ci(e2e): probe VM gateway readiness - #2544
Conversation
Signed-off-by: Evan Lezar <elezar@nvidia.com>
|
Label |
elezar
left a comment
There was a problem hiding this comment.
gator-agent
Maintainer Approval Needed
Gator validation and PR monitoring are complete.
Validation: This is a small, concentrated CI/E2E robustness fix for VM gateway readiness probing.
Head SHA: 662fb9b286e0a30830885a0c6ff4706b2eb926af
Review: Independent principal-engineer review found no blocking findings.
Docs: Not needed; this changes CI/E2E scripts and does not alter direct user-facing UX or Fern docs surfaces.
Checks: Required gates are green: OpenShell / Branch Checks, OpenShell / Helm Lint, OpenShell / E2E, and OpenShell / GPU E2E.
E2E: test:e2e is applied and OpenShell / E2E passed, including e2e / E2E (rust-vm).
Human maintainer approval is now required.
Next state: gator:approval-needed
Summary
Replace VM gateway readiness checks that depend on the
Server listeninglogmessage with active health probes. This keeps VM CI stable when gateway logging
changes and verifies the interfaces the tests actually consume.
Related Issue
No issue. Fixes a brittle CI assumption exposed by #2492.
Changes
openshell statusoutput to report a connected gRPC endpointon timeout
script instead of matching gateway logs
Testing
bash -n e2e/rust/e2e-vm.sh tasks/scripts/vm/smoke-orphan-cleanup.shconnectedand rejectsdisconnectedmise run pre-commitpassesChecklist