drbd: allow for explicit resync-finished notifications
authorLars Ellenberg <lars.ellenberg@linbit.com>
Thu, 7 Oct 2010 13:55:39 +0000 (15:55 +0200)
committerPhilipp Reisner <philipp.reisner@linbit.com>
Thu, 14 Oct 2010 16:38:47 +0000 (18:38 +0200)
Preparation patch so more drbd_send_state() usage on the peer
will not confuse drbd in receive_state().

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
drivers/block/drbd/drbd_receiver.c

index 585049d..990fe01 100644 (file)
@@ -3260,6 +3260,40 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned
        os = ns = mdev->state;
        spin_unlock_irq(&mdev->req_lock);
 
+       /* peer says his disk is uptodate, while we think it is inconsistent,
+        * and this happens while we think we have a sync going on. */
+       if (os.pdsk == D_INCONSISTENT && real_peer_disk == D_UP_TO_DATE &&
+           os.conn > C_CONNECTED && os.disk == D_UP_TO_DATE) {
+               /* If we are (becoming) SyncSource, but peer is still in sync
+                * preparation, ignore its uptodate-ness to avoid flapping, it
+                * will change to inconsistent once the peer reaches active
+                * syncing states.
+                * It may have changed syncer-paused flags, however, so we
+                * cannot ignore this completely. */
+               if (peer_state.conn > C_CONNECTED &&
+                   peer_state.conn < C_SYNC_SOURCE)
+                       real_peer_disk = D_INCONSISTENT;
+
+               /* if peer_state changes to connected at the same time,
+                * it explicitly notifies us that it finished resync.
+                * Maybe we should finish it up, too? */
+               else if (os.conn >= C_SYNC_SOURCE &&
+                        peer_state.conn == C_CONNECTED) {
+                       if (drbd_bm_total_weight(mdev) <= mdev->rs_failed)
+                               drbd_resync_finished(mdev);
+                       return TRUE;
+               }
+       }
+
+       /* peer says his disk is inconsistent, while we think it is uptodate,
+        * and this happens while the peer still thinks we have a sync going on,
+        * but we think we are already done with the sync.
+        * We ignore this to avoid flapping pdsk.
+        * This should not happen, if the peer is a recent version of drbd. */
+       if (os.pdsk == D_UP_TO_DATE && real_peer_disk == D_INCONSISTENT &&
+           os.conn == C_CONNECTED && peer_state.conn > C_SYNC_SOURCE)
+               real_peer_disk = D_UP_TO_DATE;
+
        if (ns.conn == C_WF_REPORT_PARAMS)
                ns.conn = C_CONNECTED;