howdy!
if article delete via admin hits an error, it seems to leave the article in an inconsistent state, requiring manual delete from mulgara (via runitql) and fedora (via fedora-purge)
specifically, multiple article state triples are entered in mulgara for the article and it's assets - it is marked as both published and unpublished at the same time.
on future attempts to delete or ingest, we get an error like this:
Error ingesting: pone.0000000.zip - org.topazproject.otm.OtmException: Too many values for 'private java.lang.String org.plos.models.DublinCore.description'
i suspect that something is failing to rollback.
the kinds of errors i'm seeing are permissions based - the file in the ingested queue is not writeable by topaz, or the related crossref XML file is missing and cannot be deleted.
it would also be nice to get an error on the admin page if delete runs into trouble - production staff is reporting that they see no errors when this happens.
thanks!
