ceph: retry the same mds later after the new session is opened

If max_mds > 1 and a request is submitted that chooses a random mds
rank, and the relating session is not opened yet, the request will wait
until the session has been opened and resend again.

Every time the request goes through __do_request, it will release the
req->session first and choose a random one again, which may be a
completely different rank than the one it just waited on.

In the worst case, it will open all the mds sessions one by one just
before the request can be successfully sent out.

Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

1 files changed, 16 insertions, 4 deletions

diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index e1902663f8b8..07ecdfc8438d 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -878,7 +878,8 @@ static struct inode *get_nonsnap_parent(struct dentry *dentry)
  * Called under mdsc->mutex.
  */
 static int __choose_mds(struct ceph_mds_client *mdsc,
-			struct ceph_mds_request *req)
+			struct ceph_mds_request *req,
+			bool *random)
 {
 	struct inode *inode;
 	struct ceph_inode_info *ci;
@@ -888,6 +889,9 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 	u32 hash = req->r_direct_hash;
 	bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
 
+	if (random)
+		*random = false;
+
 	/*
 	 * is there a specific mds we should try?  ignore hint if we have
 	 * no session and the mds is not up (active or recovering).
@@ -1023,6 +1027,9 @@ out:
 	return mds;
 
 random:
+	if (random)
+		*random = true;
+
 	mds = ceph_mdsmap_get_random_mds(mdsc->mdsmap);
 	dout("choose_mds chose random mds%d\n", mds);
 	return mds;
@@ -2551,6 +2558,7 @@ static void __do_request(struct ceph_mds_client *mdsc,
 	struct ceph_mds_session *session = NULL;
 	int mds = -1;
 	int err = 0;
+	bool random;
 
 	if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
 		if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
@@ -2590,7 +2598,7 @@ static void __do_request(struct ceph_mds_client *mdsc,
 
 	put_request_session(req);
 
-	mds = __choose_mds(mdsc, req);
+	mds = __choose_mds(mdsc, req, &random);
 	if (mds < 0 ||
 	    ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
 		dout("do_request no mds or not active, waiting for map\n");
@@ -2618,8 +2626,12 @@ static void __do_request(struct ceph_mds_client *mdsc,
 			goto out_session;
 		}
 		if (session->s_state == CEPH_MDS_SESSION_NEW ||
-		    session->s_state == CEPH_MDS_SESSION_CLOSING)
+		    session->s_state == CEPH_MDS_SESSION_CLOSING) {
 			__open_session(mdsc, session);
+			/* retry the same mds later */
+			if (random)
+				req->r_resend_mds = mds;
+		}
 		list_add(&req->r_wait, &session->s_waiting);
 		goto out_session;
 	}
@@ -2883,7 +2895,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 			mutex_unlock(&mdsc->mutex);
 			goto out;
 		} else  {
-			int mds = __choose_mds(mdsc, req);
+			int mds = __choose_mds(mdsc, req, NULL);
 			if (mds >= 0 && mds != req->r_session->s_mds) {
 				dout("but auth changed, so resending\n");
 				__do_request(mdsc, req);