I have been unable to enable the harvester. Basically the issue is that the /oaiprovider web app periodically polls against the /fedora/risearch triplestore, but instead of retrieving an ITQL response with the fedora objects for the harvester to mantain, this query causes a java.lang.OutOfMemoryError in the /fedora/risearch.
The query resulting in the error is this one:
select $item $itemID $date $state
where $item <http://www.openarchives.org/OAI/2.0/itemID> $itemID
and $item <info:fedora/fedora-system:def/model#state> $state
and $item <info:fedora/fedora-system:def/view#disseminates> $diss
and $diss <info:fedora/fedora-system:def/view#disseminationType> <info:fedora/*/DC>
and $item <info:fedora/fedora-system:def/view#lastModifiedDate> $date
and $date <http://mulgara.org/mulgara#after> '2010-05-12T19:37:55.289Z'^^<http://www.w3.org/2001/XMLSchema#dateTime> in <#xsd>
and $date <http://mulgara.org/mulgara#before> '2010-05-13T13:07:04.746Z'^^<http://www.w3.org/2001/XMLSchema#dateTime> in <#xsd>
order by $itemID asc
The first solution tried was to add as much memory to the jvm as possible:
-Xms2056m -Xmx2056m -XX:PermSize=256m -XX:MaxPermSize=256m
This worked once, then failed every time after in regular polling with the same java.OutOfMemoryError
Workaround: Deploy remote Mulgara, configure it as backend to fedora/risearch - hopefully the memory allocations will work having the remoteMulgara running on a separate jvm. Tried 2 ways, 1) by loading the remote Mulgara via the fedora-rebuild RI option, and 2) by attempting to load mulgara directly with a export of the original local Mulgara. The first successfully rebuilt the triplestore, but then failed with an RMI error due to a broken pipe. The second successfully imported into the remote Mulgara the export of the local Mulgara, but failed when attempting the above query with an "#xsd is not absolute" error not even after adding it as an alias.
My complete notes with detailed instructions and stack traces are attached. If anyone has insight into this problem, suggestions are very welcome.
From: Deborah Kaplan [deborah.kaplan@...]
Sent: Tuesday, June 29, 2010 1:20 PM
Subject: [Fedora-commons-users] OAI-PMH work?
I know that a lot of comments have gone by about people looking into various problems with the OAI-PMH provider built into Fedora. Is anybody working on these? We were thinking about putting some cycles into investigating and fixing many of the problems, but we don't want to reproduce ongoing efforts.
The key problems we're investigating are why the provider isn't usable by OAISTER, but I know there have been other concerns with the provider.
Digital Resources Archivist
Digital Collections and Archives
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
Fedora-commons-users mailing list