fix: add clean-data playbook to prevent storage-related deployment failures

ansible/playbooks/clean-data.yml

@@ -0,0 +1,114 @@
+---
+# Clean data playbook: Clean node data directories
+
+- name: Parse and validate node names
+  hosts: localhost
+  connection: local
+  gather_facts: no
+  vars:
+    validator_config_file: "{{ genesis_dir }}/validator-config.yaml"
+  tags:
+    - zeam
+    - ream
+    - qlean
+    - lantern
+    - lighthouse
+    - deploy
+
+  tasks:
+    - name: Validate validator-config.yaml exists
+      stat:
+        path: "{{ validator_config_file }}"
+      register: validator_config_stat
+
+    - name: Fail if validator-config.yaml missing
+      fail:
+        msg: "validator-config.yaml not found at {{ validator_config_file }}"
+      when: not validator_config_stat.stat.exists
+
+    - name: Extract all node names
+      shell: |
+        yq eval '.validators[].name' {{ validator_config_file }}
+      register: all_node_names_raw
+      changed_when: false
+
+    - name: Set all node names
+      set_fact:
+        all_node_names: "{{ all_node_names_raw.stdout_lines }}"
+
+    - name: Fail if node_names is not specified
+      fail:
+        msg: "node_names must be specified. Provide one or more node names (comma or space separated)."
+      when: node_names is not defined or node_names == ""
+
+    - name: Handle "all" node names - expand to all nodes
+      set_fact:
+        clean_nodes: "{{ all_node_names }}"
+      when:
+        - node_names is defined
+        - node_names == "all"
+
+    - name: Parse node names if provided as comma-separated string
+      set_fact:
+        clean_nodes: "{{ node_names.split(',') | map('trim') | list }}"
+      when:
+        - node_names is defined
+        - node_names is string
+        - node_names != "all"
+        - '("," in node_names)'
+
+    - name: Parse node names if provided as space-separated string
+      set_fact:
+        clean_nodes: "{{ node_names.split(' ') | map('trim') | select('length') | list }}"
+      when:
+        - node_names is defined
+        - node_names is string
+        - node_names != "all"
+        - '("," not in node_names)'
+        - '(" " in node_names)'
+
+    - name: Handle single node name
+      set_fact:
+        clean_nodes: "{{ [node_names] }}"
+      when:
+        - node_names is defined
+        - node_names is string
+        - node_names != "all"
+        - '"," not in node_names'
+        - '" " not in node_names'
+
+    - name: Display nodes to clean
+      debug:
+        msg: "Cleaning data directories for nodes: {{ clean_nodes | join(', ') }}"
+
+    - name: Add nodes to clean_targets group
+      add_host:
+        name: "{{ item }}"
+        groups: clean_targets
+      loop: "{{ clean_nodes }}"
+
+- name: Clean node data directories on remote hosts
+  hosts: clean_targets
+  gather_facts: no
+  vars:
+    # Use remote paths on remote hosts
+    data_dir: "{{ remote_data_dir | default('/opt/lean-quickstart/data') }}"
+    node_name: "{{ inventory_hostname }}"
+  tags:
+    - zeam
+    - ream
+    - qlean
+    - lantern
+    - lighthouse
+    - deploy
+
+  tasks:
+    - name: Clean node data directory
+      file:
+        path: "{{ data_dir }}/{{ node_name }}"
+        state: absent
+      become: yes
+
+    - name: Display cleaned directory
+      debug:
+        msg: "Cleaned data directory: {{ data_dir }}/{{ node_name }}"

ansible/playbooks/site.yml

@@ -1,8 +1,9 @@
 ---
 # Main site playbook: Complete deployment workflow
-# 1. Generate genesis files locally (including .key files)
-# 2. Copy genesis files to remote hosts
-# 3. Deploy nodes
+# 1. Clean data directories (if clean_data=true)
+# 2. Generate genesis files locally (including .key files)
+# 3. Copy genesis files to remote hosts
+# 4. Deploy nodes
 #
 # Usage:
 #   ansible-playbook -i inventory/hosts.yml playbooks/site.yml \
@@ -13,9 +14,17 @@
 #   network_dir      - Path to network directory (genesis_dir derived from this)
 #   genesis_dir      - Path to genesis directory (required)
 #   node_names       - Nodes to deploy: "all" or comma-separated names
+#   clean_data       - Set to true to clean data directories before deployment (default: false)
 #   skip_genesis     - Set to true to skip genesis generation (default: false)
 #   genesis_offset   - Seconds to add to current time for genesis (default: 360 for ansible mode)
 
+- name: Clean Data Directories
+  import_playbook: clean-data.yml
+  vars:
+    genesis_dir: "{{ network_dir }}/genesis"
+    node_names: "{{ node_names }}"
+  when: clean_data | default(false) | bool
+
 - name: Generate Genesis Files
   import_playbook: generate-genesis.yml
   vars:

ansible/roles/lantern/tasks/main.yml

@@ -60,11 +60,22 @@
     msg: "Node key file {{ node_name }}.key not found in {{ genesis_dir }}"
   when: not (node_key_stat.stat.exists | default(false))
 
+- name: Check if node data directory has contents
+  find:
+    paths: "{{ data_dir }}/{{ node_name }}"
+    file_type: any
+    hidden: yes
+  register: node_data_contents
+  failed_when: false
+  when: clean_data | default(false) | bool
+
 - name: Clean node data directory
   file:
     path: "{{ data_dir }}/{{ node_name }}"
     state: absent
-  when: clean_data | default(false) | bool
+  when:
+    - clean_data | default(false) | bool
+    - node_data_contents.matched | default(0) > 0
 
 - name: Create node data directory
   file:

ansible/roles/lighthouse/tasks/main.yml

@@ -60,11 +60,22 @@
     msg: "Node key file {{ node_name }}.key not found in {{ genesis_dir }}"
   when: not (node_key_stat.stat.exists | default(false))
 
+- name: Check if node data directory has contents
+  find:
+    paths: "{{ data_dir }}/{{ node_name }}"
+    file_type: any
+    hidden: yes
+  register: node_data_contents
+  failed_when: false
+  when: clean_data | default(false) | bool
+
 - name: Clean node data directory
   file:
     path: "{{ data_dir }}/{{ node_name }}"
     state: absent
-  when: clean_data | default(false) | bool
+  when:
+    - clean_data | default(false) | bool
+    - node_data_contents.matched | default(0) > 0
 
 - name: Create node data directory
   file:

ansible/roles/qlean/tasks/main.yml

@@ -72,11 +72,22 @@
     msg: "Node key file {{ node_name }}.key not found in {{ genesis_dir }}"
   when: not (node_key_stat.stat.exists | default(false))
 
+- name: Check if node data directory has contents
+  find:
+    paths: "{{ data_dir }}/{{ node_name }}"
+    file_type: any
+    hidden: yes
+  register: node_data_contents
+  failed_when: false
+  when: clean_data | default(false) | bool
+
 - name: Clean node data directory
   file:
     path: "{{ data_dir }}/{{ node_name }}"
     state: absent
-  when: clean_data | default(false) | bool
+  when:
+    - clean_data | default(false) | bool
+    - node_data_contents.matched | default(0) > 0
 
 - name: Create node data directory
   file:

ansible/roles/ream/tasks/main.yml

@@ -60,11 +60,22 @@
     msg: "Node key file {{ node_name }}.key not found in {{ genesis_dir }}"
   when: not (node_key_stat.stat.exists | default(false))
 
+- name: Check if node data directory has contents
+  find:
+    paths: "{{ data_dir }}/{{ node_name }}"
+    file_type: any
+    hidden: yes
+  register: node_data_contents
+  failed_when: false
+  when: clean_data | default(false) | bool
+
 - name: Clean node data directory
   file:
     path: "{{ data_dir }}/{{ node_name }}"
     state: absent
-  when: clean_data | default(false) | bool
+  when:
+    - clean_data | default(false) | bool
+    - node_data_contents.matched | default(0) > 0
 
 - name: Create node data directory
   file:

ansible/roles/zeam/tasks/main.yml

@@ -65,11 +65,22 @@
     msg: "Node key file {{ node_name }}.key not found in {{ genesis_dir }}"
   when: not (node_key_stat.stat.exists | default(false))
 
+- name: Check if node data directory has contents
+  find:
+    paths: "{{ data_dir }}/{{ node_name }}"
+    file_type: any
+    hidden: yes
+  register: node_data_contents
+  failed_when: false
+  when: clean_data | default(false) | bool
+
 - name: Clean node data directory
   file:
     path: "{{ data_dir }}/{{ node_name }}"
     state: absent
-  when: clean_data | default(false) | bool
+  when:
+    - clean_data | default(false) | bool
+    - node_data_contents.matched | default(0) > 0
 
 - name: Create node data directory
   file: