stacklok / toolhive / 30335252401

70%

Build Type

push

github

Committed by web-flow

Commit Message

Deflake CI: TUF warn mode, readiness diagnostics, test races (#6063)

* Stop warn-mode verification failing runs on TUF outages

In the default warn mode, VerifyImage returned verifier construction
errors and non-signature verification errors as fatal, while an actual
failed verification only warned. A sigstore TUF CDN connection reset
therefore hard-failed 'thv run' for any registry server with provenance
(observed repeatedly in CI, e.g. run 30297766499). Warn mode now warns
and continues on every verification failure; enabled mode still fails
closed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y5WXPQr5Da9Cox2nZWhg6n

* Surface why a workload failed or stalls in starting

startWorkload and the detached-spawn failure path wrote status error
with an empty context, clobbering the failure reason RunWorkload's
retry loop had just recorded; CI readiness timeouts then showed only a
bare status "error". Record err.Error() instead.

The readiness probe logged per-attempt outcomes at DEBUG only, so a
workload stuck in starting left no visible trace of what the probe
observed. Log an INFO progress line every 30s with the last observed
outcome, and include it in the timeout error.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y5WXPQr5Da9Cox2nZWhg6n

* Dump proxy logs when an e2e readiness wait times out

DebugServerState only printed container logs, which show a healthy
server while the workload sits in starting or error: the failure is in
the detached supervisor, whose output goes to the proxy log file. Dump
'thv logs --proxy' too so a CI timeout names the actual blocker.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y5WXPQr5Da9Cox2nZWhg6n

* Quarantine all four server-to-client conformance flakes

Suite v0.1.16 has four scenarios that drive a server... (continued)

Coverage Stats

19 of 31 new or added lines in 3 files covered. (61.29%)

9 existing lines in 3 files now uncovered.

80784 of 115917 relevant lines covered (69.69%)

86.9 hits per line