Bulk Loading Tools + SSH Tunnel

[isaacrlee]

If use_bulk is set to true, a bulk loading tool such as bcp is installed, and ssh_tunnel and ssh_private_key are set, would the connection to the SSH tunnel be used by bcp?

[flarco]

Ah, interesting premise. No, bcp is only checked and used locally.
See https://github.com/slingdata-io/sling-cli/blob/main/core/dbio/database/database_sqlserver.go#L197.
Actually, thinking of it, even if ssh is set, if bcp is detected locally, sling will forget about the ssh and attempt to load from the local machine. Not ideal.
This logic needs to be implemented. Opened #280 for this.

[flarco]

@isaacrlee to be clear, bcp would need to be installed on the ssh machine. As far as I understand, bcp does not connect to ssh to import into SQL server.

[isaacrlee]

Thanks for the reply. I might be misunderstanding Sling/bcp/SSH tunnels, but would it not be possible for bcp to use an updated connection URL with local host and port after the SSH tunnel is created?

I see the code for the updated connection URL used by later non-bulk inserts here, but here the original connection URL will always be used for BCP imports.

[flarco]

Actually, that's a good one. Instead of having bcp on the ssh machine, bcp should simply point to the forwarded port. Good.

[flarco]

Added to branch v1.2.7

[flarco]

ecd9bdb

[isaacrlee]

Awesome, thank you!