Pool recovery

[elsbrock]

We recently had what seems like a connection hiccup to our HANA database (connection reset). While the database was reachable shortly after, it seems like all connections in the pool failed and did not recover, leading to a downtime.

Can you comment on the intended behavior? What can we do to make sure the pool automatically recovers from such failures and connections are correctly invalidated and recreated? I was unable to find information on this topic.

We are using hdb-pool in the context of typeorm.

Thanks!

[ckyycc]

Before delivering a connection, the pool will check the status of the connection. The pool will deliver the connection only when the state of the connection is 'connected' (https://help.sap.com/viewer/0eec0d68141541d1b07893a39944924e/2.0.04/en-US/aeb50c109d19452ba6515a1fab515e5c.html). If the status is not 'connected', the related connection will be destroyed and moved out from the pool automatically (that specific request will be failed with CONNECTION_VALIDATION_ERROR).

According to your description, the status of the connection was 'connected' but actually the connection was not working (The pool will not know this).

So, I think:

1. it's better to have a reproduce scenario.
2. check with SAP about the connection.state( ), why connection.state() returns 'connected' but the connection actually doesn't work.
3. for the pool, you can manually trigger pool.destroy(connection) to destroy a specific connection or pool.clear() to clear the entire pool.