From 4a837e07ed7fda4178101ebff39bcde58d3c57af Mon Sep 17 00:00:00 2001
From: SysAdmin <sysadmin@silverlabs.uk>
Date: Thu, 7 May 2026 17:59:48 +0100
Subject: [PATCH] fix(linux/build): discover job container ID from cgroup, not
 hostname (M1.1 iter21)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Run #4268's build-and-verify died <1s into Build A:

    docker: Error response from daemon: No such container: docker

Cause: build.sh's CI path uses `--volumes-from "$(hostname)"` to
inherit the parent job container's /workspace mount, but in the new
runner config (network: host applied via the now-actually-loaded
config.yaml) `hostname` returns the literal string "docker" inside
catthehacker/ubuntu:act-latest — the image bakes that into /etc/hostname
and act_runner doesn't override it. So `--volumes-from docker` looks for
a container literally named "docker", finds nothing, exits.

This worked in earlier runs (#4260) only because config.yaml *wasn't
being loaded* (see iter18 commit), so the runner ran on its built-in
defaults — which kept the container's hostname as the auto-generated
container ID. Fixing config.yaml exposed this latent bug.

Right way to learn your own container ID inside a Linux container is
/proc/self/cgroup, which contains the 64-char hex ID on every cgroup
driver:
  cgroup v1: 12:devices:/docker/<64-hex>
  cgroup v2: 0::/system.slice/docker-<64-hex>.scope

awk extracts the first 64-hex run; that becomes the --volumes-from
argument. If extraction fails (would only happen on a non-docker
runtime), fail loud rather than silent.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 linux/build/scripts/build.sh | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/linux/build/scripts/build.sh b/linux/build/scripts/build.sh
index 32a5f0f..db46cc1 100755
--- a/linux/build/scripts/build.sh
+++ b/linux/build/scripts/build.sh
@@ -99,8 +99,25 @@ echo "build.sh: output -> ${BUILD_DIR}"
 # job container, which inherits its /workspace mount intact. That keeps
 # paths identical inside and outside, so the inner heredoc below is the
 # same in both environments.
+#
+# Discovering the job container's own ID: `hostname` is unreliable on
+# act_runner / catthehacker (returned the literal string "docker" once
+# the runner was running with config.yaml's `network: host` applied —
+# see run #4268). /proc/self/cgroup is the portable way:
+#   * cgroup v1: lines look like `12:devices:/docker/<64-hex>`
+#   * cgroup v2: `0::/system.slice/docker-<64-hex>.scope`
+# Either way the 64-char hex container ID is in the path. Extract the
+# first one.
 if [[ -n "${GITHUB_ACTIONS:-}" ]]; then
-    BIND_ARGS=(--volumes-from "$(hostname)")
+    SELF_CID="$(awk 'match($0, /[a-f0-9]{64}/) { print substr($0, RSTART, RLENGTH); exit }' /proc/self/cgroup 2>/dev/null || true)"
+    if [[ -z "${SELF_CID}" ]]; then
+        echo "build.sh: could not determine own container ID from /proc/self/cgroup" >&2
+        echo "build.sh: cgroup contents:" >&2
+        cat /proc/self/cgroup >&2 || true
+        exit 1
+    fi
+    echo "build.sh: --volumes-from ${SELF_CID:0:12}"
+    BIND_ARGS=(--volumes-from "${SELF_CID}")
 else
     BIND_ARGS=(-v "${REPO_ROOT}:${REPO_ROOT}:rw")
     # If BUILD_DIR lives outside REPO_ROOT (uncommon, but the env-var