Skip to content

armgit / xerj

Public Imported from GitHub
0

AI search engine and MCP server for bots and agents. One Rust binary indexes code, docs, logs, JSONL/CSV and PDFs for BM25 full-text, vector and hybrid search, code search without grep, agent memory and RAG. Elasticsearch-compatible API; Corpus Hub serves hundreds of reference memory corpora.

ci(#751): add a 50-min job-level cap as defense-in-depth against any Build+Test hang

xerj-org0b321ef41ae3
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index bfc7c2999..4564c47d6 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -263,6 +263,19 @@ jobs:
build-test:
name: Build + Test
runs-on: ubuntu-latest
+ # #751's second-order damage is the whole job hanging against GitHub's 6h
+ # default while main's `cancel-in-progress: false` never auto-cancels — main
+ # then sits ~4h with no CI verdict. The step-6 `timeout-minutes: 12` below
+ # bounds the one documented hang (step 6, default parallelism) precisely and
+ # fast. This job-level cap is defense-in-depth: it bounds ANY hang in the job
+ # (setup, cache, a future test race the step timeout would not cover) to
+ # 50 min instead of the 6h default. It is deliberately generous because
+ # step 5 (build + threads=2 tests) can legitimately take ~36 min on a loaded
+ # 2-4 vCPU runner — a slow compile, not a hang (measured: step 5 built for
+ # ~36 min then step 6 ran and passed) — so a tighter cap would false-fail
+ # slow-but-healthy runs. Neither cap fixes the root race (#751 stays open for
+ # the bisect + port/fixture serialisation).
+ timeout-minutes: 50
steps:
- uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262 # v4.4.0