OHDSI Home | Forums | Wiki | Github

Initial process after building ATHENA on my system?


#1

I’m a researcher in a lab developing a system works with CDM. To understand vocabulary’s structure and implement a simple search, I installed and ran athena on my local system (including postgresql). I also imported SNOMED v5 and v4.5 vocabulary into my local database for testing. However, when I connect to local athena, the filters on the left are all empty and no search results are returned. There seems to be no vocabulary in the database. Is there anything else I need to do when I install athena on my local system? After importing the vocabulary into the database, it seems to recognize the vocabulary list through initializing.


#2

In the console, I saw the following exceptions:
“org.apache.solr.client.solrj.SolrServerException: Server refused connection at: http: // localhost: 8984 / solr / concepts”
It seems that I need solr to run athena, and do some setup of solr. What should I do with solr?


(Alex Cumarav) #3

Hello @stkim,

Athena app utilizes Solr to index concepts hierarchy and execute search requests against Solr index. So you need Solr to make Athena working. Vocabularies is stored in the database. In order to build search index Solr reading data directly from the database and Athena query Solr engine with the search queries.

Your steps should be:

  1. Load vocabularies into Postgres database.
  2. Configure Solr and perform data import
  3. Configure Athena to use Solr
  4. Ensure that you can see search results in the Athena UI

You can find Solr configuration in the Athena repository:

Data Import default config is here:

<requestHandler name="/dataimport" class="solr.DataImportHandler">

<lst name="defaults">

<str name="config">data-config.xml</str>
….


#4

Thank you @acumarav for your detailed description.
It seems that the solr app was able to search athena so quickly even with so many vocabularies.


(Alex Cumarav) #5

Yes, initial data import into Solr can lasts around 10 minutes but search should be quick


t