diff --git a/CHANGELOG b/CHANGELOG
index 6894363d3c..dc0b9d0acb 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,6 +1,8 @@
 devel
 -----
 
+* DB server locking / unlocking for hot backup revisited and enhanced
+
 * Re-enabled the AQL sort-limit optimization rule in conjunction with fullCount
   in the cluster. It now also may speed up fullCount with sorted indexes and a
   limit.
@@ -8,7 +10,7 @@ devel
 * Fix config directory handling, so we don't trap into UNC path lookups on windows
 
 * Prevent spurious log message "Scheduler queue is filled more than 50% in last x s"
-  from occurring when this is not the case. Due to a data race, the message could 
+  from occurring when this is not the case. Due to a data race, the message could
   previously also occur if the queue was empty.
 
 * Make the scheduler enforce the configured queue lengths. The values of the options
@@ -34,8 +36,8 @@ devel
 * Add replicationFactor, minReplicationFactor and sharding strategy to database creation
   dialog in web UI. Preselect database default values for collection creation in web UI.
 
-* Add new JavaScipt function `db._properties()` and REST route GET 
-  `/_api/database/properties` that provides information about the current database's 
+* Add new JavaScipt function `db._properties()` and REST route GET
+  `/_api/database/properties` that provides information about the current database's
   properties.
 
 * Add new options `sharding` and `replicationFactor` for database creation methods. The
diff --git a/arangod/Cluster/ClusterMethods.cpp b/arangod/Cluster/ClusterMethods.cpp
index ddf77d12ab..3b22e7f377 100644
--- a/arangod/Cluster/ClusterMethods.cpp
+++ b/arangod/Cluster/ClusterMethods.cpp
@@ -3526,8 +3526,7 @@ arangodb::Result hotRestoreCoordinator(ClusterFeature& feature, VPackSlice const
   return arangodb::Result();
 }
 
-std::vector<std::string> lockPath =
-    std::vector<std::string>{"result", "lockId"};
+std::vector<std::string> lockPath = std::vector<std::string>{"result", "lockId"};
 
 arangodb::Result lockDBServerTransactions(std::string const& backupId,
                                           std::vector<ServerID> const& dbServers,
@@ -3549,6 +3548,7 @@ arangodb::Result lockDBServerTransactions(std::string const& backupId,
     VPackObjectBuilder o(&lock);
     lock.add("id", VPackValue(backupId));
     lock.add("timeout", VPackValue(lockWait));
+    lock.add("unlockTimeout", VPackValue(5.0 + lockWait));
   }
 
   LOG_TOPIC("707ed", DEBUG, Logger::BACKUP)
@@ -3916,10 +3916,8 @@ arangodb::Result hotBackupCoordinator(ClusterFeature& feature, VPackSlice const
     }
     std::vector<ServerID> dbServers = ci.getCurrentDBServers();
     std::vector<ServerID> lockedServers;
-    double lockWait = 2.0;
+    double lockWait(0.1);
     while (cc != nullptr && steady_clock::now() < end) {
-      auto iterEnd = steady_clock::now() + duration<double>(lockWait);
-
       result = lockDBServerTransactions(backupId, dbServers, lockWait, lockedServers);
       if (!result.ok()) {
         unlockDBServerTransactions(backupId, lockedServers);
@@ -3931,12 +3929,9 @@ arangodb::Result hotBackupCoordinator(ClusterFeature& feature, VPackSlice const
         break;
       }
       if (lockWait < 30.0) {
-        lockWait *= 1.1;
-      }
-      double tmp = duration<double>(iterEnd - steady_clock::now()).count();
-      if (tmp > 0) {
-        std::this_thread::sleep_for(duration<double>(tmp));
+        lockWait *= 1.25;
       }
+      std::this_thread::sleep_for(seconds(1));
     }
 
     bool gotLocks = result.ok();