Server stops querying downstream databases (Docker Stardog to Azure MySQL DB)

Support Bug
1

Hi,

I am using the latest version of Stardog in a docker container. After it is spun up I am adding an Azure SQL (MySQL) database and mapping it in based on the vkg tutorial however Stardog seems to drop the connection and not reconnect.

When executing a SPARQL query like below:

SELECT ?member ?name {
   GRAPH <virtual://music> {
      ?band a :Band ;
            :name "The Beatles" .
      ?member :member ?band ;
            :name ?name .
   }
}

This query will work a bunch of times and then stop working (supposedly) randomly. When it fails I get the following error:

bash-4.2$ /opt/stardog/bin/stardog query execute music /tmp/correct_query.sparql 
com.complexible.stardog.plan.eval.operator.OperatorException: Unable to execute virtual graph query. SQL string: SELECT `membership`.`artist`, `artist`.`name`
FROM `TEST`.`membership`
INNER JOIN `TEST`.`artist` ON `membership`.`artist` = `artist`.`id` AND `artist`.`name` IS NOT NULL
INNER JOIN `TEST`.`artist` AS `artist0` ON `membership`.`band` = `artist0`.`id` AND `artist0`.`name` = 'The Beatles' AND `artist0`.`type` = 2

If I copy and paste the generated SQL into my database it returns results, so it isn't an issue with the SQL generation:

mysql> SELECT `membership`.`artist`, `artist`.`name`
    -> FROM `TEST`.`membership`
    -> INNER JOIN `TEST`.`artist` ON `membership`.`artist` = `artist`.`id` AND `artist`.`name` IS NOT NULL
    -> INNER JOIN `TEST`.`artist` AS `artist0` ON `membership`.`band` = `artist0`.`id` AND `artist0`.`name` = 'The Beatles' AND `artist0`.`type` = 2;
+--------+-----------------+
| artist | name            |
+--------+-----------------+
|      1 | John Lennon     |
|      2 | Paul McCartney  |
|      3 | Ringo Starr     |
|      4 | George Harrison |
|      1 | John Lennon     |
|      2 | Paul McCartney  |
|      3 | Ringo Starr     |
|      4 | George Harrison |
+--------+-----------------+
8 rows in set (0.20 sec)

The only pattern I have seen that might help is that after I run an incorrectly formatted query it stops working (as if it crashes the SQL connection and it doesn't get re-established)

Any thoughts?

2

Can you check for the error in the stardog.log file? This is common on MySQL because persistent connections will timeout after 8 hours by default but Stardog's default connection pooling configuration should avoid this.

3

Hi @jess ,

error.log (32.4 KB)

Attached the logfile. Hopefully this helps

4

Does restarting the Stardog server consistently fix the issue? Have you altered any of the connection pooling settings in the data source?