bdev_nvme: add timeout parameters to start_discovery RPC

These parameters will be used for any controller created
by the discovery service.

Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I221b791f38b9c5797ba084c647a98b82c102a121
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11942
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>

doc/jsonrpc.md

@@ -3227,6 +3227,9 @@ traddr                     | Required | string      | NVMe-oF target address: ip
 adrfam                     | Optional | string      | NVMe-oF target adrfam: ipv4, ipv6
 trsvcid                    | Optional | string      | NVMe-oF target trsvcid: port number
 hostnqn                    | Optional | string      | NVMe-oF target hostnqn
+ctrlr_loss_timeout_sec     | Optional | number      | Time to wait until ctrlr is reconnected before deleting ctrlr.  -1 means infinite reconnects. 0 means no reconnect.
+reconnect_delay_sec        | Optional | number      | Time to delay a reconnect trial. 0 means no reconnect.
+fast_io_fail_timeout_sec   | Optional | number      | Time to wait until ctrlr is reconnected before failing I/O to ctrlr. 0 means no such timeout.
 
 #### Example
 

module/bdev/nvme/bdev_nvme.c

@@ -4332,6 +4332,7 @@ struct discovery_ctx {
 	struct discovery_entry_ctx		*entry_ctx_in_use;
 	struct spdk_poller			*poller;
 	struct spdk_nvme_ctrlr_opts		drv_opts;
+	struct nvme_ctrlr_opts			bdev_opts;
 	struct spdk_nvmf_discovery_log_page	*log_page;
 	TAILQ_ENTRY(discovery_ctx)		tailq;
 	TAILQ_HEAD(, discovery_entry_ctx)	nvm_entry_ctxs;
@@ -4568,7 +4569,7 @@ discovery_log_page_cb(void *cb_arg, int rc, const struct spdk_nvme_cpl *cpl,
 			snprintf(new_ctx->drv_opts.hostnqn, sizeof(new_ctx->drv_opts.hostnqn), "%s", ctx->hostnqn);
 			rc = bdev_nvme_create(&new_ctx->trid, new_ctx->name, NULL, 0,
 					      discovery_attach_controller_done, new_ctx,
-					      &new_ctx->drv_opts, NULL, true);
+					      &new_ctx->drv_opts, &ctx->bdev_opts, true);
 			if (rc == 0) {
 				TAILQ_INSERT_TAIL(&ctx->nvm_entry_ctxs, new_ctx, tailq);
 				ctx->attach_in_progress++;
@@ -4716,7 +4717,8 @@ start_discovery_poller(void *arg)
 int
 bdev_nvme_start_discovery(struct spdk_nvme_transport_id *trid,
 			  const char *base_name,
-			  struct spdk_nvme_ctrlr_opts *drv_opts)
+			  struct spdk_nvme_ctrlr_opts *drv_opts,
+			  struct nvme_ctrlr_opts *bdev_opts)
 {
 	struct discovery_ctx *ctx;
 	struct discovery_entry_ctx *discovery_entry_ctx;
@@ -4732,6 +4734,7 @@ bdev_nvme_start_discovery(struct spdk_nvme_transport_id *trid,
 		return -ENOMEM;
 	}
 	memcpy(&ctx->drv_opts, drv_opts, sizeof(*drv_opts));
+	memcpy(&ctx->bdev_opts, bdev_opts, sizeof(*bdev_opts));
 	ctx->calling_thread = spdk_get_thread();
 	TAILQ_INIT(&ctx->nvm_entry_ctxs);
 	TAILQ_INIT(&ctx->discovery_entry_ctxs);

module/bdev/nvme/bdev_nvme.h

@@ -282,7 +282,7 @@ int bdev_nvme_create(struct spdk_nvme_transport_id *trid,
 		     bool multipath);
 
 int bdev_nvme_start_discovery(struct spdk_nvme_transport_id *trid, const char *base_name,
-			      struct spdk_nvme_ctrlr_opts *drv_opts);
+			      struct spdk_nvme_ctrlr_opts *drv_opts, struct nvme_ctrlr_opts *bdev_opts);
 int bdev_nvme_stop_discovery(const char *name, spdk_bdev_nvme_stop_discovery_fn cb_fn,
 			     void *cb_ctx);
 

module/bdev/nvme/bdev_nvme_rpc.c

@@ -1607,6 +1607,7 @@ struct rpc_bdev_nvme_start_discovery {
 	char *trsvcid;
 	char *hostnqn;
 	struct spdk_nvme_ctrlr_opts opts;
+	struct nvme_ctrlr_opts bdev_opts;
 };
 
 static void
@@ -1627,6 +1628,9 @@ static const struct spdk_json_object_decoder rpc_bdev_nvme_start_discovery_decod
 	{"adrfam", offsetof(struct rpc_bdev_nvme_start_discovery, adrfam), spdk_json_decode_string, true},
 	{"trsvcid", offsetof(struct rpc_bdev_nvme_start_discovery, trsvcid), spdk_json_decode_string, true},
 	{"hostnqn", offsetof(struct rpc_bdev_nvme_start_discovery, hostnqn), spdk_json_decode_string, true},
+	{"ctrlr_loss_timeout_sec", offsetof(struct rpc_bdev_nvme_start_discovery, bdev_opts.ctrlr_loss_timeout_sec), spdk_json_decode_int32, true},
+	{"reconnect_delay_sec", offsetof(struct rpc_bdev_nvme_start_discovery, bdev_opts.reconnect_delay_sec), spdk_json_decode_uint32, true},
+	{"fast_io_fail_timeout_sec", offsetof(struct rpc_bdev_nvme_start_discovery, bdev_opts.fast_io_fail_timeout_sec), spdk_json_decode_uint32, true},
 };
 
 struct rpc_bdev_nvme_start_discovery_ctx {
@@ -1712,7 +1716,7 @@ rpc_bdev_nvme_start_discovery(struct spdk_jsonrpc_request *request,
 	}
 
 	ctx->request = request;
-	rc = bdev_nvme_start_discovery(&trid, ctx->req.name, &ctx->req.opts);
+	rc = bdev_nvme_start_discovery(&trid, ctx->req.name, &ctx->req.opts, &ctx->req.bdev_opts);
 	if (rc == 0) {
 		spdk_jsonrpc_send_bool_response(ctx->request, true);
 	} else {

scripts/rpc.py

@@ -693,7 +693,10 @@ if __name__ == "__main__":
                                            traddr=args.traddr,
                                            adrfam=args.adrfam,
                                            trsvcid=args.trsvcid,
-                                           hostnqn=args.hostnqn)
+                                           hostnqn=args.hostnqn,
+                                           ctrlr_loss_timeout_sec=args.ctrlr_loss_timeout_sec,
+                                           reconnect_delay_sec=args.reconnect_delay_sec,
+                                           fast_io_fail_timeout_sec=args.fast_io_fail_timeout_sec)
 
     p = subparsers.add_parser('bdev_nvme_start_discovery', help='Start automatic discovery')
     p.add_argument('-b', '--name', help="Name of the NVMe controller prefix for each bdev name", required=True)
@@ -706,6 +709,26 @@ if __name__ == "__main__":
     p.add_argument('-s', '--trsvcid',
                    help='NVMe-oF target trsvcid: e.g., a port number')
     p.add_argument('-q', '--hostnqn', help='NVMe-oF host subnqn')
+    p.add_argument('-l', '--ctrlr-loss-timeout-sec',
+                   help="""Time to wait until ctrlr is reconnected before deleting ctrlr.
+                   -1 means infinite reconnect retries. 0 means no reconnect retry.
+                   If reconnect_delay_sec is zero, ctrlr_loss_timeout_sec has to be zero.
+                   If reconnect_delay_sec is non-zero, ctrlr_loss_timeout_sec has to be -1 or not less than
+                   reconnect_delay_sec.""",
+                   type=int)
+    p.add_argument('-o', '--reconnect-delay-sec',
+                   help="""Time to delay a reconnect retry.
+                   If ctrlr_loss_timeout_sec is zero, reconnect_delay_sec has to be zero.
+                   If ctrlr_loss_timeout_sec is -1, reconnect_delay_sec has to be non-zero.
+                   If ctrlr_loss_timeout_sec is not -1 or zero, reconnect_delay_sec has to be non-zero and
+                   less than ctrlr_loss_timeout_sec.""",
+                   type=int)
+    p.add_argument('-u', '--fast-io-fail-timeout-sec',
+                   help="""Time to wait until ctrlr is reconnected before failing I/O to ctrlr.
+                   0 means no such timeout.
+                   If fast_io_fail_timeout_sec is not zero, it has to be not less than reconnect_delay_sec and
+                   less than ctrlr_loss_timeout_sec if ctrlr_loss_timeout_sec is not -1.""",
+                   type=int)
     p.set_defaults(func=bdev_nvme_start_discovery)
 
     def bdev_nvme_stop_discovery(args):

scripts/rpc/bdev.py

@@ -724,7 +724,8 @@ def bdev_nvme_reset_controller(client, name):
 
 
 def bdev_nvme_start_discovery(client, name, trtype, traddr, adrfam=None, trsvcid=None,
-                              hostnqn=None):
+                              hostnqn=None, ctrlr_loss_timeout_sec=None, reconnect_delay_sec=None,
+                              fast_io_fail_timeout_sec=None):
     """Start discovery with the specified discovery subsystem
 
     Args:
@@ -734,6 +735,20 @@ def bdev_nvme_start_discovery(client, name, trtype, traddr, adrfam=None, trsvcid
         adrfam: address family ("IPv4", "IPv6", "IB", or "FC")
         trsvcid: transport service ID (port number for IP-based addresses)
         hostnqn: NQN to connect from (optional)
+        ctrlr_loss_timeout_sec: Time to wait until ctrlr is reconnected before deleting ctrlr.
+        -1 means infinite reconnect retries. 0 means no reconnect retry.
+        If reconnect_delay_sec is zero, ctrlr_loss_timeout_sec has to be zero.
+        If reconnect_delay_sec is non-zero, ctrlr_loss_timeout_sec has to be -1 or not less than reconnect_delay_sec.
+        (optional)
+        reconnect_delay_sec: Time to delay a reconnect retry.
+        If ctrlr_loss_timeout_sec is zero, reconnect_delay_sec has to be zero.
+        If ctrlr_loss_timeout_sec is -1, reconnect_delay_sec has to be non-zero.
+        If ctrlr_loss_timeout_sec is not -1 or zero, reconnect_sec has to be non-zero and less than ctrlr_loss_timeout_sec.
+        (optional)
+        fail_io_fast_timeout_sec: Time to wait until ctrlr is reconnected before failing I/O to ctrlr.
+        0 means no such timeout.
+        If fast_io_fail_timeout_sec is not zero, it has to be not less than reconnect_delay_sec and less than
+        ctrlr_loss_timeout_sec if ctrlr_loss_timeout_sec is not -1. (optional)
     """
     params = {'name': name,
               'trtype': trtype,
@@ -748,6 +763,15 @@ def bdev_nvme_start_discovery(client, name, trtype, traddr, adrfam=None, trsvcid
     if trsvcid:
         params['trsvcid'] = trsvcid
 
+    if ctrlr_loss_timeout_sec is not None:
+        params['ctrlr_loss_timeout_sec'] = ctrlr_loss_timeout_sec
+
+    if reconnect_delay_sec is not None:
+        params['reconnect_delay_sec'] = reconnect_delay_sec
+
+    if fast_io_fail_timeout_sec is not None:
+        params['fast_io_fail_timeout_sec'] = fast_io_fail_timeout_sec
+
     return client.call('bdev_nvme_start_discovery', params)