ci: collect dockerd pprof dumps on test timeout

Add timeout signaling from test2json2gha to GITHUB_OUTPUT so subsequent
CI steps can detect when tests timed out. On timeout, the dump logs step
now collects goroutine stacks, a binary heap profile, and the dockerd
binary from the runner for offline analysis with go tool pprof.

Signed-off-by: Brian Goff <cpuguy83@gmail.com>

Author: cpuguy83

.github/workflows/ci.yml

@@ -299,6 +299,7 @@ jobs:
         env:
           TEST_SUITE: ${{ matrix.suite }}
       - name: Run integration tests
+        id: run-tests
         run: |
           set -ex
           if [ -n "${TEST_SUITE}" ] && [ ! "${TEST_SUITE}" = "other" ]; then
@@ -342,15 +343,29 @@ jobs:
           set -e
 
           dir="$(mktemp -d)"
-          f="${dir}/dockerd.log"
-          echo "DOCKERD_LOG_PATH=${f}" >> $GITHUB_OUTPUT
-          sudo journalctl -u docker > "${f}"
+          echo "DOCKERD_LOG_PATH=${dir}" >> $GITHUB_OUTPUT
+
+          if [ "${{ steps.run-tests.outputs.test_timeout }}" = "true" ]; then
+            echo "::group::Collecting pprof data from dockerd (test timeout detected)"
+            curl --unix-socket /var/run/docker.sock \
+              -o "${dir}/goroutine-stacks.txt" \
+              "http://localhost/debug/pprof/goroutine?debug=2" || true
+
+            curl --unix-socket /var/run/docker.sock \
+              -o "${dir}/heap-profile.bin" \
+              "http://localhost/debug/pprof/heap" || true
+
+            cp "$(which dockerd)" "${dir}/dockerd" || true
+            echo "::endgroup::"
+          fi
+
+          sudo journalctl -u docker > "${dir}/dockerd.log"
       - name: Upload buildkit logs
         if: failure()
         uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
           name: e2e-dockerd-logs-${{ matrix.suite }}
-          path: ${{ steps.dump-logs.outputs.DOCKERD_LOG_PATH }}
+          path: ${{ steps.dump-logs.outputs.DOCKERD_LOG_PATH }}/*
           retention-days: 1
 
   unit:

cmd/test2json2gha/main.go

@@ -5,6 +5,7 @@ import (
 	"flag"
 	"fmt"
 	"io"
+	"iter"
 	"log/slog"
 	"os"
 	"runtime/debug"
@@ -73,6 +74,7 @@ func do(in io.Reader, out io.Writer, cfg config) (bool, error) {
 		var wg waitGroup
 
 		results.markUnfinishedAsTimeout()
+		signalTimeout(results.Results())
 
 		wg.Go(func() {
 			var rf ResultsFormatter
@@ -135,6 +137,28 @@ func do(in io.Reader, out io.Writer, cfg config) (bool, error) {
 	return bool(anyFailed), nil
 }
 
+// signalTimeout writes test_timeout=true to GITHUB_OUTPUT if any test timed out.
+// This allows subsequent CI steps to detect that a timeout occurred.
+func signalTimeout(results iter.Seq[*TestResult]) {
+	ghOutput := os.Getenv("GITHUB_OUTPUT")
+	if ghOutput == "" {
+		return
+	}
+
+	for r := range results {
+		if r.timeout {
+			f, err := os.OpenFile(ghOutput, os.O_WRONLY|os.O_APPEND, 0)
+			if err != nil {
+				slog.Error("Error opening GITHUB_OUTPUT", "error", err)
+				return
+			}
+			fmt.Fprintln(f, "test_timeout=true")
+			f.Close()
+			return
+		}
+	}
+}
+
 type waitGroup struct {
 	sync.WaitGroup
 }