Pawn:FAQ

General

What is the purpose of <nop>PAWN?

PAWN is designed provide a platform for handling the ingest portion of digital archiving and preservation. It is a gateway between the the physical back-end archive and hundreds or thousands of eager data producers. It provides a way for an archive to allow many distributed distructful parties to prepare and transfer data for archiving in a standard way.

How is PAWN different from DSPACE/Fedora?

PAWN is not a long term archiving or content management system. It can be used to provide bulk ingestion into DSPACE and Fedora.

Where can I download PAWN?

Currently PAWN is being developed as a research prototype to explore digital accessioning. A demo of the current version is available here.

Architecture Questions

What is the difference between management server and a domain?

The management server can contain multiple domains. While domains do not trust each other, it will still be useful to have many domains aggregated onto one of more management servers to save on hardware and administrative overhead. Even though domains are housed on the same management server, they are completely independant.

Why scheduling and not load balancing?

The initial design discussions attempted to use standard web load balancing techniques to allocate data between receiving servers, however this failed for a number of reasons. First, since there is repeat access into each client access, the load balancer would have to keep state for a much longer time than normal web clients (think bank session). Second, the load balancer would have no way of knowing ahead of time how much data a client is going to push. This is opposite from most web sites where the size of outgoing traffic is fairly predictable and a load balancer can be configured as necessary. Using a scheduler solves these problems as it allows clients to request resources based on it's needs and is able to track packages over their lifetime as this information is needed to make scheduling decisions.

Why condor classads?

Condor classads are traditionally used to allocate compute resources, however in our case it provides a nice way to allow resources and requests to be flexible and still compatible for matching. As we didn't need any sophisticated backfill or other hpc type scheduling features, the classad library allows for easy on-the-fly matching of clients and resources.

Client Questions

Why can't I modify a package?

If the modify button is greyed out, the package you are trying to change has been locked. From the package list window, you must first 'unlock' the package to make changes. (your account must have permission to unlock).

What does 'manifest or children already archived, no changes allowed ...' mean?

You are trying to change part of a package that has already been archived. This could mean the item you modified is already archived, or a folder under it has already been archived. If a subdirectory / folder has been archived you can't modify anything above it w/o risking orphaning that items that have already been archived.

What does 'WSDoAllReceiver:<nop> security processing failed; nested exception is: org.apache.ws.security.WSSecurityException: An error was discovered processing the <wsse:Security> header. (WSSecurityEngine:<nop> Invalid timestamp The security semantics of message have expired') mean ?

This means that your computer and the receiving server don't have clocks that are close to the same time. Due to the security used in PAWN, there cannot be more than a 5 minute difference between any client and the archive.