Hello, I use the static linux build of sedna 3.2.91 on linux AMD64.
I have a 600MB XML file (dblp.xml taken from: http://dblp.uni-trier.de/xml/). I create a 'dblp' database with all default settings. Then I load the dblp.xml file to a 'dblp' collection. It loads OK the sedata file is about 3G large. Then I use the Java XQJ driver to execute a query. I use the writeSequenceToSAX method on the result sequence. It is then processed by my SAX handler. The query itself is attached below. The query is executed correctly and returns a couple of hundred thousand results (<resource> tags). The problem is that the temp file rises to 18GB until it fills the whole disk. Sedna writes some error in the log about not being able to extend the temp file "cannot write the block" or something similar. Then it breaks, all processes are killed (both the trn and the sm ones). XQJ throws an exception. Therefore some questions. 1. Is the 18GB temp file to be expected on such a query for such a dataset? Maybe I should live with it and get a bigger disk. 2. If I limit the temp file size with the -tmp-file-max-size option will sedna deal with a limited temp file and bring the query to an end or will it simply break earlier when the temp file hits the limit? 3. What is the correct way to specify the -tmp-file-max-size option. I tried it a couple of time and always get se_cdb -tmp-file-init-size 100 -tmp-file-ext-portion 100 -tmp-file-max-size 200 dblp Creating a data base (it can take a few minutes)... SEDNA Message: ERROR SE1012 Temporary file has reached its maximum size. The same with following settings (init-size,ext-portion,max-size) (100,1000,1000) (100,1500,100) (100,100,1000) Basically anything apart from 0 in the -tmp-file-max-size throws the SE1012 error. All kinds of comments welcome. I can provide any additional information if needed. Antoni Mylka [email protected] The quuery is meant to extract information from a given file in a given collection. for $x at $pos in collection('dblp')[compare(document-uri(.),'dblp.xml') = 0]/./dblp/* return <resource> <d1>{for $d1 in $x/data(@key) return <s>{$d1}</s>}</d1> <d2>{for $d2 in $x/title/text() return <s>{$d2}</s>}</d2> <d3>{for $d3 in $x/source/text() return <s>{$d3}</s>}</d3> <d4>{for $d4 in $x/year/text() return <s>{$d4}</s>}</d4> <d5>{for $d5 in $x/series/text() return <s>{$d5}</s>}</d5> <d6>{for $d6 in $x/volume/text() return <s>{$d6}</s>}</d6> <d7>{for $d7 in $x/number/text() return <s>{$d7}</s>}</d7> <d8>{for $d8 in $x/month/text() return <s>{$d8}</s>}</d8> <d9>{for $d9 in $x/pages/text() return <s>{$d9}</s>}</d9> <d10>{for $d10 in $x/url[starts-with(text(),'http://')]/text() return <s>{$d10}</s>} </d10> <d11>{for $d11 in $x/url[not(starts-with(text(),'http://'))]/text() return <s>{$d11}</s>} </d11> <d12>{for $d12 in $x/ee[starts-with(text(),'http://')]/text() return <s>{$d12}</s>} </d12> <d13>{for $d13 in $x/ee[not(starts-with(text(),'http://'))]/text() return <s>{$d13}</s>} </d13> <d14>{for $d14 in $x/publisher/text() return <s>{$d14}</s>}</d14> <d15>{for $d15 in $x/isbn/text() return <s>{$d15}</s>}</d15> <d16>{for $d16 in $x/crossref/text() return <s>{$d16}</s>}</d16> <d17>{for $d17 in $x/cite[not(starts-with(text(),'.'))]/text() return <s>{$d17}</s>} </d17> <d18>{for $d18 in $x/booktitle/text() return <s>{$d18}</s>}</d18> <d19>{for $d19 in $x/author/text() return <s>{$d19}</s>}</d19> <d20>{for $d20 in $x/editor/text() return <s>{$d20}</s>}</d20> <d21>{for $d21 in $x/titleSignature/text() return <s>{$d21}</s>}</d21> <d22>{for $d22 in $x/mdate/text() return <s>{$d22}</s>}</d22> <p>{$pos}</p> </resource> ------------------------------------------------------------------------------ This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev _______________________________________________ Sedna-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sedna-discussion
