The bitrepository for newspapers will consist of:
- A tape based pillar, managed by JHLJ. The pillar's cache will act as processing area the jpeg2000 datafiles.
- A tape based pillar, managed by LFM.
- A checksum pillar, managed by ???
Data processing
As part of the general architecture files are processed when they have been ingested in the repositories. I.e. Bitrepository for datafiles (jpeg2000) and DOMS for metadata.
A tape based pillar normally gives no guarantees of when files are online, which poses a problem when using it for processing. To solve this problem the tape backend api (layer 2, python) is being extended with additional methods so that:
- We can control which files are kept in cache while processing.
- We can minimize the number of files that has been rejected that ends up on tape storage.
The api methods for the tape backend is:
- force-online <prefix>
- Method to keep files with <prefix> online. The method should be called prior to ingesting any files with the prefix, to prevent them from being rolled out on tape.
- A call to the method may fail if the qouta for online files has been exceeded.
- In the future, a call to this method with a prefix for which files already is on tape may bring the files online.
- release-online <prefix>
- Method to release files with a prefix to be rolled out on tape.
- status-online
- Method to list which prefixes are kept online.
- May in the future provide more information of the online/offline state.