# Set-up Virtuoso+Xapian on Production

## Tags

* assigned: bonfacem, zachs, fredm
* priority: high
* type: ops
* keywords: virtuoso

## Description

We already have virtuoso set-up in tux02.  Right now, to be able to interact with RDF, we need to have virtuoso set-up.  This issue will unblock:

* Global Search in Production

=> https://github.com/genenetwork/genenetwork3/pull/137 Update RDF endpoints

=> https://github.com/genenetwork/genenetwork2/pull/808 UI/RDF frontend


## HOWTO: Updating Virtuoso in Production (Tux01)


Note where the virtuoso data directory is mapped from the "production.sh" script as you will use this in the consequent steps:

> --share=/export2/guix-containers/genenetwork/var/lib/virtuoso=/var/lib/virtuoso

### Generating the TTL Files

* Run "generate-ttl-files" to generate the TTL files:

```
time guix shell guile-dbi -m manifest.scm -- ./generate-ttl-files.scm --settings conn-dev.scm --output /export2/guix-containers/genenetwork-development/var/lib/virtuoso --documentation /tmp/doc-directory
```

=> https://git.genenetwork.org/gn-transform-databases/tree/generate-ttl-files.scm generate-ttl-files.scm

* (Recommended) Alternatively, copy over the TTL files (in Tux02) to the correct shared directory in the container ("--share=/export2/guix-containers/genenetwork-development/var/lib/virtuoso=/var/lib/virtuoso"):

> cp /home/bonfacem/ttl-files/*ttl /export2/guix-containers/genenetwork/var/lib/virtuoso/

### Loading the TTL Files

* Make sure that the virtuoso service type has the "dirs-allowed" variable set correctly:

```
(service virtuoso-service-type
         (virtuoso-configuration
          (server-port 7892)
          (http-server-port 7893)
          (dirs-allowed "/var/lib/virtuoso")))
```

* Get into isql:

> guix shell virtuoso-ose -- isql 7892

* Make sure that no pre-existing files exist in "DB.DBA.LOAD_LIST":

> SQL> select * from DB.DBA.LOAD_LIST;
> SQL> delete from DB.DBA.load_list;

* Delete the genenetwork graph:

> SQL> DELETE FROM rdf_quad WHERE g = iri_to_id('http://genenetwork.org');

* Load all the TTL files (This takes some time):

> SQL> ld_dir('/var/lib/virtuoso', '*.ttl', 'http://genenetwork.org');
> SQL> rdf_loader_run();
> SQL> CHECKPOINT;

* Verify you have some RDF data by running:

```
SQL> SPARQL
PREFIX gn: <http://genenetwork.org/id/> 
PREFIX gnc: <http://genenetwork.org/category/> 
PREFIX owl: <http://www.w3.org/2002/07/owl#> 
PREFIX gnt: <http://genenetwork.org/term/> 
PREFIX skos: <http://www.w3.org/2004/02/skos/core#> 
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> 
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
PREFIX taxon: <http://purl.uniprot.org/taxonomy/> 

SELECT * WHERE { 
    ?s skos:member gn:Mus_musculus .
    ?s ?p ?o .
};
```

* Update GN3 Configurations to point to the correct Virtuoso instance:

> SPARQL_ENDPOINT="http://localhost:7893/sparql"

## Generating the Xapian Index

* Make sure you are using the correct guix profile or that you have your PYTHONPATH pointing to the GN3 repository.

* Generate the Xapian Index using "genenetwork3/scripts/create-xapian-index" against the correct output directory (The build takes around 71 minutes on an SSD Drive):

> time python index-genenetwork create-xapian-index /export/data/genenetwork-xapian/ mysql://<user>:<password>@localhost/db_webqtl http://localhost:7893/sparql

* After the build, you can verify that the index works by:

> guix shell xapian -- xapian-delve /export/data/genenetwork-xapian/

* Update GN3 configuration files to point to the right Xapian path:

> XAPIAN_DB_PATH="/export/data/genenetwork-xapian/"