ocfs2/dlm: Handle EAGAIN for compatibility - v2
authorSunil Mushran <sunil.mushran@oracle.com>
Tue, 26 Jan 2010 01:58:30 +0000 (17:58 -0800)
committerJoel Becker <joel.becker@oracle.com>
Wed, 3 Feb 2010 02:56:34 +0000 (18:56 -0800)
Mainline commit aad1b15310b9bcd59fa81ab8f2b1513b59553ea8 made the
dlm_begin_reco_handler() return -EAGAIN instead of EAGAIN.

As this error is transmitted over the wire, we want the receiver,
dlm_send_begin_reco_message(), to understand both the older EAGAIN and
the newer -EAGAIN, to allow rolling upgrade of the cluster nodes.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
fs/ocfs2/dlm/dlmrecovery.c

index cfb2ae9ab5383deaca8927a4b5131a3ab38838d9..ad712211d4eacbb4f02203fd2626fdeaacbeb717 100644 (file)
@@ -2639,7 +2639,13 @@ retry:
                             "begin reco msg (%d)\n", dlm->name, nodenum, ret);
                        ret = 0;
                }
-               if (ret == -EAGAIN) {
+
+               /*
+                * Prior to commit aad1b15310b9bcd59fa81ab8f2b1513b59553ea8,
+                * dlm_begin_reco_handler() returned EAGAIN and not -EAGAIN.
+                * We are handling both for compatibility reasons.
+                */
+               if (ret == -EAGAIN || ret == EAGAIN) {
                        mlog(0, "%s: trying to start recovery of node "
                             "%u, but node %u is waiting for last recovery "
                             "to complete, backoff for a bit\n", dlm->name,