ci(e2e): skip final cleanup on disposable PR runners

Destroying runner-local state after a completed suite can push the job
past its deadline even though the runner is discarded. Keep cleanup
between preparation attempts and for reused hosts instead.

Fixes: 186547d835 ("test(e2e): run the merge-gating lanes on Talos containers")
Assisted-by: LLM
Signed-off-by: Aleksei Sviridkin <f@lex.la>

.github/workflows/e2e-fork.yaml

@@ -771,21 +771,6 @@ jobs:
           name: cozyreport
           path: _out/cozyreport.tgz
           if-no-files-found: warn
-      # Both halves, and in this order: the compose project owns the nodes and
-      # the zpools outlive it.
-      - name: Tear down sandbox
-        if: always()
-        run: |
-          # Annotated, not fatal: same reasoning as the in-tree lane. The runner
-          # is discarded after the job, so a dirty host has no next tenant, and
-          # failing here would post the required "E2E Tests" status as a failure
-          # indistinguishable from a real one.
-          cleanup_rc=0
-          make -C packages/core/testing SANDBOX_NAME=$SANDBOX_NAME delete-cluster-container || cleanup_rc=$?
-          make -C packages/core/testing SANDBOX_NAME=$SANDBOX_NAME delete || true
-          if [ "$cleanup_rc" -ne 0 ]; then
-            echo "::error title=sandbox teardown::teardown exited $cleanup_rc; the host may hold a stale compose project or zpool"
-          fi
 
   # Flip the pending "E2E Tests" commit status posted by `resolve` to its
   # terminal state. Runs on failure too, so a failed publish or e2e turns the

.github/workflows/pull-requests.yaml

@@ -1348,28 +1348,6 @@ jobs:
             Resume from inside: `breakpoint resume`. Otherwise the breakpoint
             exits 10 minutes after the last SSH session disconnects.
 
-      # ▸ Tear down environment (always runs)
-      #
-      #   Both halves, and in this order: the compose project owns the nodes and
-      #   the zpools outlive it. A leftover pool makes the next run on a reused
-      #   host adopt storage that already carries a dead cluster's volumes.
-      - name: Tear down sandbox
-        if: always()
-        run: |
-          # Loud, but never the job's verdict. A failed teardown means a dirty
-          # host, and this runner is discarded after the job, so there is no next
-          # run to protect -- while failing the job here posts the required
-          # "E2E Tests" status as a failure that no reader can tell apart from a
-          # real one. The annotation is what a reused host (a developer machine)
-          # needs; `delete-cluster-container` returning non-zero is not evidence
-          # about the product.
-          cleanup_rc=0
-          make -C packages/core/testing SANDBOX_NAME=$SANDBOX_NAME delete-cluster-container || cleanup_rc=$?
-          make -C packages/core/testing SANDBOX_NAME=$SANDBOX_NAME delete || true
-          if [ "$cleanup_rc" -ne 0 ]; then
-            echo "::error title=sandbox teardown::teardown exited $cleanup_rc; the host may hold a stale compose project or zpool"
-          fi
-
   # Report the required "E2E Tests" gate as a commit status rather than relying
   # on a job named "E2E Tests". A required *job* that is skipped counts
   # as PASSED by branch protection — desired for docs-only PRs, but the exact

hack/cozyreport.bats

@@ -966,6 +966,37 @@ STUB
   }
 }
 
+@test "PR E2E jobs clean up between preparation attempts but not after testing" {
+  tmp="$(mktemp -d)"
+  mkdir -p "$tmp/bin"
+  cat >"$tmp/bin/make" <<'STUB'
+#!/bin/sh
+printf '%s\n' "$*" >>"$CLEANUP_CALLS"
+case "$*" in
+  *prepare-env-container*)
+    if [ ! -f "$CLEANUP_ATTEMPT" ]; then
+      touch "$CLEANUP_ATTEMPT"
+      exit 1
+    fi
+    ;;
+esac
+STUB
+  chmod +x "$tmp/bin/make"
+  for wf in pull-requests e2e-fork; do
+    workflow="$HACK_DIR/../.github/workflows/$wf.yaml"
+    yq --unwrapScalar '.jobs.e2e.steps[] | select(.name == "Prepare environment") | .run' "$workflow" >"$tmp/prepare.sh"
+    CLEANUP_CALLS="$tmp/$wf.calls" CLEANUP_ATTEMPT="$tmp/$wf.attempt" SANDBOX_NAME=test PATH="$tmp/bin:$PATH" bash -eu "$tmp/prepare.sh"
+    printf '%s\n' 'SANDBOX_NAME=test prepare-env-container' '-C packages/core/testing SANDBOX_NAME=test delete-cluster-container' 'SANDBOX_NAME=test prepare-env-container' >"$tmp/expected"
+    diff "$tmp/expected" "$tmp/$wf.calls"
+    scripts="$(yq --unwrapScalar '.jobs.e2e.steps[] | select(.name != "Prepare environment") | .run // ""' "$workflow")"
+    if printf '%s\n' "$scripts" | grep -Eq 'delete-cluster-container|docker[[:space:]]+(compose.*down|rm)|zpool[[:space:]]+destroy|make.*[[:space:]]delete([[:space:]]|$)'; then
+      echo "FAIL: $wf destroys runner-local state outside preparation"
+      false
+    fi
+  done
+  rm -rf "$tmp"
+}
+
 @test "the step that runs this collector has a ceiling of its own everywhere" {
   # Every bound inside the script protects one read. None of them protects the
   # STEP: the tarball is written on the collector's last line, the upload is the