ocfs2: better error handling during assert master message
authorKurt Hackel <kurt.hackel@oracle.com>
Fri, 28 Apr 2006 02:26:15 +0000 (19:26 -0700)
committerMark Fasheh <mark.fasheh@oracle.com>
Mon, 26 Jun 2006 21:42:57 +0000 (14:42 -0700)
handle errors during lock assert master by either killing self or other node

Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
fs/ocfs2/dlm/dlmmaster.c

index 427c0af0d219100c4bb5bbf270edf1ec6d66ecdd..81ceee2e0d50312d98bf89c597606eaf31250110 100644 (file)
@@ -1633,6 +1633,8 @@ again:
        dlm_node_iter_init(nodemap, &iter);
        while ((to = dlm_node_iter_next(&iter)) >= 0) {
                int r = 0;
+               struct dlm_master_list_entry *mle = NULL;
+
                mlog(0, "sending assert master to %d (%.*s)\n", to,
                     namelen, lockname);
                memset(&assert, 0, sizeof(assert));
@@ -1657,7 +1659,15 @@ again:
                        /* ok, something horribly messed.  kill thyself. */
                        mlog(ML_ERROR,"during assert master of %.*s to %u, "
                             "got %d.\n", namelen, lockname, to, r);
-                       dlm_dump_lock_resources(dlm);
+                       spin_lock(&dlm->spinlock);
+                       spin_lock(&dlm->master_lock);
+                       if (dlm_find_mle(dlm, &mle, (char *)lockname,
+                                        namelen)) {
+                               dlm_print_one_mle(mle);
+                               __dlm_put_mle(mle);
+                       }
+                       spin_unlock(&dlm->master_lock);
+                       spin_unlock(&dlm->spinlock);
                        BUG();
                } else if (r == EAGAIN) {
                        mlog(0, "%.*s: node %u create mles on other "
@@ -1922,12 +1932,12 @@ done:
 
 kill:
        /* kill the caller! */
+       mlog(ML_ERROR, "Bad message received from another node.  Dumping state "
+            "and killing the other node now!  This node is OK and can continue.\n");
+       __dlm_print_one_lock_resource(res);
        spin_unlock(&res->spinlock);
        spin_unlock(&dlm->spinlock);
        dlm_lockres_put(res);
-       mlog(ML_ERROR, "Bad message received from another node.  Dumping state "
-            "and killing the other node now!  This node is OK and can continue.\n");
-       dlm_dump_lock_resources(dlm);
        dlm_put(dlm);
        return -EINVAL;
 }