diff --git a/src/mgmt/srv_target.c b/src/mgmt/srv_target.c
index 4098a3225ad..c8307df90ec 100644
--- a/src/mgmt/srv_target.c
+++ b/src/mgmt/srv_target.c
@@ -1,5 +1,5 @@
 /*
- * (C) Copyright 2016-2022 Intel Corporation.
+ * (C) Copyright 2016-2024 Intel Corporation.
  *
  * SPDX-License-Identifier: BSD-2-Clause-Patent
  */
@@ -171,12 +171,17 @@ struct tgt_destroy_args {
 	int			 tda_rc;
 };
 
+struct tgt_kill_arg {
+	uuid_t	     pool_uuid;
+	unsigned int flags;
+};
+
 static inline int
-tgt_kill_pool(void *args)
+tgt_kill_pool(void *data)
 {
-	struct d_uuid	*id = args;
+	struct tgt_kill_arg	*arg = data;
 
-	return vos_pool_kill(id->uuid, 0);
+	return vos_pool_kill(arg->pool_uuid, arg->flags);
 }
 
 /**
@@ -267,18 +272,22 @@ struct dead_pool {
 static int
 clear_vos_pool(uuid_t uuid)
 {
-	int			 rc;
-	struct d_uuid		 id;
+	struct tgt_kill_arg kill_arg;
+	int		    rc;
 
 	/* destroy blobIDs */
 	D_DEBUG(DB_MGMT, "Clear SPDK blobs for pool "DF_UUID"\n", DP_UUID(uuid));
-	rc = vos_pool_kill(uuid, VOS_POF_RDB);
+	uuid_copy(kill_arg.pool_uuid, uuid);
+	kill_arg.flags = VOS_POF_RDB;
+	/* This maybe called from main thread, so let's execute the kill on system xtream */
+	rc = dss_ult_execute(tgt_kill_pool, &kill_arg, NULL, NULL, DSS_XS_SYS, 0, 0);
 	if (rc != 0) {
 		D_ERROR(DF_UUID": kill pool service VOS pool: "DF_RC"\n", DP_UUID(uuid), DP_RC(rc));
 		return rc;
 	}
-	uuid_copy(id.uuid, uuid);
-	rc = dss_thread_collective(tgt_kill_pool, &id, 0);
+
+	kill_arg.flags = 0;
+	rc = dss_thread_collective(tgt_kill_pool, &kill_arg, 0);
 	if (rc != 0) {
 		D_ERROR("tgt_kill_pool, rc: "DF_RC"\n", DP_RC(rc));
 		return rc;
@@ -464,7 +473,6 @@ recreate_pooltgts()
 	int			 rc = 0;
 	int			 pool_list_cnt;
 	daos_size_t		 rdb_blob_sz = 0;
-	struct d_uuid		 id;
 
 	D_ASSERT(bio_nvme_configured(SMD_DEV_TYPE_META));
 	D_INIT_LIST_HEAD(&pool_list);
@@ -478,10 +486,13 @@ recreate_pooltgts()
 		/* Cleanup Newborns */
 		if ((pool_info->spi_blob_sz[SMD_DEV_TYPE_META] == 0) ||
 		    (pool_info->spi_flags[SMD_DEV_TYPE_META] & SMD_POOL_IN_CREATION)) {
+			struct tgt_kill_arg arg;
+
 			D_INFO("cleaning up newborn pool "DF_UUID"\n",
 			       DP_UUID(pool_info->spi_id));
-			uuid_copy(id.uuid, pool_info->spi_id);
-			rc = dss_thread_collective(tgt_kill_pool, &id, 0);
+			uuid_copy(arg.pool_uuid, pool_info->spi_id);
+			arg.flags = 0;
+			rc = dss_thread_collective(tgt_kill_pool, &arg, 0);
 			if (rc) {
 				D_ERROR("failed to cleanup newborn pool "DF_UUID": "DF_RC"\n",
 					DP_UUID(pool_info->spi_id), DP_RC(rc));
@@ -1216,15 +1227,18 @@ static int
 tgt_destroy(uuid_t pool_uuid, char *path)
 {
 	struct tgt_destroy_args	 tda = {0};
+	struct tgt_kill_arg	 arg;
 	pthread_t		 thread;
 	int			 rc;
 
 	/* destroy blobIDs first */
-	uuid_copy(tda.tda_id.uuid, pool_uuid);
-	rc = dss_thread_collective(tgt_kill_pool, &tda.tda_id, 0);
+	uuid_copy(arg.pool_uuid, pool_uuid);
+	arg.flags = 0;
+	rc = dss_thread_collective(tgt_kill_pool, &arg, 0);
 	if (rc && rc != -DER_BUSY)
 		goto out;
 
+	uuid_copy(tda.tda_id.uuid, pool_uuid);
 	tda.tda_path = path;
 	tda.tda_rc   = rc;
 	tda.tda_dx   = dss_current_xstream();