Write the stress test for gRPC, http, and tcp probes

Summary of Open Issue: Write Stress Test for gRPC, HTTP, and TCP Probes

Issue Overview: The objective of this issue is to create a new End-to-End (e2e) stress test for liveness probes using gRPC, HTTP, and TCP protocols, addressing the bug fix documented in issue #89898. This test will replace a previously skipped unit test (#115329) and will be designated as [Serial] to ensure sequential execution.

Implementation Plan:

  • The test will involve launching a significant number of containers, each configured with a liveness probe that checks every second.
  • After a defined duration, the test will validate that none of the containers have a restartCount greater than zero, confirming their continued operation.

Context and Importance: This effort aligns with the goals set out in KEP #2727 regarding GA requirements for Kubernetes. The test needs to encompass all three probe types—gRPC, HTTP, and TCP—to ensure robustness in liveness check implementations.

Current Contributors:

  • SergeyKanzhelev (initiator)
  • Multiple contributors have been assigned throughout the discussion, including Aojea, Anis-cpu-13, Prateekaryyan, Alok1929, and Lanycrost, among others.
  • Lanycrost has notably begun work on this issue, sharing links to a benchmark test in progress.

Open Questions and Activity:

  • A PR (#117439) was opened but was subsequently closed automatically. Lanycrost has indicated plans to rework and reopen this PR.
  • Others, including Mukeshgupta7 and Hrishikeshdkakkad, have expressed interest in contributing and clarifying the current status.

Labels:

  • good first issue, help wanted, kind/cleanup, needs-triage, priority/important-soon, sig/node

Next Steps: Advanced developers interested in contributing should collaborate with current assignees, particularly Lanycrost, to ensure synchronization with ongoing work and to address the remaining tasks in the issue’s objectives.