Currently the check-log-errors case is a separate test case that runs after the entire cilium-cli connectivity test suite runs. This makes it hard to correlate the error with the condition that caused the error (ie under which test did it run?)

It would be good to correlate these errors with the individual test. A couple of different proposals on how to achieve this:

• Refactor so that the CLI gathers agent logs before and after each individual test (could be costly)
• Gather timestamps before and after every test run, and delay the test output until after all tests are run including the check-log-errors case, then correlate the failure log timestamps with the time of the test run.
  • Note, clocks are not consistent across hosts and the CLI may run outside the cluster, so we may want some fuzz factor for correlation if it's ambiguous when the failure occurred.