Perhaps, my way to create databases (located where i need to is) could be useful * first, I create interactively a dataset in the Fuseki UI (Persistent database), say mybase * then, I move the directory from run/databases/mybase to the place convenient for me * finally, I change the tdb:location item in the file run/configuration/mybase.ttl, to point to the place where the dataset lives now Better suggestions welcome
-- Jean-Claude Moissinac Le ven. 7 févr. 2020 à 12:26, Andy Seaborne <[email protected]> a écrit : > > > On 07/02/2020 09:02, Immanuel Normann wrote: > > Hi there, > > I have Fuseki 3.14 running inside my Tomcat 9 and installed > > apache-jena-3.14. > > Now I am trying to work with tdbloader2 to create a database from > ttl-files > > for Fuseki: > > > > bin/tdbloader2 --loc mydb mydata.ttl > > This generates a directory mydb with all the indexes. My question is: how > > do I make Fuseki aware of my newly created mydb? > > > > My idea was to simply move it into /etc/fuseki/databases where all other > > data indexes are located and restart Fuseki. But that is apparently not > > enough: The Fuseki WebGUI does not list mydb , where as it lists all > other > > datasets which I have created via the Fuseki WebGUI. > > > > How do I notify Fuseki about my freshly created index via tbloader2 > without > > GUI interaction? > > Put the database, .../run/database and a config file in > ../run/configuration and restart the server. > > ---- > > If you allow a GUI action, create the database in the eventual location, > then create the dataset in the UI - it will pick up the prebuilt > database, no restart required. > > # Initially ds2 must not exist > tdb2.tdbloader --loc run/databases/ds2 ...datafiles... > <create ds2 in the interface> > > Fuseki and any tdbloader can both be using the database at the same > time. But in the pattern above, the database is created while Fuseki is > unaware of it, then the Fuseki connects to it. > > In fact, the UI is POSTing a request to the server to create the dataset > but that's not documented properly. > > If you have a TDB2 database in Fuseki > (confusion alert: tdbloader2 create TDB1 databases - legacy) > you can, instead, directly load the data into Fuseki with the server > running. It is slower to load at scale than the bulk loader, but there > are no other steps so it may still be faster overall. > > If you have that much data, and the machine has fast I/O > > tdb2.tdbloader --loader=parallel .... > > (there are other --loader options) > > There is no hard-and-fast rule which loader is faster. It is hardware > dependent. > > Andy > > > > > Regards, > > Immanuel > > >
