[sheepdog] [PATCH v4 06/10] sheep: fetch vdi copy list after sheep joins the cluster
levin li
levin108 at gmail.com
Thu Aug 9 07:27:41 CEST 2012
From: levin li <xingke.lwp at taobao.com>
The new joined node doesn't have the vdi copy list, or have
incomplete vdi copy list, so we need to fetch the copy list
data from other nodes
Signed-off-by: levin li <xingke.lwp at taobao.com>
---
include/internal_proto.h | 1 +
sheep/group.c | 11 +++++--
sheep/ops.c | 24 +++++++++++++--
sheep/recovery.c | 42 +++++++++++++++++++++++++--
sheep/sheep_priv.h | 5 ++-
sheep/vdi.c | 70 ++++++++++++++++++++++++++++++++++++++++++++++
6 files changed, 142 insertions(+), 11 deletions(-)
diff --git a/include/internal_proto.h b/include/internal_proto.h
index 83d98f1..3d70ba9 100644
--- a/include/internal_proto.h
+++ b/include/internal_proto.h
@@ -63,6 +63,7 @@
#define SD_OP_ENABLE_RECOVER 0xA8
#define SD_OP_DISABLE_RECOVER 0xA9
#define SD_OP_INFO_RECOVER 0xAA
+#define SD_OP_GET_VDI_COPIES 0xAB
/* internal flags for hdr.flags, must be above 0x80 */
#define SD_FLAG_CMD_RECOVERY 0x0080
diff --git a/sheep/group.c b/sheep/group.c
index 05ffb3e..fd631ec 100644
--- a/sheep/group.c
+++ b/sheep/group.c
@@ -853,6 +853,8 @@ static void update_cluster_info(struct join_message *msg,
if (msg->inc_epoch) {
if (!sys->disable_recovery) {
+ int is_newly_joined = 0;
+
uatomic_inc(&sys->epoch);
log_current_epoch();
clear_exceptional_node_lists();
@@ -863,8 +865,11 @@ static void update_cluster_info(struct join_message *msg,
nodes, nr_nodes);
}
- start_recovery(current_vnode_info,
- old_vnode_info);
+ if (node_eq(joined, &sys->this_node))
+ is_newly_joined = 1;
+
+ start_recovery(current_vnode_info, old_vnode_info,
+ is_newly_joined);
} else
prepare_recovery(joined, nodes, nr_nodes);
}
@@ -1148,7 +1153,7 @@ void sd_leave_handler(struct sd_node *left, struct sd_node *members,
case SD_STATUS_OK:
uatomic_inc(&sys->epoch);
log_current_epoch();
- start_recovery(current_vnode_info, old_vnode_info);
+ start_recovery(current_vnode_info, old_vnode_info, 0);
if (!have_enough_zones())
sys->status = SD_STATUS_HALT;
diff --git a/sheep/ops.c b/sheep/ops.c
index efaf979..ce0f8a4 100644
--- a/sheep/ops.c
+++ b/sheep/ops.c
@@ -282,13 +282,16 @@ static int cluster_shutdown(const struct sd_req *req, struct sd_rsp *rsp,
static int cluster_enable_recover(const struct sd_req *req,
struct sd_rsp *rsp, void *data)
{
- int i;
+ int i, is_newly_joined = 0;
struct vnode_info *old_vnode_info, *vnode_info;
if (nr_joining_nodes) {
- for (i = 0; i < nr_joining_nodes; i++)
+ for (i = 0; i < nr_joining_nodes; i++) {
all_nodes[nr_all_nodes++] = joining_nodes[i];
+ if (node_eq(&joining_nodes[i], &sys->this_node))
+ is_newly_joined = 1;
+ }
old_vnode_info = get_vnode_info();
vnode_info = alloc_vnode_info(all_nodes, nr_all_nodes);
@@ -298,7 +301,7 @@ static int cluster_enable_recover(const struct sd_req *req,
log_current_epoch();
clear_exceptional_node_lists();
- start_recovery(vnode_info, old_vnode_info);
+ start_recovery(vnode_info, old_vnode_info, is_newly_joined);
put_vnode_info(old_vnode_info);
}
@@ -447,6 +450,13 @@ static int local_get_obj_list(struct request *req)
(struct sd_list_rsp *)&req->rp, req->data);
}
+static int local_get_vdi_copies(struct request *req)
+{
+ req->rp.data_length = fill_vdi_copy_list(req->data);
+
+ return SD_RES_SUCCESS;
+}
+
static int local_get_epoch(struct request *req)
{
uint32_t epoch = req->rq.obj.tgt_epoch;
@@ -510,7 +520,7 @@ static int cluster_force_recover(const struct sd_req *req, struct sd_rsp *rsp,
sys->status = SD_STATUS_HALT;
vnode_info = get_vnode_info();
- start_recovery(vnode_info, old_vnode_info);
+ start_recovery(vnode_info, old_vnode_info, 1);
put_vnode_info(vnode_info);
out:
put_vnode_info(old_vnode_info);
@@ -989,6 +999,12 @@ static struct sd_op_template sd_ops[] = {
.process_work = local_get_obj_list,
},
+ [SD_OP_GET_VDI_COPIES] = {
+ .name = "GET_VDI_COPIES",
+ .type = SD_OP_TYPE_LOCAL,
+ .process_work = local_get_vdi_copies,
+ },
+
[SD_OP_GET_EPOCH] = {
.name = "GET_EPOCH",
.type = SD_OP_TYPE_LOCAL,
diff --git a/sheep/recovery.c b/sheep/recovery.c
index 5164aa7..3fdcad2 100644
--- a/sheep/recovery.c
+++ b/sheep/recovery.c
@@ -592,7 +592,38 @@ static inline bool node_is_gateway_only(void)
return sys->this_node.nr_vnodes == 0 ? true : false;
}
-int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo)
+static void prepare_vdi_copy_list(struct work *work)
+{
+ struct recovery_work *rw = container_of(work, struct recovery_work,
+ work);
+ struct sd_node *nodes = rw->cur_vinfo->nodes;
+ int i, nr_nodes = rw->cur_vinfo->nr_nodes;
+
+ for (i = 0; i < nr_nodes; i++) {
+ if (node_eq(nodes + i, &sys->this_node))
+ continue;
+
+ fetch_vdi_copies_from(nodes + i);
+ }
+}
+
+static void finish_vdi_copy_list(struct work *work)
+{
+ struct recovery_work *rw = container_of(work, struct recovery_work,
+ work);
+
+ if (next_rw) {
+ run_next_rw(rw);
+ return;
+ }
+
+ rw->work.fn = prepare_object_list;
+ rw->work.done = finish_object_list;
+ queue_work(sys->recovery_wqueue, &rw->work);
+}
+
+int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo,
+ int is_newly_joined)
{
struct recovery_work *rw;
@@ -613,8 +644,13 @@ int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo)
rw->cur_vinfo = grab_vnode_info(cur_vinfo);
rw->old_vinfo = grab_vnode_info(old_vinfo);
- rw->work.fn = prepare_object_list;
- rw->work.done = finish_object_list;
+ if (is_newly_joined) {
+ rw->work.fn = prepare_vdi_copy_list;
+ rw->work.done = finish_vdi_copy_list;
+ } else {
+ rw->work.fn = prepare_object_list;
+ rw->work.done = finish_object_list;
+ }
if (sd_store->begin_recover) {
struct siocb iocb = { 0 };
diff --git a/sheep/sheep_priv.h b/sheep/sheep_priv.h
index 3f763c4..335e337 100644
--- a/sheep/sheep_priv.h
+++ b/sheep/sheep_priv.h
@@ -207,6 +207,8 @@ int get_max_copy_number(void);
int get_req_copy_number(struct request *req);
int add_vdi_copies(uint32_t vid, int nr_copies);
int load_vdi_copies(void);
+int fetch_vdi_copies_from(struct sd_node *node);
+int fill_vdi_copy_list(void *data);
int vdi_exist(uint32_t vid);
int add_vdi(struct vdi_iocb *iocb, uint32_t *new_vid);
@@ -272,7 +274,8 @@ uint64_t get_cluster_ctime(void);
int get_obj_list(const struct sd_list_req *, struct sd_list_rsp *, void *);
int objlist_cache_cleanup(uint32_t vid);
-int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo);
+int start_recovery(struct vnode_info *cur_vinfo, struct vnode_info *old_vinfo,
+ int is_newly_joined);
void resume_recovery_work(void);
bool oid_in_recovery(uint64_t oid);
int is_recovery_init(void);
diff --git a/sheep/vdi.c b/sheep/vdi.c
index 0581ad6..72fbd7b 100644
--- a/sheep/vdi.c
+++ b/sheep/vdi.c
@@ -18,6 +18,11 @@
#include "sheepdog_proto.h"
#include "sheep_priv.h"
+struct vdi_copy {
+ uint32_t vid;
+ uint32_t nr_copies;
+};
+
struct vdi_copy_entry {
uint32_t vid;
unsigned int nr_copies;
@@ -226,6 +231,71 @@ out:
return ret;
}
+int fetch_vdi_copies_from(struct sd_node *node)
+{
+ char host[128];
+ struct sd_req hdr;
+ struct sd_rsp *rsp = (struct sd_rsp *)&hdr;
+ struct vdi_copy *vc;
+ int fd, ret, i;
+ unsigned int wlen, rlen;
+ int count = 1 << 15;
+
+ addr_to_str(host, sizeof(host), node->nid.addr, 0);
+ dprintf("fetch vdi copy list from %s:%d\n", host, node->nid.port);
+ fd = connect_to(host, node->nid.port);
+ if (fd < 0) {
+ dprintf("fail: %m\n");
+ return SD_RES_NETWORK_ERROR;
+ }
+
+ sd_init_req(&hdr, SD_OP_GET_VDI_COPIES);
+ hdr.epoch = sys->epoch;
+ hdr.data_length = count * sizeof(*vc);
+ rlen = hdr.data_length;
+ wlen = 0;
+
+ vc = xzalloc(rlen);
+
+ ret = exec_req(fd, &hdr, (char *)vc, &wlen, &rlen);
+ close(fd);
+
+ if (ret || rsp->result != SD_RES_SUCCESS) {
+ eprintf("fail to get VDI copy list (%d, %d)\n",
+ ret, rsp->result);
+ ret = SD_RES_NETWORK_ERROR;
+ goto out;
+ }
+
+ count = rsp->data_length / sizeof(*vc);
+ dprintf("got %d vdi copy data\n", count);
+ for (i = 0; i < count; i++)
+ add_vdi_copies(vc[i].vid, vc[i].nr_copies);
+out:
+ free(vc);
+ return ret;
+}
+
+int fill_vdi_copy_list(void *data)
+{
+ int nr = 0;
+ struct rb_node *n;
+ struct vdi_copy *vc = data;
+ struct vdi_copy_entry *entry;
+
+ pthread_rwlock_rdlock(&vdi_copy_lock);
+ for (n = rb_first(&vdi_copy_root); n; n = rb_next(n)) {
+ entry = rb_entry(n, struct vdi_copy_entry, node);
+ vc->vid = entry->vid;
+ vc->nr_copies = entry->nr_copies;
+ vc++;
+ nr++;
+ }
+ pthread_rwlock_unlock(&vdi_copy_lock);
+
+ return nr * sizeof(*vc);
+}
+
int vdi_exist(uint32_t vid)
{
struct sheepdog_inode *inode;
--
1.7.1
More information about the sheepdog
mailing list