chore(misc): gha improve retry logic and

Signed-off-by: Fluent Crafter <205769460+fluentcrafter@users.noreply.github.com>

Author: fluentcrafter

.github/actions/run-e2e-tests/action.yml

@@ -114,7 +114,7 @@ runs:
         path: ${{ github.workspace }}/${{ inputs.working-directory }}/tmp/artifacts
         merge-multiple: true
 
-    - name: Load Docker images from artifacts or pull develop for unchanged services
+    - name: Load Docker images from artifacts
       shell: bash
       working-directory: ${{ inputs.working-directory }}
       env:
@@ -192,20 +192,12 @@ runs:
       working-directory: ${{ inputs.working-directory }}
       run: |
         set -euo pipefail
-        timeout 10m bash -c '
-          max=10
-          attempt=0
-          until make start-env-with-tracing-v2-ci CLEAN_PREVIOUS_ENV=false; do
-            attempt=$((attempt + 1))
-            if [ "$attempt" -ge "$max" ]; then
-              echo "All $max attempts failed"
-              exit 1
-            fi
-            echo "Attempt $attempt failed — cleaning up and retrying in 30 s..."
-            make clean-environment || true
-            sleep 30
-          done
-        '
+        make retry \
+          TARGET="start-env-with-tracing-v2-ci CLEAN_PREVIOUS_ENV=false" \
+          RETRY_LIMIT=10 \
+          RETRY_TIMEOUT=10m \
+          RETRY_BACKOFF=30s \
+          RETRY_CLEANUP_TARGET=clean-environment
 
     - name: List docker containers/images
       if: ${{ always() && inputs.e2e-tests-containers-list == 'true' }}

Makefile

@@ -109,6 +109,42 @@ staterecovery-replay-from-block:
 	docker compose -f docker/compose-tracing-v2-staterecovery-extension.yml down zkbesu-shomei-sr shomei-sr
 	L1_ROLLUP_CONTRACT_ADDRESS=$(L1_ROLLUP_CONTRACT_ADDRESS) STATERECOVERY_OVERRIDE_START_BLOCK_NUMBER=$(STATERECOVERY_OVERRIDE_START_BLOCK_NUMBER) docker compose -f docker/compose-tracing-v2-staterecovery-extension.yml up zkbesu-shomei-sr shomei-sr -d
 
+# Retries TARGET up to RETRY_LIMIT times, aborting each attempt after
+# RETRY_TIMEOUT. Uses perl's alarm+exec instead of GNU coreutils `timeout`,
+# which isn't installed by default on macOS. Optionally runs
+# RETRY_CLEANUP_TARGET between failed attempts (e.g. to reset docker state).
+retry: RETRY_LIMIT:=5
+retry: RETRY_TIMEOUT:=10m
+retry: RETRY_BACKOFF:=10s
+retry:
+	@if [ -z "$(TARGET)" ]; then \
+		echo "Usage: make retry TARGET=<target> [RETRY_LIMIT=n] [RETRY_TIMEOUT=Xm] [RETRY_BACKOFF=Xs] [RETRY_CLEANUP_TARGET=<target>]"; \
+		exit 1; \
+	fi; \
+	to_seconds() { \
+		case "$$1" in \
+			*h) echo $$(( $${1%h} * 3600 )) ;; \
+			*m) echo $$(( $${1%m} * 60 )) ;; \
+			*s) echo $${1%s} ;; \
+			*) echo "$$1" ;; \
+		esac; \
+	}; \
+	timeout_secs=$$(to_seconds $(RETRY_TIMEOUT)); \
+	backoff_secs=$$(to_seconds $(RETRY_BACKOFF)); \
+	attempt=1; \
+	until perl -e 'alarm shift; exec @ARGV' $$timeout_secs $(MAKE) $(TARGET); do \
+		if [ $$attempt -ge $(RETRY_LIMIT) ]; then \
+			echo "Retry limit ($(RETRY_LIMIT)) reached for target: $(TARGET)"; \
+			exit 1; \
+		fi; \
+		echo "Attempt $$attempt of '$(TARGET)' failed. Retrying in $(RETRY_BACKOFF)..."; \
+		if [ -n "$(RETRY_CLEANUP_TARGET)" ]; then \
+			$(MAKE) $(RETRY_CLEANUP_TARGET) || true; \
+		fi; \
+		attempt=$$((attempt + 1)); \
+		sleep $$backoff_secs; \
+	done
+
 stop_pid:
 		if [ -f $(PID_FILE) ]; then \
 			kill `cat $(PID_FILE)`; \