stacklok / toolhive / 30268388184
70%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 840
Run time 3min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 69.546% (+0.001%) from 69.545%
30268388184

push

github

web-flow 
Make e2e workload readiness waits reliable (#6012)

* Make e2e workload readiness waits reliable

The group specs time out waiting for a freshly started workload to reach
the running state on loaded CI runners. Reaching running covers the
container start plus ToolHive's own initialize probe against the proxy,
which the product waits up to 5 minutes for, so a 60s spec-side budget
fails while the workload is still legitimately starting.

WaitForMCPServer also matched the name and the word running anywhere in
the list table, so it could return before the named workload was running,
and its timeout carried no diagnostics.

Separately, CI pre-pulled hardcoded image tags for servers the tests
reach by registry name. Renovate bumps toolhive-catalog daily but has no
manager for those docker pull lines, so fetch (1.0.2 vs 1.0.5) and time
(2026.1.26 vs 2026.7.10) had drifted and the pre-pull no longer covered
what the tests run. Resolve those tags from the binary under test.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* Bound the readiness poll's thv list call

Run() applies TestConfig.TestTimeout (10 minutes), so a single hung
list could outlast the readiness budget it is polling for. Callers
poll, so an overrunning list is retried rather than fatal.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* Share the workload lookup with the isolated-env wait

waitForIsolatedMCPServer was a copy of WaitForMCPServer's old body and
carried the same defect: it matched the name and the word running
anywhere in the list table. Export FindWorkload, which takes the command
builder so an isolated config/home/data spec can supply its own, and
have both waits read the named workload's own record.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

80214 of 115339 relevant lines covered (69.55%)

83.72 hits per line

Coverage Regressions

Lines Coverage ∆ File
3
96.36
 -0.47% pkg/authz/authorizers/cedar/core.go
3
80.56
 -0.7% pkg/transport/proxy/httpsse/http_proxy.go
2
62.12
 0.0% pkg/workloads/manager.go
Jobs
ID Job ID Ran Files Coverage
1 30268388184.1 840
69.55
 GitHub Action Run
Source Files on build 30268388184