PLoS Journal Websites – Upgrade to Topaz 0.9.1
At 5pm PST, we will update the PLoS journal websites to Topaz 0.9.1.. The journal websites will be offline for approximately one hour. Once the upgrade is complete, the websites will be a bit slow for the first couple of hours while the caches re-fill. We will run scripts before go-live to re-fill the caches for articles linked from the homepage and the current issue, so the most recent articles should display quickly to the end user.
The PLoS development team used Amazon EC2 extensively for performance tests and for benchmark tests vs. Topaz 0.9.0. We’ve also finished the herculean task of ingesting 6372 PLoS Medicine and PLoS Biology corpus articles required for the migration of those journals to Ambra/Topaz in the next few months.
Features implemented in Topaz 0.9.1 include:
- Revamp of Ambra search to use Mulgara (does not include update to the search UI)
- Support of article packages greater than 4Gb in size
- Support of article packages in tar format
- Support for very large assets (e.g. supporting info files) outside of memory
- Fixes to formal corrections
- Fixes to citations
- Fixes to citation downloads
- Article-level feeds for top-level notes/comments on an article
- Updates to admin panel for administering notes/comments
- Many fixes to the UI including line wrap for long gene sequences, browser-specific bugs, wrapping long URLs in annotations/articles, etc.
- Cleanup of content models, back-end bug fixes, etc.
It seems that the new search integration with Mulgara can cause a performance slowdown with many queries. We have disabled search until the developers have a chance to look into the problem.
We’ve enabled simple search for all the journal websites. Currently, this will only search the title and the abstract text of articles. Searches on author names and/or full body text is still slow due to the large amount of data in those fields. We have disabled advanced search until we can either update the advanced search form to exclude author names and full body text and/or Topaz can fix the Mulgara queries.