Dear Hugh, many thanks for your reply! Sorry, I forgot to mentioned that I increased up to 24 core and 228 GB memory.
I'm still on trying to optimize the performance and to avoid transaction deadlocks... but unfortunately without any success so far... I've already tried to create some additional indexes, like PSOG, but the evaluation of some filter parts take in some cases really long (more than 40 minutes)... For me it is also very strange why the evaluation of query define input:default-graph-uri <http://clashsniffer.hs-albsig.de/benchset> PREFIX xsd: <http://www.w3.org/2001/XMLSchema#> PREFIX owl: <http://www.w3.org/2002/07/owl#> PREFIX quest: <http://obda.org/quest#> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> SELECT ?x ?b WHERE { ?x <http://purl.org/ontology/bibo/volume> ?b . FILTER ( ?x >= <http://eprints.wmin.ac.uk/id/eprint/1620> ) . FILTER ( ?x < <http://pub.epsilon.slu.se/id/eprint/11557> ) . } at instance running at port 8895 will take only some ms, but executing query define input:default-graph-uri <http://clashsniffer.hs-albsig.de/benchset> PREFIX xsd: <http://www.w3.org/2001/XMLSchema#> PREFIX owl: <http://www.w3.org/2002/07/owl#> PREFIX quest: <http://obda.org/quest#> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> SELECT ?x ?b WHERE { SERVICE <http://141.87.4.8:8895/sparql> { ?x <http://purl.org/ontology/bibo/volume> ?b . FILTER ( ?x >= <http://eprints.wmin.ac.uk/id/eprint/1620> ) . FILTER ( ?x < <http://pub.epsilon.slu.se/id/eprint/11557> ) . } . } at another instance is running possibly indefinitely... (stopped the evaluation after 45 minutes). Best Andy -----Ursprüngliche Nachricht----- Von: Hugh Williams [mailto:hwilli...@openlinksw.com] Gesendet: Sonntag, 28. Februar 2016 04:14 An: Nolle, Andreas <no...@hs-albsig.de> Cc: virtuoso-users@lists.sourceforge.net Betreff: Re: [Virtuoso-users] infrequent errors on parallel querying Hi Andreas, In you INI file “MaxClientConnections” and “ServerThreads” are the same thing the latter being the original name which is retained for compatibility reasons. Why do you set these params to 1000 on both sections of the INI file ? I would reduce them to some thing like 200 as especially the HTTP Server ones are pre-allocated on server startup which will consume significant server resources for these threads. Looking at the Vectored execution INI file params you have set: MaxQueryMem = 32G ; memory allocated to query processor VectorSize = 1000 ; initial parallel query vector (array of query operations) size MaxVectorSize = 1000000 ; query vector size threshold. AdjustVectorSize = 1 ThreadsPerQuery = 24 Why is MaxQueryMem set so high ie 32G as this will all consume significant system memory when multiple queries are being run, thus I would set it to something like the 2G default to start off with. How much memory is available on the system ? AdjustVectorSize should be set to 0 in most use cases … I assume you machine has 24 cores which is why you set "ThreadsPerQuery = 24” ? See the following documentation on vectored query execution: http://docs.openlinksw.com/virtuoso/databaseadmsrv.html#confvectexec - Configuring Vectored Execution http://docs.openlinksw.com/virtuoso/databaseadmsrv.html#tunparamsmworkload - Tuning Parameters for Multiuser Workloads and the following on Transaction deadlocks and how to determine and avoid them: http://docs.openlinksw.com/virtuoso/databaseadmsrv.html#perfdiag - Performance diagnostics http://docs.openlinksw.com/virtuoso/databaseadmsrv.html#TRANSACTION_ISOLATION_LEVELS - Transaction Metrics, Diagnostics and Optimization Best Regards Hugh Williams Professional Services OpenLink Software, Inc. // http://www.openlinksw.com/ Weblog -- http://www.openlinksw.com/blogs/ LinkedIn -- http://www.linkedin.com/company/openlink-software/ Twitter -- http://twitter.com/OpenLink Google+ -- http://plus.google.com/100570109519069333827/ Facebook -- http://www.facebook.com/OpenLinkSoftware Universal Data Access, Integration, and Management Technology Providers > On 26 Feb 2016, at 08:03, Nolle, Andreas <no...@hs-albsig.de> wrote: > > The critical error is now that after some query evaluations I get a lot of > "Virtuoso 40001 Error SR172: Transaction deadlocked" errors. For me this is > really strange because I've set the DefaultIsolation = 2 and only SELECT > queries are executed. Please notice that I've slightly changed some > parameters in the ini files. > > It would be really nice if you could help me on that and give me some > suggestions how transaction errors can be avoided on evaluating SELECT > queries. Please find the current log and ini files of each Virtuoso instance > as well as the corresponding output of status() at > https://www.dropbox.com/s/v6hvw7j7uz9r9uq/virtuoso_files.zip?dl=0 > > Best > Andy > > > -----Ursprüngliche Nachricht----- > Von: Nolle, Andreas [mailto:no...@hs-albsig.de] > Gesendet: Freitag, 26. Februar 2016 09:00 > An: Hugh Williams <hwilli...@openlinksw.com> > Cc: virtuoso-users@lists.sourceforge.net > Betreff: Re: [Virtuoso-users] infrequent errors on parallel querying > > Hi Hugh, > > in the meantime I solved some of the errors. Assuming the following query is > evaluated at endpoint running at port 8893: > > SELECT ?x ?a ?b ?P0src ?P1src > WHERE { > { > SERVICE <http://141.87.4.8:8899/sparql> { > ?x <http://purl.org/dc/terms/partOf> ?a . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < > <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) . > } . > BIND (<http://141.87.4.8:8899/sparql> AS ?P0src) > } > UNION > { > SERVICE <http://141.87.4.8:8891/sparql> { > ?x <http://purl.org/dc/terms/partOf> ?a . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < > <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) . > } . > BIND (<http://141.87.4.8:8891/sparql> AS ?P0src) > } > UNION > { > ?x <http://purl.org/dc/terms/partOf> ?a . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < > <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) . > BIND (<http://141.87.4.8:8893/sparql> AS ?P0src) > } > UNION > { > SERVICE <http://141.87.4.8:8895/sparql> { > ?x <http://purl.org/dc/terms/partOf> ?a . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < > <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) . > } . > BIND (<http://141.87.4.8:8895/sparql> AS ?P0src) > BIND (<http://141.87.4.8:8895/sparql> AS ?P0src) > } . > ?b <http://www.aktors.org/ontology/portal#article-of-journal> ?x . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) > . > BIND (<http://141.87.4.8:8893/sparql> AS ?P1src) } > > If this endpoint has no ?x matching pattern at line 22 of this query, an > exception is thrown. It is possible to avoid that by replacing line 20 to 26 > by > > UNION > { > { > ?x <http://purl.org/dc/terms/partOf> ?a . > FILTER ( ?x >= > <http://bibsonomy.org/uri/bibtexkey/conf/icmcs/NagatomoYOIT02/dblp> ) . > FILTER ( ?x < > <http://bibsonomy.org/uri/bibtexkey/conf/wpmc/EndoMA14/dblp> ) . > } > BIND (<http://141.87.4.8:8893/sparql> AS ?P0src) > } > > Best > Andy > ------------------------------------------------------------------------------ Site24x7 APM Insight: Get Deep Visibility into Application Performance APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month Monitor end-to-end web transactions and take corrective actions now Troubleshoot faster and improve end-user experience. Signup Now! http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140 _______________________________________________ Virtuoso-users mailing list Virtuoso-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/virtuoso-users
