Our telemetry test using remote write is increasingly flaky. The recurring error is:
Any failed samples will cause this test to fail. This is perhaps a too strict requirement. We could consider it good enough if some samples are send successfully. The current version tests telemeter behavior on top of CMO behavior.