[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Linux-cluster] node kicked out of cluster



You might like to try this patch. It seems to reduce the number of cman retries
considerably in my testing.

-- 

patrick


Index: cman-kernel/src/cnxman.c
===================================================================
RCS file: /cvs/cluster/cluster/cman-kernel/src/cnxman.c,v
retrieving revision 1.51
diff -u -p -r1.51 cnxman.c
--- cman-kernel/src/cnxman.c    25 Feb 2005 10:16:36 -0000      1.51
+++ cman-kernel/src/cnxman.c    25 Feb 2005 14:48:44 -0000
@@ -340,10 +340,6 @@ static int cluster_kthread(void *unused)
                if (quit_threads)
                        break;
 
-               if (test_and_clear_bit(ACK_TIMEOUT, &mainloop_flags)) {
-                       check_for_unacked_nodes();
-               }
-
                /* Now receive any messages waiting for us */
                spin_lock_irq(&active_socket_lock);
                list_for_each_safe(socklist, temp, &active_socket_list) {
@@ -368,6 +364,10 @@ static int cluster_kthread(void *unused)
                }
                spin_unlock_irq(&active_socket_lock);
 
+               if (test_and_clear_bit(ACK_TIMEOUT, &mainloop_flags)) {
+                       check_for_unacked_nodes();
+               }
+
                /* Resend any unacked messages */
                if (test_and_clear_bit(RESEND_NEEDED, &mainloop_flags)
                    && acks_expected) {



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]