Invalid ping query for TypeOrmHealthIndicator using oracle connection

[mvandervliet]

I'm submitting a...

[ ] Regression 
[x] Bug report
[ ] Feature request
[ ] Documentation issue or request
[ ] Support request => Please do not submit support request here, instead post your question on Stack Overflow.

Current behavior

Fails when using TypeOrmHealthIndicator with TypeOrmModule configuration for an type: 'oracle' connection. The issue is due to the invalid SELECT 1 query used as the ping.

Expected behavior

The proper statement for type: 'oracle' should be SELECT 1 FROM DUAL

Minimal reproduction of the problem with instructions

1. Create TypeOrm connection with type: 'oracle'
2. Create HealthIndicator for TypeORM as described in docs
3. Load /health and see 503 with status: 'down' and no message

What is the motivation / use case for changing the behavior?

Make TypeORM health indicator function for all supported connection types

Environment

Nest version: 6.7.1

 
For Tooling issues:
- Node version: v10.16.3  
- Platform:  Mac, Docker/Linux

Others:

[BrunnerLivio]

Thanks a lot for reporting! Seems to be a valid issue. Would you like to create a PR? You could easily add this to the existing TypeOrmHealthIndicator#pingDb function.

terminus/lib/health-indicators/database/typeorm.health.ts

Lines 78 to 86 in f558ca5

	private async pingDb(connection: Connection, timeout: number) {
	const check =
	connection.options.type === 'mongodb'
	? connection.isConnected
	? Promise.resolve()
	: Promise.reject()
	: connection.query('SELECT 1');
	return await promiseTimeout(timeout, check);
	}

Maybe you'll need to refactor the statement to a switch/case or if/else statement, since the "if"-shorthand gets hard to read with 3+ options.

[mvandervliet]

sure @BrunnerLivio, I can definitely open the PR... I'd actually solved in a workaround by extending the class for now, but I can refactor the method and let y'all review 😉

[BrunnerLivio]

Included in 6.5.2 🎊