Hi Erlend, I hope you are not seeing memory issues on large files with ManifoldCF itself. That should not happen, and if it does we need to figure out why.
Solr memory issues, on the other hand, I can believe. If that is the problem, then I agree we should try to do something about it. Probably the right thing to do is (since it is a Solr limitation) adding a configuration parameter to the Solr connector that specifies the maximum size of a file the connection will accept. Files larger than that should return a 400 if indexing is attempted, etc. Perhaps we should also consider adding a new method to the IOutputConnector interface that returns a maximum file size value, and expose that in IVersionActivity and IProcessActivity. That would allow connectors to make output-based decisions as to whether they should fetch large files in the first place. Karl On Thu, Jun 23, 2011 at 7:32 AM, Erlend Garåsen <[email protected]> wrote: > > I will create a ticket today. Post filtering sounds like a good idea. > > Another thing. We are facing memory problems with huge documents. Maybe we > should add another future in order to cope with such documents, for instance > skip documents which exceed a preset size. We have discovered pdfs on 500 > MB. What do you think? Do we need such a future as well? > > Erlend > > On 23.06.11 12.08, Karl Wright wrote: >> >> Have there been any further developments on this thread? >> Karl >> >> On Tue, Jun 21, 2011 at 6:08 AM, Karl Wright<[email protected]> wrote: >>> >>> Sure. But you've already convinced me we need a new feature. ;-) >>> >>> Karl >>> >>> On Tue, Jun 21, 2011 at 3:50 AM, Erlend Garåsen<[email protected]> >>> wrote: >>>> >>>> Sure, I can create a ticket. But first I want to discuss this issue with >>>> the >>>> two search consultants we have hired. >>>> >>>> I decided to post to the dev list in order to get some feedback on this >>>> issue. >>>> >>>> Erlend >>>> >>>> On 20.06.11 18.00, Karl Wright wrote: >>>>> >>>>> Hi Erlend, >>>>> >>>>> The inclusions and exclusions are based solely on URL, and block the >>>>> connector from fetching the file. Otherwise you would easily wind up >>>>> fetching the entire web. >>>>> >>>>> However, this raises an interesting issue as to whether there's a way >>>>> in the web connector to do what you are trying to do, which is to >>>>> filter based on URL after links have been extracted. The current >>>>> inclusions/exclusions work fine for any URLs without links but do not >>>>> allow for the case you are looking for. >>>>> >>>>> Can you create a ticket? The suggestion would be to introduce >>>>> post-extraction inclusions and exclusions into the connector. >>>>> >>>>> Karl >>>>> >>>>> >>>>> On Mon, Jun 20, 2011 at 10:53 AM, Erlend Garåsen >>>>> <[email protected]> wrote: >>>>>> >>>>>> I just realized that if I exclude html files for a job, links in these >>>>>> files >>>>>> will not be followed. Is this a desirable behaviour? Should links be >>>>>> followed regardless of the exclude filter? >>>>>> >>>>>> I discovered this issue when I was going to crawl only pdfs and >>>>>> realized >>>>>> that the job ended without finding any documents at all. I think I had >>>>>> something like this in my include list: >>>>>> http://foreninger.uio.no/.*\.pdf$ >>>>>> http://folk.uio.no/.*\.pdf$ >>>>>> >>>>>> Erlend >>>>>> >>>>>> -- >>>>>> Erlend Garåsen >>>>>> Center for Information Technology Services >>>>>> University of Oslo >>>>>> P.O. Box 1086 Blindern, N-0317 OSLO, Norway >>>>>> Ph: (+47) 22840193, Fax: (+47) 22852970, Mobile: (+47) 91380968, VIP: >>>>>> 31050 >>>>>> >>>> >>>> >>>> -- >>>> Erlend Garåsen >>>> Center for Information Technology Services >>>> University of Oslo >>>> P.O. Box 1086 Blindern, N-0317 OSLO, Norway >>>> Ph: (+47) 22840193, Fax: (+47) 22852970, Mobile: (+47) 91380968, VIP: >>>> 31050 >>>> >>> > > > -- > Erlend Garåsen > Center for Information Technology Services > University of Oslo > P.O. Box 1086 Blindern, N-0317 OSLO, Norway > Ph: (+47) 22840193, Fax: (+47) 22852970, Mobile: (+47) 91380968, VIP: 31050 >
