Hmmm, I'd go considerably higher than 2.5G. Problem is you the Tika processing will need memory, I have no idea how much. Then you'll have a bunch of stuff for Solr to index it etc.
But I also suspect that this will be about useless to index (assuming you're talking lots of data, not say just the meta-data associated with a video or something). How do you provide a meaningful snippet of such a huge amount of data? If it *is* say a video or whatever where almost all of the data won't make it into the index anyway, you're probably better off using tika directly on the client and only sending the bits to Solr that you need in the form of a SolrInputDocument (I'm thinking that you'll be doing this in SolrJ) rather than transmit 2.5G over the network and throwing almost all of it away.... If the entire 2.5G is data to be indexed, you'll probably want to consider breaking it up into smaller chunks in order to make it useful. Best Erick On Fri, Jan 27, 2012 at 3:43 AM, Augusto Camarotti <augu...@prpb.mpf.gov.br> wrote: > I'm talking about 2 GB files. It means that I'll have to allocate something > bigger than that for the JVM? Something like 2,5 GB? > > Thanks, > > Augusto Camarotti > >>>> Erick Erickson <erickerick...@gmail.com> 1/25/2012 1:48 pm >>> > Mostly it depends on your container settings, quite often that's > where the limits are. I don't think Solr imposes any restrictions. > > What size are we talking about anyway? There are implicit > issues with how much memory parsing the file requires, but you > can allocate lots of memory to the JVM to handle that. > > Best > Erick > > On Tue, Jan 24, 2012 at 10:24 AM, Augusto Camarotti > <augu...@prpb.mpf.gov.br> wrote: >> Hi everybody >> >> Does anyone knows if there is a maximum file size that can be uploaded to >> the extractingrequesthandler via http request? >> >> Thanks in advance, >> >> Augusto Camarotti
