ccl/streamingccl/streamingest: TestTenantStreaming failed

[cockroach-teamcity]

ccl/streamingccl/streamingest.TestTenantStreaming failed with artifacts on master @ e1cdb469d4828e2e1df0f801ac50d4e4f9943e1e:

=== RUN   TestTenantStreaming
    test_log_scope.go:79: test logs captured to: /artifacts/tmp/_tmp/045fecfe9e18c9da6f776a53d446f2c5/logTestTenantStreaming711556021
    test_log_scope.go:80: use -show-logs to present logs inline
    stream_ingestion_job_test.go:191: error executing 'SELECT * FROM d.t1': pq: no healthy sql instances available for planning
    panic.go:661: -- test log scope end --
--- FAIL: TestTenantStreaming (62.10s)

Parameters: TAGS=bazel,gss

Help

See also: How To Investigate a Go Test Failure (internal)

/cc @cockroachdb/bulk-io

This test on roachdash | Improve this report!

Jira issue: CRDB-16620

[ajwerner]

https://teamcity.cockroachdb.com/buildConfiguration/Cockroach_UnitTests_Test/5460040?showRootCauses=false&expandBuildChangesSection=true&expandBuildProblemsSection=true&expandBuildDeploymentsSection=true&expandBuildTestsSection=true here's a bors failure

[ajwerner]

I looked into this a bit. The deal is that sometimes we don't find any sql instances here:

cockroach/pkg/sql/distsql_physical_planner.go

Lines 1171 to 1177 in 9e4364a

	instances, err := dsp.sqlInstanceProvider.GetAllInstances(ctx)
	if err != nil {
	return nil, err
	}
	if len(instances) == 0 {
	return nil, errors.New("no healthy sql instances available for planning")
	}

Probably we should always find ourself. That will take a tweak down in the SQLInstance layer.

[ajwerner]

@rharding6373 this may be of interest to you.