LU-290 Reconnects are not throttled

[fs/lustre-release.git] / lustre / ptlrpc / import.c

diff --git a/lustre/ptlrpc/import.c b/lustre/ptlrpc/import.c
index cef2a40..0934764 100644 (file)
--- a/lustre/ptlrpc/import.c
+++ b/lustre/ptlrpc/import.c
@@ -26,10 +26,13 @@
  * GPL HEADER END
  */
 /*
- * Copyright  2008 Sun Microsystems, Inc. All rights reserved
+ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
  * Use is subject to license terms.
  */
 /*
+ * Copyright (c) 2011 Whamcloud, Inc.
+ */
+/*
  * This file is part of Lustre, http://www.lustre.org/
  * Lustre is a trademark of Sun Microsystems, Inc.
  *
@@ -476,12 +479,6 @@ static int import_select_connection(struct obd_import *imp)
                        imp->imp_obd->obd_name,
                        libcfs_nid2str(conn->oic_conn->c_peer.nid),
                        conn->oic_last_attempt);
-                /* Don't thrash connections */
-                if (cfs_time_before_64(cfs_time_current_64(),
-                                     conn->oic_last_attempt +
-                                     cfs_time_seconds(CONNECTION_SWITCH_MIN))) {
-                        continue;
-                }
 
                 /* If we have not tried this connection since
                    the last successful attempt, go with this one */
@@ -732,42 +729,12 @@ EXPORT_SYMBOL(ptlrpc_connect_import);
 static void ptlrpc_maybe_ping_import_soon(struct obd_import *imp)
 {
 #ifdef __KERNEL__
-        struct obd_import_conn *imp_conn;
-#endif
-        int wake_pinger = 0;
-
-        ENTRY;
-
-        cfs_spin_lock(&imp->imp_lock);
-        if (cfs_list_empty(&imp->imp_conn_list))
-                GOTO(unlock, 0);
-
-#ifdef __KERNEL__
-        imp_conn = cfs_list_entry(imp->imp_conn_list.prev,
-                                  struct obd_import_conn,
-                                  oic_item);
-
-        /* XXX: When the failover node is the primary node, it is possible
-         * to have two identical connections in imp_conn_list. We must
-         * compare not conn's pointers but NIDs, otherwise we can defeat
-         * connection throttling. (See bug 14774.) */
-        if (imp->imp_conn_current->oic_conn->c_peer.nid !=
-                                imp_conn->oic_conn->c_peer.nid) {
-                ptlrpc_ping_import_soon(imp);
-                wake_pinger = 1;
-        }
+        /* the pinger takes care of issuing the next reconnect request */
+        return;
 #else
         /* liblustre has no pinger thread, so we wakeup pinger anyway */
-        wake_pinger = 1;
+        ptlrpc_pinger_wake_up();
 #endif
-
- unlock:
-        cfs_spin_unlock(&imp->imp_lock);
-
-        if (wake_pinger)
-                ptlrpc_pinger_wake_up();
-
-        EXIT;
 }
 
 static int ptlrpc_busy_reconnect(int rc)
@@ -1071,10 +1038,12 @@ finish:
                         cli->cl_cksum_type = OBD_CKSUM_CRC32;
                 }
 
-                if (ocd->ocd_connect_flags & OBD_CONNECT_BRW_SIZE) {
+                if (ocd->ocd_connect_flags & OBD_CONNECT_BRW_SIZE)
                         cli->cl_max_pages_per_rpc =
                                 ocd->ocd_brw_size >> CFS_PAGE_SHIFT;
-                }
+                else if (imp->imp_connect_op == MDS_CONNECT ||
+                         imp->imp_connect_op == MGS_CONNECT)
+                        cli->cl_max_pages_per_rpc = 1;
 
                 /* Reset ns_connect_flags only for initial connect. It might be
                  * changed in while using FS and if we reset it in reconnect
@@ -1209,6 +1178,9 @@ static int signal_completed_replay(struct obd_import *imp)
         struct ptlrpc_request *req;
         ENTRY;
 
+        if (unlikely(OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_FINISH_REPLAY)))
+                RETURN(0);
+
         LASSERT(cfs_atomic_read(&imp->imp_replay_inflight) == 0);
         cfs_atomic_inc(&imp->imp_replay_inflight);
 
@@ -1273,7 +1245,7 @@ static int ptlrpc_invalidate_import_thread(void *data)
  * If we came to server that is in recovery, we enter IMP_REPLAY import state.
  * We go through our list of requests to replay and send them to server one by
  * one.
- * After sending all request from the list we change import state to 
+ * After sending all request from the list we change import state to
  * IMP_REPLAY_LOCKS and re-request all the locks we believe we have from server
  * and also all the locks we don't yet have and wait for server to grant us.
  * After that we send a special "replay completed" request and change import
@@ -1317,8 +1289,8 @@ int ptlrpc_import_recovery_state_machine(struct obd_import *imp)
                  * invalidate thread without reference to import and import can
                  * be freed at same time. */
                 class_import_get(imp);
-                rc = cfs_kernel_thread(ptlrpc_invalidate_import_thread, imp,
-                                       CLONE_VM | CLONE_FILES);
+                rc = cfs_create_thread(ptlrpc_invalidate_import_thread, imp,
+                                       CFS_DAEMON_FLAGS);
                 if (rc < 0) {
                         class_import_put(imp);
                         CERROR("error starting invalidate thread: %d\n", rc);