We've had similar issues at Rutgers trying to ingest large objects
including video datastreams. There seem to be various problems:
system file size, apache limits., etc. We would have liked to be able
to ingest datastreams copied from the local file system without
having to pass through through apache, but this does not seem to be
possible with the current Fedora (if I'm wrong about this, I'd love
to hear about it!). Our current workaround has been to split very
large files into chunks no larger than 2GBs, which we ingest as
multiple datastreams. As these are mainly "archiival" datastreams in
our case, the fact that they need to be reasssembled has not been a
problem, bujt if you wish to serve a huge file whole from within
Fedora, our approach would not work for you. We'd be interested to
hear if anyone else has wrestled with this problem and to learn our
> Hi, everyone. I've got a question about whether it is possible to save
> large datastreams (> 5GB) in Fedora objects and, if so, what is the
> way to go about it.
> I've tried a method of copying a large archival video file to the
> running Fedora and manually creating a datastream in Fedora
> Administrator to import it, but I get a "Broken Pipe" error when
> a 9 GB datastream.
> I could obviously reference the large file outside of Fedora, but I
> would prefer to store the large file in Fedora to preserve it,
> bit-for-bit. Besides, externally referenced content is inherently
> insecure, if it is publicly available, even if you get the benefit of
> organizing the file by associating it with a Fedora object holding
> metadata about and other derivatives of that file.
> I've seen references in the list archives about folks placing files
> greater than 3 GB in Fedora, but what is the largest datastream
> by folks on this list? And if over a few GB, how did you do it?
> Thanks in advance for any suggestions.
> Stacy Pennington
> Information Technology Services
> Rhodes College