TestLongRunningAgentForLeaks is failing due to two reasons:

• Flaky for linux-amd64
  • It fails due to the monitoring module staying in a degraded mode for a while, it will eventually return to a healthy state. It takes a few dozen seconds for libbeat to start the metrics endpoint and we only wait for 10 seconds.
  • FIX: Waiting for 60s should solve this

EDIT: Following is fixed:

• Agent remaining DEGRADED for windows, even while running as an administrator.
  • This seems like a genuine issue with windows and this requires some debugging and troubleshooting. I'm currently working on it
  • This only affects system.process and system.process_summary.
  • TEMPORARY FIX TO UNBLOCK CI: disable system.process and system.process_summary from TestLongRunningAgentForLeaks. (ps: Let me know if this sounds good to you?)
  • PERMANENT FIX: Investigate and resolve the underlying issue affecting these metrics.

Example of an failure:
https://buildkite.com/elastic/elastic-agent-extended-testing/builds/1850#01913472-043d-4264-b524-8c1dfa828833