[Sheepdog] [PATCH 6/8] sheep: fix retry conditions of forwarding operations

[MORITA Kazutaka]

We must take into account the following cases:
  - the target node is down and epoch will change soon
  - the target node does not finish join operations

Signed-off-by: MORITA Kazutaka <morita.kazutaka at lab.ntt.co.jp>
---
 sheep/sdnet.c      |    6 ++++--
 sheep/sheep_priv.h |    2 ++
 sheep/store.c      |   10 +++++-----
 3 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/sheep/sdnet.c b/sheep/sdnet.c
index ae1c3b5..c8cbd87 100644
--- a/sheep/sdnet.c
+++ b/sheep/sdnet.c
@@ -123,8 +123,10 @@ static void __done(struct work *work, int idx)
 
 		if (!(req->rq.flags & SD_FLAG_CMD_DIRECT) &&
 		    (req->rp.result == SD_RES_OLD_NODE_VER ||
-		     req->rp.result == SD_RES_NEW_NODE_VER)) {
-
+		     req->rp.result == SD_RES_NEW_NODE_VER ||
+		     req->rp.result == SD_RES_NETWORK_ERROR ||
+		     req->rp.result == SD_RES_WAIT_FOR_JOIN ||
+		     req->rp.result == SD_RES_WAIT_FOR_FORMAT)) {
 
 			req->rq.epoch = sys->epoch;
 			req->nr_nodes = setup_ordered_sd_node_list(req);
diff --git a/sheep/sheep_priv.h b/sheep/sheep_priv.h
index 05c0052..c66baf4 100644
--- a/sheep/sheep_priv.h
+++ b/sheep/sheep_priv.h
@@ -38,6 +38,8 @@
 #define SD_STATUS_SHUTDOWN          0x03
 #define SD_STATUS_JOIN_FAILED       0x04
 
+#define SD_RES_NETWORK_ERROR    0x81 /* Network error between sheeps */
+
 enum cpg_event_type {
 	CPG_EVENT_CONCHG,
 	CPG_EVENT_DELIVER,
diff --git a/sheep/store.c b/sheep/store.c
index 2952c4e..a4d6155 100644
--- a/sheep/store.c
+++ b/sheep/store.c
@@ -431,7 +431,7 @@ static int forward_read_obj_req(struct request *req)
 
 	fd = connect_to(name, e[n].port);
 	if (fd < 0) {
-		ret = SD_RES_EIO;
+		ret = SD_RES_NETWORK_ERROR;
 		goto out;
 	}
 
@@ -443,7 +443,7 @@ static int forward_read_obj_req(struct request *req)
 	close(fd);
 
 	if (ret) /* network errors */
-		ret = SD_RES_EIO;
+		ret = SD_RES_NETWORK_ERROR;
 	else {
 		memcpy(&req->rp, rsp, sizeof(*rsp));
 		ret = rsp->result;
@@ -504,13 +504,13 @@ static int forward_write_obj_req(struct request *req)
 		fd = connect_to(name, e[n].port);
 		if (fd < 0) {
 			eprintf("failed to connect to %s:%"PRIu32"\n", name, e[n].port);
-			ret = SD_RES_EIO;
+			ret = SD_RES_NETWORK_ERROR;
 			goto out;
 		}
 
 		ret = send_req(fd, (struct sd_req *)hdr, req->data, &wlen);
 		if (ret) { /* network errors */
-			ret = SD_RES_EIO;
+			ret = SD_RES_NETWORK_ERROR;
 			dprintf("fail %"PRIu32"\n", ret);
 			goto out;
 		}
@@ -562,7 +562,7 @@ again:
 
 		if (ret) {
 			eprintf("failed to get a rsp, %m\n");
-			ret = SD_RES_EIO;
+			ret = SD_RES_NETWORK_ERROR;
 			goto out;
 		}
 
-- 
1.5.6.5