[sheepdog] [PATCH] Revert "sheep: forbid revival of orphan objects"

Liu Yuan namei.unix at gmail.com
Sat Feb 28 11:42:56 CET 2015


From: Liu Yuan <liuyuan at cmss.chinamobile.com>

The reverted patch actually couldn't pass our 061, which means that we will lose
data in the following case:

0 A cluster with 3 copies
1 3 nodes of the cluster failed at the same time
2 after sometime, the cluster is recovered. As we can expect, some poor objects
  that happened to have all its copies dwell on 3 failed nodes, will not be
  recovered. Instead, all the copies are in those failed nodes. We DON'T lose
  them yet.
3 we start those 3 nodes to expect recover the previously unrecovered objects.
  Unfortunately, the reverted patch will mark all 3 nodes as excluded and not
  try to recover the objects at all. So we LOSE the objects permanentally after
  a new epoch is reached, because a cleanup operation across the cluster will be
  triggered and those good objects are cleaned up.

P.S., Merge the patch not passing our test framework is dangerouse!

Cc: Signed-off-by: Hitoshi Mitake <mitake.hitoshi at lab.ntt.co.jp>
Signed-off-by: Liu Yuan <liuyuan at cmss.chinamobile.com>
---
 sheep/group.c      | 20 ++++++--------------
 sheep/md.c         |  2 +-
 sheep/ops.c        |  4 ++--
 sheep/recovery.c   | 47 +----------------------------------------------
 sheep/sheep_priv.h |  3 +--
 5 files changed, 11 insertions(+), 65 deletions(-)

diff --git a/sheep/group.c b/sheep/group.c
index 5ab2c0c..0fce7c0 100644
--- a/sheep/group.c
+++ b/sheep/group.c
@@ -868,7 +868,7 @@ static bool membership_changed(const struct cluster_info *cinfo,
 static void update_cluster_info(const struct cluster_info *cinfo,
 				const struct sd_node *joined,
 				const struct rb_root *nroot,
-				size_t nr_nodes, enum sd_status prev_status)
+				size_t nr_nodes)
 {
 	struct vnode_info *old_vnode_info;
 
@@ -920,8 +920,6 @@ static void update_cluster_info(const struct cluster_info *cinfo,
 
 		if (membership_changed(cinfo, nroot, nr_nodes)) {
 			int ret;
-			struct sd_node *excluded = NULL;
-
 			if (old_vnode_info)
 				put_vnode_info(old_vnode_info);
 
@@ -931,17 +929,12 @@ static void update_cluster_info(const struct cluster_info *cinfo,
 				panic("cannot log current epoch %d",
 				      sys->cinfo.epoch);
 
-			if (prev_status == SD_STATUS_OK ||
-			    (prev_status == SD_STATUS_WAIT &&
-			     !node_cmp(joined, &sys->this_node)))
-				excluded = (struct sd_node *)joined;
-
 			start_recovery(main_thread_get(current_vnode_info),
-				       old_vnode_info, true, excluded);
+				       old_vnode_info, true);
 		} else if (!was_cluster_shutdowned()) {
 			start_recovery(main_thread_get(current_vnode_info),
 				       main_thread_get(current_vnode_info),
-				       false, NULL);
+				       false);
 		}
 		set_cluster_shutdown(false);
 	}
@@ -1169,7 +1162,6 @@ main_fn void sd_accept_handler(const struct sd_node *joined,
 {
 	const struct cluster_info *cinfo = opaque;
 	struct sd_node *n;
-	enum sd_status prev_status = sys->cinfo.status;
 	uint16_t flags;
 
 	if (node_is_local(joined) && sys->gateway_only
@@ -1196,7 +1188,7 @@ main_fn void sd_accept_handler(const struct sd_node *joined,
 	if (sys->cinfo.status == SD_STATUS_SHUTDOWN)
 		return;
 
-	update_cluster_info(cinfo, joined, nroot, nr_nodes, prev_status);
+	update_cluster_info(cinfo, joined, nroot, nr_nodes);
 
 	if (node_is_local(joined)) {
  		/* this output is used for testing */
@@ -1259,7 +1251,7 @@ main_fn void sd_leave_handler(const struct sd_node *left,
 		if (ret != 0)
 			panic("cannot log current epoch %d", sys->cinfo.epoch);
 		start_recovery(main_thread_get(current_vnode_info),
-			       old_vnode_info, true, NULL);
+			       old_vnode_info, true);
 	}
 
 	put_vnode_info(old_vnode_info);
@@ -1303,7 +1295,7 @@ static void kick_node_recover(void)
 	ret = inc_and_log_epoch();
 	if (ret != 0)
 		panic("cannot log current epoch %d", sys->cinfo.epoch);
-	start_recovery(main_thread_get(current_vnode_info), old, true, NULL);
+	start_recovery(main_thread_get(current_vnode_info), old, true);
 	put_vnode_info(old);
 }
 
diff --git a/sheep/md.c b/sheep/md.c
index 4f0cd49..c00d7a5 100644
--- a/sheep/md.c
+++ b/sheep/md.c
@@ -531,7 +531,7 @@ static inline void kick_recover(void)
 	if (is_cluster_diskmode(&sys->cinfo))
 		sys->cdrv->update_node(&sys->this_node);
 	else {
-		start_recovery(vinfo, vinfo, false, NULL);
+		start_recovery(vinfo, vinfo, false);
 		put_vnode_info(vinfo);
 	}
 }
diff --git a/sheep/ops.c b/sheep/ops.c
index cba5573..8084011 100644
--- a/sheep/ops.c
+++ b/sheep/ops.c
@@ -667,7 +667,7 @@ static int cluster_force_recover_main(const struct sd_req *req,
 
 	vnode_info = get_vnode_info();
 	old_vnode_info = alloc_vnode_info(&nroot);
-	start_recovery(vnode_info, old_vnode_info, true, NULL);
+	start_recovery(vnode_info, old_vnode_info, true);
 	put_vnode_info(vnode_info);
 	put_vnode_info(old_vnode_info);
 	return ret;
@@ -814,7 +814,7 @@ static int cluster_alter_vdi_copy(const struct sd_req *req, struct sd_rsp *rsp,
 	add_vdi_state(vid, nr_copies, false, 0, block_size_shift, 0);
 
 	vinfo = get_vnode_info();
-	start_recovery(vinfo, vinfo, false, NULL);
+	start_recovery(vinfo, vinfo, false);
 	put_vnode_info(vinfo);
 
 	return SD_RES_SUCCESS;
diff --git a/sheep/recovery.c b/sheep/recovery.c
index adbf5ba..8d94d4f 100644
--- a/sheep/recovery.c
+++ b/sheep/recovery.c
@@ -73,9 +73,6 @@ struct recovery_info {
 	int max_epoch;
 	struct vnode_info **vinfo_array;
 	struct sd_mutex vinfo_lock;
-
-	struct sd_node *excluded;
-
 	uint32_t max_exec_count;
 	uint64_t queue_work_interval;
 	bool throttling;
@@ -197,8 +194,6 @@ static void *read_erasure_object(uint64_t oid, uint8_t idx,
 	uint8_t policy = get_vdi_copy_policy(oid_to_vid(oid));
 	int edp = ec_policy_to_dp(policy, NULL, NULL);
 	int ret;
-	struct sd_node *excluded = row->base.rinfo->excluded;
-
 again:
 	if (unlikely(old->nr_zones < edp)) {
 		if (search_erasure_object(oid, idx, &old->nroot, rw,
@@ -213,11 +208,6 @@ again:
 		 oid, epoch, tgt_epoch, idx, node_to_str(node));
 	if (invalid_node(node, rw->cur_vinfo))
 		goto rollback;
-	if (excluded && !node_cmp(excluded, node)) {
-		sd_debug("skipping reading object from %s", node_to_str(node));
-		goto rollback;
-	}
-
 	sd_init_req(&hdr, SD_OP_READ_PEER);
 	hdr.epoch = epoch;
 	hdr.flags = SD_FLAG_CMD_RECOVERY;
@@ -332,7 +322,6 @@ static int recover_object_from_replica(struct recovery_obj_work *row,
 	uint32_t epoch = row->base.epoch;
 	int nr_copies, ret = SD_RES_SUCCESS, start = 0;
 	bool fully_replicated = true;
-	struct sd_node *excluded = row->base.rinfo->excluded;
 
 	nr_copies = get_obj_copy_number(oid, old->nr_zones);
 
@@ -358,12 +347,6 @@ static int recover_object_from_replica(struct recovery_obj_work *row,
 		if (invalid_node(node, row->base.cur_vinfo))
 			continue;
 
-		if (excluded && !node_cmp(excluded, node)) {
-			sd_debug("skipping reading from %s because it is"
-				 " excluded", node_to_str(node));
-			continue;
-		}
-
 		ret = recover_object_from(row, node, tgt_epoch);
 		switch (ret) {
 		case SD_RES_SUCCESS:
@@ -575,20 +558,12 @@ static void recover_object_work(struct work *work)
 	uint64_t oid = row->oid;
 	struct vnode_info *cur = rw->cur_vinfo;
 	int ret, epoch;
-	struct sd_node *excluded = rw->rinfo->excluded;
 
 	if (sd_store->exist(oid, local_ec_index(cur, oid))) {
 		sd_debug("the object is already recovered");
 		return;
 	}
 
-	if (excluded && !node_cmp(excluded, &sys->this_node)) {
-		sd_debug("skipping recovery from stale dir because"
-			 " I'm newly joining node (%s)",
-			 node_to_str(&sys->this_node));
-		goto remote_recover;
-	}
-
 	/* find object in the stale directory */
 	if (!is_erasure_oid(oid))
 		for (epoch = sys_epoch() - 1; epoch >= last_gathered_epoch;
@@ -602,7 +577,6 @@ static void recover_object_work(struct work *work)
 			}
 		}
 
-remote_recover:
 	ret = do_recover_object(row);
 	if (ret != 0)
 		sd_err("failed to recover object %"PRIx64, oid);
@@ -1133,7 +1107,6 @@ static void prepare_object_list(struct work *work)
 	int start = random() % nr_nodes, i, end = nr_nodes;
 	uint64_t *oids;
 	struct sd_node *nodes;
-	struct sd_node *excluded = rw->rinfo->excluded;
 
 	if (node_is_gateway_only())
 		return;
@@ -1154,13 +1127,6 @@ again:
 			goto out;
 		}
 
-		if (excluded && !node_cmp(excluded, node)) {
-			sd_info("skipping object list reading from %s because"
-				"it is marked as excluded node",
-				node_to_str(excluded));
-			continue;
-		}
-
 		oids = fetch_object_list(node, rw->epoch, &nr_oids);
 		if (!oids)
 			continue;
@@ -1179,17 +1145,8 @@ out:
 	free(nodes);
 }
 
-/*
- * notes on the parameter excluded of start_recovery():
- * When a status of a cluster is SD_STATUS_OK and a new node is joining to the
- * cluster, the recovery process should avoid reading objects from the newly
- * joining node. Because the newly joining node can have orphan objects.
- * If the cluster status is already SD_STATUS_OK, the cluster must have every
- * replica, so the newly joining node shouldn't provide any objects for the
- * recovery process.
- */
 int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo,
-		   bool epoch_lifted, struct sd_node *excluded)
+		   bool epoch_lifted)
 {
 	struct recovery_info *rinfo;
 
@@ -1214,8 +1171,6 @@ int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo,
 	rinfo->cur_vinfo = grab_vnode_info(cur_vinfo);
 	rinfo->old_vinfo = grab_vnode_info(old_vinfo);
 
-	rinfo->excluded = excluded;
-
 	if (!node_is_gateway_only())
 		sd_store->update_epoch(rinfo->tgt_epoch);
 
diff --git a/sheep/sheep_priv.h b/sheep/sheep_priv.h
index 85af7a0..26afa89 100644
--- a/sheep/sheep_priv.h
+++ b/sheep/sheep_priv.h
@@ -428,8 +428,7 @@ int init_config_file(void);
 int get_obj_list(const struct sd_req *, struct sd_rsp *, void *);
 int objlist_cache_cleanup(uint32_t vid);
 
-int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *, bool,
-		   struct sd_node *);
+int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *, bool);
 bool oid_in_recovery(uint64_t oid);
 bool node_in_recovery(void);
 void get_recovery_state(struct recovery_state *state);
-- 
1.9.1




More information about the sheepdog mailing list