Hello,
We have recently built a custom extension to fetch data from a SQL data source.
Our underlying data source is an Exasol cluster and in order to establish the connection we have enabled the JDBC driver of the vendor according to the official instructions documented here.
Initially everything went according to plan, the data was fetched every minute so we started expanding the number of jdbcRemote endpoints to have complete coverage according to our needs.
After a few days of observing we noticed that from time to time some endpoints are failing randomly with the following error message:
[out]org.apache.tomcat.jdbc.pool.PoolExhaustedException: [pool-4-thread-1] Timeout: Pool empty. Unable to fetch a connection in 30 seconds, none available[size:1; busy:1; idle:0; lastwait:30000].
This was not a limitation on the data source side as the connection pool there is way larger and we were also not even seeing an attempt to connect. We quickly concluded that the default connection pool size of our JDBC driver is actually 64, so what mechanism in the extension decides to overwrite the default connection pool size and "shrink" it to 1 (as seen in the message above)?
Furthermore, this message was randomly affecting endpoints that were grouped under the same monitoring configuration and "evicting" the endpoint to its own monitoring configuration remediated the issue.
We need more details around how the Extension is building a JDBC connector as right now there is nothing in the extension log that hints to how to connector is initialised.
Thanks,
Adrian
