add DMP.ignore to avoid surprises when upgrading w/ old config file

conf/cassandra.yaml

@@ -98,6 +98,7 @@ commitlog_directory: /var/lib/cassandra/commitlog
 # best_effort: stop using the failed disk and respond to requests based on
 #              remaining available sstables.  This means you WILL see obsolete
 #              data at CL.ONE!
+# ignore: ignore fatal errors and let requests fail, as in pre-1.2 Cassandra
 disk_failure_policy: stop
 
 # Maximum size of the key cache in memory.

src/java/org/apache/cassandra/config/Config.java

@@ -40,7 +40,7 @@ public class Config
     public SeedProviderDef seed_provider;
     public DiskAccessMode disk_access_mode = DiskAccessMode.auto;
 
-    public DiskFailurePolicy disk_failure_policy = DiskFailurePolicy.best_effort;
+    public DiskFailurePolicy disk_failure_policy = DiskFailurePolicy.ignore;
 
     /* initial token in the ring */
     public String initial_token;
@@ -198,7 +198,8 @@ public class Config
     public static enum DiskFailurePolicy
     {
         best_effort,
-        stop
+        stop,
+        ignore,
     }
 
     public static enum RequestSchedulerId

src/java/org/apache/cassandra/service/CassandraDaemon.java

@@ -162,6 +162,9 @@ public class CassandraDaemon
                                 Table.removeUnreadableSSTables(directory);
                         }
                         break;
+                    case ignore:
+                        // already logged, so left nothing to do
+                        break;
                     default:
                         throw new IllegalStateException();
                 }