARPA2 Reservoir. It manipulates metadata in LDAP as well as the files that hold the litaral objects.

Reservoir is an object store with maximum exposure over standard protocols. It stores metadata in LDAP, which may be searched from the outside. The actual data is stored in a hierarchy of a given, simple structure. This API controls both.

To access the API, import its package:

from arpa2 import reservoir

This only includes the API. You will separately have to add support from other tools if you need it. Some tools are in fact commandline utilities that get installed through their entry_point as additional commands for the OS.

Layers on Top of this API

On top of this API, there are a few other packages:

Work in progress:

• arpa2reservoir is a commandline shell over which administrators can perform the actions offered by the API. It also has an automation interface centered around JSON messages.

• wsgidav is a WebDAV implementation with a backend plugin for the Reservoir. It digs up metadata in LDAP and serves the results over a common WebDAV interface. Even though LDAP would be more efficient to use, there are many ways of making a WebDAV repository available as a virtual drive or directory on a client machine.

• reservoir.sftp can be a plugin to OpenSSH2 that serves the SFTP protocol, but in reality interacts with the Reservoir instead of (just) local files.

• reservoir.rmt solutions may use the Reservoir like a remote tape for backup. It may implement the rmt(1) command set for anyone who arrives over a secure shell connection.

Further options for which your help could make all the difference:

• reservoir.amqp servers will send and receive from queues in the Reservoir, by connecting to other domains with AMQP 1.0, performing mutual authentication and then passing objects with as much metadata as the protocol can carry.

• reservoir.lmtp servers may receive information for Reservoir from a variety of sources, including but not limited to a domain's mail server.

• reservoir.volume may be a service to provide images for a large variety of purposes; these may include disk images for virtual machines, CD-ROM images for OS installation, TFTP images of operating systems and their initial disks, and so on. Note that Reservoir can store hashes of content in the metadata.

Functionalities

The Python documentation will help you get going, but this is a quick start.

There are a few classes of interest:

• Resource is like a file; it represents a stored object, including its metadata.
• Index is like a directory; it can progress to other indexes, as well as list or obtain resources held in them.
• The domain (or a user under a domain) have what is called a home index; this works a little differently; the first step is to provide a hinted application name that may have its own Index setup; if not, a default/unnamed Index that serves as a fallback is used instead.

An Index points to a Collection of Resources. You can access a name in the index to move to another such Collection. New Collections may be added with a name, existing ones may be aliased to a name, and this is how Indexes grow and expand.

Finally, Collections have Resources added to them. Indexes that are home indexes for (a user under) a domain are not Collections, and they cannot have Resources underneath them. This can be remedied simply with an application hint.