fix: drop removed modules from backup schedules

core/imageroot/var/lib/nethserver/cluster/actions/list-backups/50list

@@ -66,6 +66,9 @@ for bid, backup in backups.items():
     instances = rdb.hget(f"cluster/backup/{bid}", "instances") or ""
     for mid in instances.split():
         module_uuid = rdb.hget("cluster/module_uuid", mid)
+        if module_uuid is None:
+            # Skip stale references to modules removed from the cluster
+            continue
         path_prefix = agent.get_image_name_from_url(rdb.hget(f"module/{mid}/environment", "IMAGE_URL") or "/:")
         status = backup_statuses.get((bid, mid))
         if status is not None:

core/imageroot/var/lib/nethserver/cluster/actions/remove-module/50update

@@ -114,6 +114,15 @@ rdb.delete(f'cluster/authorizations/module/{module_id}')
 rdb.hdel('cluster/module_node', module_id)
 rdb.hdel('cluster/module_uuid', module_id)
 
+# Remove the module from any backup schedule and purge its backup status
+for kbackup in rdb.scan_iter('cluster/backup/*'):
+    instances = (rdb.hget(kbackup, 'instances') or '').split()
+    if module_id in instances:
+        instances.remove(module_id)
+        rdb.hset(kbackup, 'instances', ' '.join(instances))
+for kstatus in rdb.scan_iter('node/*/backup_status/*'):
+    rdb.hdel(kstatus, module_id)
+
 # Revoke module permissions, clean up permissions of other modules
 rdb.delete(f'roles/module/{module_id}')
 permpipe = rdb.pipeline()

core/imageroot/var/lib/nethserver/cluster/actions/remove-node/50remove_node

@@ -44,10 +44,25 @@ for xmodule_id, xnode_id in rdb.hgetall('cluster/module_node').items():
 # Save ACLs on the disk and propagate to worker nodes
 cluster.grants.save_acls(rdb)
 
+# Collect the backup schedule/status changes needed to forget the removed modules
+backup_updates = {} # {cluster/backup/<id> key: trimmed "instances" string}
+backup_status_keys = []
+if remove_module_set:
+    for kbackup in rdb.scan_iter('cluster/backup/*'):
+        instances_set = set((rdb.hget(kbackup, 'instances') or '').split())
+        if instances_set.intersection(remove_module_set):
+            backup_updates[kbackup] = ' '.join(sorted(instances_set.difference(remove_module_set)))
+    backup_status_keys = list(rdb.scan_iter('node/*/backup_status/*'))
+
 trx = rdb.pipeline()
 if remove_module_set:
     trx.hdel("cluster/module_node", *remove_module_set)
     trx.hdel("cluster/module_domains", *remove_module_set)
+    trx.hdel("cluster/module_uuid", *remove_module_set)
+    for kbackup, instances in backup_updates.items():
+        trx.hset(kbackup, 'instances', instances)
+    for kstatus in backup_status_keys:
+        trx.hdel(kstatus, *remove_module_set)
 for xmodule_id in remove_module_set:
     trx.publish('cluster/event/module-removed', json.dumps({
         'module': xmodule_id,

core/imageroot/var/lib/nethserver/cluster/update-core-pre-modules.d/60prune_removed_module_backups

@@ -0,0 +1,36 @@
+#!/usr/bin/env python3
+
+#
+# Copyright (C) 2026 Nethesis S.r.l.
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+
+import agent
+
+#
+# Bug NethServer/dev#8094: modules removed from the cluster before backup
+# schedule cleanup was implemented were left dangling in backup schedules,
+# backup status entries and cluster/module_uuid. Prune them, using
+# cluster/module_node as the authoritative source of live module ids.
+#
+
+rdb = agent.redis_connect(privileged=True)
+live_modules = set(rdb.hkeys('cluster/module_node'))
+
+trx = rdb.pipeline()
+
+dangling_mids = [mid for mid in rdb.hkeys('cluster/module_uuid') if mid not in live_modules]
+if dangling_mids:
+    trx.hdel('cluster/module_uuid', *dangling_mids)
+
+for kbackup in rdb.scan_iter('cluster/backup/*'):
+    instances_set = set((rdb.hget(kbackup, 'instances') or '').split())
+    if not instances_set.issubset(live_modules):
+        trx.hset(kbackup, 'instances', ' '.join(sorted(instances_set.intersection(live_modules))))
+
+for kstatus in rdb.scan_iter('node/*/backup_status/*'):
+    dangling_mid_hkeys = [mid for mid in rdb.hkeys(kstatus) if mid not in live_modules]
+    if dangling_mid_hkeys:
+        trx.hdel(kstatus, *dangling_mid_hkeys)
+
+trx.execute()