diff --git a/cmd/zpool/zpool_main.c b/cmd/zpool/zpool_main.c
index 5af626558a..4845956e51 100644
--- a/cmd/zpool/zpool_main.c
+++ b/cmd/zpool/zpool_main.c
@@ -97,6 +97,7 @@ static int zpool_do_replace(int, char **);
 static int zpool_do_split(int, char **);
 
 static int zpool_do_scrub(int, char **);
+static int zpool_do_resilver(int, char **);
 
 static int zpool_do_import(int, char **);
 static int zpool_do_export(int, char **);
@@ -149,6 +150,7 @@ typedef enum {
 	HELP_REPLACE,
 	HELP_REMOVE,
 	HELP_SCRUB,
+	HELP_RESILVER,
 	HELP_STATUS,
 	HELP_UPGRADE,
 	HELP_EVENTS,
@@ -276,6 +278,7 @@ static zpool_command_t command_table[] = {
 	{ "split",	zpool_do_split,		HELP_SPLIT		},
 	{ NULL },
 	{ "scrub",	zpool_do_scrub,		HELP_SCRUB		},
+	{ "resilver",	zpool_do_resilver,	HELP_RESILVER		},
 	{ NULL },
 	{ "import",	zpool_do_import,	HELP_IMPORT		},
 	{ "export",	zpool_do_export,	HELP_EXPORT		},
@@ -358,6 +361,8 @@ get_usage(zpool_help_t idx)
 		return (gettext("\treopen [-n] <pool>\n"));
 	case HELP_SCRUB:
 		return (gettext("\tscrub [-s | -p] <pool> ...\n"));
+	case HELP_RESILVER:
+		return (gettext("\tresilver <pool> ...\n"));
 	case HELP_STATUS:
 		return (gettext("\tstatus [-c [script1,script2,...]] [-gLPvxD]"
 		    "[-T d|u] [pool] ... \n"
@@ -1874,11 +1879,14 @@ print_status_config(zpool_handle_t *zhp, status_cbdata_t *cb, const char *name,
 	(void) nvlist_lookup_uint64_array(root, ZPOOL_CONFIG_SCAN_STATS,
 	    (uint64_t **)&ps, &c);
 
-	if (ps != NULL && ps->pss_state == DSS_SCANNING &&
-	    vs->vs_scan_processed != 0 && children == 0) {
-		(void) printf(gettext("  (%s)"),
-		    (ps->pss_func == POOL_SCAN_RESILVER) ?
-		    "resilvering" : "repairing");
+	if (ps != NULL && ps->pss_state == DSS_SCANNING && children == 0) {
+		if (vs->vs_scan_processed != 0) {
+			(void) printf(gettext("  (%s)"),
+			    (ps->pss_func == POOL_SCAN_RESILVER) ?
+			    "resilvering" : "repairing");
+		} else if (vs->vs_resilver_deferred) {
+			(void) printf(gettext("  (awaiting resilver)"));
+		}
 	}
 
 	if (cb->vcdl != NULL) {
@@ -6251,7 +6259,7 @@ scrub_callback(zpool_handle_t *zhp, void *data)
 	 * Ignore faulted pools.
 	 */
 	if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) {
-		(void) fprintf(stderr, gettext("cannot scrub '%s': pool is "
+		(void) fprintf(stderr, gettext("cannot scan '%s': pool is "
 		    "currently unavailable\n"), zpool_get_name(zhp));
 		return (1);
 	}
@@ -6319,6 +6327,44 @@ zpool_do_scrub(int argc, char **argv)
 	return (for_each_pool(argc, argv, B_TRUE, NULL, scrub_callback, &cb));
 }
 
+/*
+ * zpool resilver <pool> ...
+ *
+ *	Restarts any in-progress resilver
+ */
+int
+zpool_do_resilver(int argc, char **argv)
+{
+	int c;
+	scrub_cbdata_t cb;
+
+	cb.cb_type = POOL_SCAN_RESILVER;
+	cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;
+	cb.cb_argc = argc;
+	cb.cb_argv = argv;
+
+	/* check options */
+	while ((c = getopt(argc, argv, "")) != -1) {
+		switch (c) {
+		case '?':
+			(void) fprintf(stderr, gettext("invalid option '%c'\n"),
+			    optopt);
+			usage(B_FALSE);
+		}
+	}
+
+	argc -= optind;
+	argv += optind;
+
+	if (argc < 1) {
+		(void) fprintf(stderr, gettext("missing pool name argument\n"));
+		usage(B_FALSE);
+	}
+
+	return (for_each_pool(argc, argv, B_TRUE, NULL, scrub_callback, &cb));
+}
+
+
 /*
  * Print out detailed scrub status.
  */
diff --git a/configure.ac b/configure.ac
index 301258e7f7..59c8c5f7f3 100644
--- a/configure.ac
+++ b/configure.ac
@@ -252,6 +252,7 @@ AC_CONFIG_FILES([
 	tests/zfs-tests/tests/functional/cli_root/zpool_online/Makefile
 	tests/zfs-tests/tests/functional/cli_root/zpool_remove/Makefile
 	tests/zfs-tests/tests/functional/cli_root/zpool_reopen/Makefile
+	tests/zfs-tests/tests/functional/cli_root/zpool_resilver/Makefile
 	tests/zfs-tests/tests/functional/cli_root/zpool_replace/Makefile
 	tests/zfs-tests/tests/functional/cli_root/zpool_scrub/Makefile
 	tests/zfs-tests/tests/functional/cli_root/zpool_set/Makefile
diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h
index 6bbf843461..4f0e13dfdc 100644
--- a/include/sys/fs/zfs.h
+++ b/include/sys/fs/zfs.h
@@ -710,6 +710,7 @@ typedef struct zpool_load_policy {
 #define	ZPOOL_CONFIG_VDEV_TOP_ZAP	"com.delphix:vdev_zap_top"
 #define	ZPOOL_CONFIG_VDEV_LEAF_ZAP	"com.delphix:vdev_zap_leaf"
 #define	ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS	"com.delphix:has_per_vdev_zaps"
+#define	ZPOOL_CONFIG_RESILVER_DEFER	"com.datto:resilver_defer"
 #define	ZPOOL_CONFIG_CACHEFILE		"cachefile"	/* not stored on disk */
 #define	ZPOOL_CONFIG_MMP_STATE		"mmp_state"	/* not stored on disk */
 #define	ZPOOL_CONFIG_MMP_TXG		"mmp_txg"	/* not stored on disk */
@@ -988,6 +989,7 @@ typedef struct vdev_stat {
 	uint64_t	vs_scan_processed;	/* scan processed bytes	*/
 	uint64_t	vs_fragmentation;	/* device fragmentation */
 	uint64_t	vs_checkpoint_space;    /* checkpoint-consumed space */
+	uint64_t	vs_resilver_deferred;	/* resilver deferred	*/
 } vdev_stat_t;
 
 /*
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index 9dbdcfcf52..404aaa9ee3 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -281,6 +281,13 @@ struct spa {
 	uint64_t	spa_scan_pass_scrub_spent_paused; /* total paused */
 	uint64_t	spa_scan_pass_exam;	/* examined bytes per pass */
 	uint64_t	spa_scan_pass_issued;	/* issued bytes per pass */
+
+	/*
+	 * We are in the middle of a resilver, and another resilver
+	 * is needed once this one completes. This is set iff any
+	 * vdev_resilver_deferred is set.
+	 */
+	boolean_t	spa_resilver_deferred;
 	kmutex_t	spa_async_lock;		/* protect async state */
 	kthread_t	*spa_async_thread;	/* thread doing async task */
 	int		spa_async_suspended;	/* async tasks suspended */
diff --git a/include/sys/vdev.h b/include/sys/vdev.h
index b37b60bdd1..2091892b27 100644
--- a/include/sys/vdev.h
+++ b/include/sys/vdev.h
@@ -149,6 +149,8 @@ extern int vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg);
 extern void vdev_state_dirty(vdev_t *vd);
 extern void vdev_state_clean(vdev_t *vd);
 
+extern void vdev_set_deferred_resilver(spa_t *spa, vdev_t *vd);
+
 typedef enum vdev_config_flag {
 	VDEV_CONFIG_SPARE = 1 << 0,
 	VDEV_CONFIG_L2CACHE = 1 << 1,
diff --git a/include/sys/vdev_impl.h b/include/sys/vdev_impl.h
index d308685478..7b07fe6c1a 100644
--- a/include/sys/vdev_impl.h
+++ b/include/sys/vdev_impl.h
@@ -335,6 +335,7 @@ struct vdev {
 	boolean_t	vdev_isspare;	/* was a hot spare		*/
 	boolean_t	vdev_isl2cache;	/* was a l2cache device		*/
 	boolean_t	vdev_copy_uberblocks;  /* post expand copy uberblocks */
+	boolean_t	vdev_resilver_deferred;  /* resilver deferred */
 	vdev_queue_t	vdev_queue;	/* I/O deadline schedule queue	*/
 	vdev_cache_t	vdev_cache;	/* physical block cache		*/
 	spa_aux_vdev_t	*vdev_aux;	/* for l2cache and spares vdevs	*/
diff --git a/include/zfeature_common.h b/include/zfeature_common.h
index 3804d7b1ab..089a7e29d1 100644
--- a/include/zfeature_common.h
+++ b/include/zfeature_common.h
@@ -65,6 +65,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_POOL_CHECKPOINT,
 	SPA_FEATURE_SPACEMAP_V2,
 	SPA_FEATURE_ALLOCATION_CLASSES,
+	SPA_FEATURE_RESILVER_DEFER,
 	SPA_FEATURES
 } spa_feature_t;
 
diff --git a/include/zfs_gitrev.h b/include/zfs_gitrev.h
new file mode 100644
index 0000000000..13a831cb38
--- /dev/null
+++ b/include/zfs_gitrev.h
@@ -0,0 +1 @@
+#define	ZFS_META_GITREV "unknown"
diff --git a/man/man5/zpool-features.5 b/man/man5/zpool-features.5
index 5cc4db45e0..c1f6ee6e4e 100644
--- a/man/man5/zpool-features.5
+++ b/man/man5/zpool-features.5
@@ -756,6 +756,27 @@ can also be triggered on filesystems via `zfs set version=current <pool/fs>`.
 The upgrade process runs in the background and may take a while to complete
 for the filesystems containing a large number of files.
 
+.RE
+.sp
+.ne 2
+.na
+\fB\fBresilver_defer\fR\fR
+.ad
+.RS 4n
+.TS
+l l .
+GUID	com.datto:resilver_defer
+READ\-ONLY COMPATIBLE	yes
+DEPENDENCIES	none
+.TE
+
+This feature allows zfs to postpone new resilvers if an existing one is already
+in progress. Without this feature, any new resilvers will cause the currently
+running one to be immediately restarted from the beginning.
+
+This feature becomes \fBactive\fR once a resilver has been defered, and returns
+to being \fBenabled\fR when the defered resilver begins.
+
 .RE
 
 .sp
diff --git a/man/man8/zpool.8 b/man/man8/zpool.8
index 821b1e4d6b..8fe6b494be 100644
--- a/man/man8/zpool.8
+++ b/man/man8/zpool.8
@@ -162,6 +162,9 @@
 .Oo Fl o Ar property Ns = Ns Ar value Oc
 .Ar pool Ar device Op Ar new_device
 .Nm
+.Cm resilver
+.Ar pool Ns ...
+.Nm
 .Cm scrub
 .Op Fl s | Fl p
 .Ar pool Ns ...
@@ -2069,6 +2072,14 @@ again.
 .El
 .It Xo
 .Nm
+.Cm resilver
+.Ar pool Ns ...
+.Xc
+Starts a resilver. If an existing resilver is already running it will be
+restarted from the beginning. Any drives that were scheduled for a deferred
+resilver will be added to the new one.
+.It Xo
+.Nm
 .Cm set
 .Ar property Ns = Ns Ar value
 .Ar pool
diff --git a/module/zcommon/zfeature_common.c b/module/zcommon/zfeature_common.c
index a0ad108c50..40ce01b947 100644
--- a/module/zcommon/zfeature_common.c
+++ b/module/zcommon/zfeature_common.c
@@ -445,6 +445,11 @@ zpool_feature_init(void)
 	    "Support for separate allocation classes.",
 	    ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);
 	}
+
+	zfeature_register(SPA_FEATURE_RESILVER_DEFER,
+	    "com.datto:resilver_defer", "resilver_defer",
+	    "Support for defering new resilvers when one is already running.",
+	    ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);
 }
 
 #if defined(_KERNEL)
diff --git a/module/zfs/dsl_scan.c b/module/zfs/dsl_scan.c
index b84c2aa45f..aff99f275e 100644
--- a/module/zfs/dsl_scan.c
+++ b/module/zfs/dsl_scan.c
@@ -175,6 +175,8 @@ enum ddt_class zfs_scrub_ddt_class_max = DDT_CLASS_DUPLICATE;
 /* max number of blocks to free in a single TXG */
 unsigned long zfs_async_block_max_blocks = 100000;
 
+int zfs_resilver_disable_defer = 0; /* set to disable resilver deferring */
+
 /*
  * We wait a few txgs after importing a pool to begin scanning so that
  * the import / mounting code isn't held up by scrub / resilver IO.
@@ -720,6 +722,11 @@ dsl_scan(dsl_pool_t *dp, pool_scan_func_t func)
 	spa->spa_scrub_reopen = B_FALSE;
 	(void) spa_vdev_state_exit(spa, NULL, 0);
 
+	if (func == POOL_SCAN_RESILVER) {
+		dsl_resilver_restart(spa->spa_dsl_pool, 0);
+		return (0);
+	}
+
 	if (func == POOL_SCAN_SCRUB && dsl_scan_is_paused_scrub(scn)) {
 		/* got scrub start cmd, resume paused scrub */
 		int err = dsl_scrub_set_pause_resume(scn->scn_dp,
@@ -736,6 +743,41 @@ dsl_scan(dsl_pool_t *dp, pool_scan_func_t func)
 	    dsl_scan_setup_sync, &func, 0, ZFS_SPACE_CHECK_EXTRA_RESERVED));
 }
 
+/*
+ * Sets the resilver defer flag to B_FALSE on all leaf devs under vd. Returns
+ * B_TRUE if we have devices that need to be resilvered and are available to
+ * accept resilver I/Os.
+ */
+static boolean_t
+dsl_scan_clear_deferred(vdev_t *vd, dmu_tx_t *tx)
+{
+	boolean_t resilver_needed = B_FALSE;
+	spa_t *spa = vd->vdev_spa;
+
+	for (int c = 0; c < vd->vdev_children; c++) {
+		resilver_needed |=
+		    dsl_scan_clear_deferred(vd->vdev_child[c], tx);
+	}
+
+	if (vd == spa->spa_root_vdev &&
+	    spa_feature_is_active(spa, SPA_FEATURE_RESILVER_DEFER)) {
+		spa_feature_decr(spa, SPA_FEATURE_RESILVER_DEFER, tx);
+		vdev_config_dirty(vd);
+		spa->spa_resilver_deferred = B_FALSE;
+		return (resilver_needed);
+	}
+
+	if (!vdev_is_concrete(vd) || vd->vdev_aux ||
+	    !vd->vdev_ops->vdev_op_leaf)
+		return (resilver_needed);
+
+	if (vd->vdev_resilver_deferred)
+		vd->vdev_resilver_deferred = B_FALSE;
+
+	return (!vdev_is_dead(vd) && !vd->vdev_offline &&
+	    vdev_resilver_needed(vd, NULL, NULL));
+}
+
 /* ARGSUSED */
 static void
 dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx)
@@ -835,6 +877,25 @@ dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx)
 		 * Let the async thread assess this and handle the detach.
 		 */
 		spa_async_request(spa, SPA_ASYNC_RESILVER_DONE);
+
+		/*
+		 * Clear any deferred_resilver flags in the config.
+		 * If there are drives that need resilvering, kick
+		 * off an asynchronous request to start resilver.
+		 * dsl_scan_clear_deferred() may update the config
+		 * before the resilver can restart. In the event of
+		 * a crash during this period, the spa loading code
+		 * will find the drives that need to be resilvered
+		 * when the machine reboots and start the resilver then.
+		 */
+		boolean_t resilver_needed =
+		    dsl_scan_clear_deferred(spa->spa_root_vdev, tx);
+		if (resilver_needed) {
+			spa_history_log_internal(spa,
+			    "starting deferred resilver", tx,
+			    "errors=%llu", spa_get_errlog_size(spa));
+			spa_async_request(spa, SPA_ASYNC_RESILVER);
+		}
 	}
 
 	scn->scn_phys.scn_end_time = gethrestime_sec();
@@ -2966,6 +3027,26 @@ dsl_scan_active(dsl_scan_t *scn)
 	return (used != 0);
 }
 
+static boolean_t
+dsl_scan_check_deferred(vdev_t *vd)
+{
+	boolean_t need_resilver = B_FALSE;
+
+	for (int c = 0; c < vd->vdev_children; c++) {
+		need_resilver |=
+		    dsl_scan_check_deferred(vd->vdev_child[c]);
+	}
+
+	if (!vdev_is_concrete(vd) || vd->vdev_aux ||
+	    !vd->vdev_ops->vdev_op_leaf)
+		return (need_resilver);
+
+	if (!vd->vdev_resilver_deferred)
+		need_resilver = B_TRUE;
+
+	return (need_resilver);
+}
+
 static boolean_t
 dsl_scan_need_resilver(spa_t *spa, const dva_t *dva, size_t psize,
     uint64_t phys_birth)
@@ -3013,6 +3094,13 @@ dsl_scan_need_resilver(spa_t *spa, const dva_t *dva, size_t psize,
 	if (!vdev_dtl_need_resilver(vd, DVA_GET_OFFSET(dva), psize))
 		return (B_FALSE);
 
+	/*
+	 * Check that this top-level vdev has a device under it which
+	 * is resilvering and is not deferred.
+	 */
+	if (!dsl_scan_check_deferred(vd))
+		return (B_FALSE);
+
 	return (B_TRUE);
 }
 
@@ -3173,12 +3261,19 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
 	spa_t *spa = dp->dp_spa;
 	state_sync_type_t sync_type = SYNC_OPTIONAL;
 
+	if (spa->spa_resilver_deferred &&
+	    !spa_feature_is_active(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER))
+		spa_feature_incr(spa, SPA_FEATURE_RESILVER_DEFER, tx);
+
 	/*
 	 * Check for scn_restart_txg before checking spa_load_state, so
 	 * that we can restart an old-style scan while the pool is being
-	 * imported (see dsl_scan_init).
+	 * imported (see dsl_scan_init). We also restart scans if there
+	 * is a deferred resilver and the user has manually disabled
+	 * deferred resilvers via the tunable.
 	 */
-	if (dsl_scan_restarting(scn, tx)) {
+	if (dsl_scan_restarting(scn, tx) ||
+	    (spa->spa_resilver_deferred && zfs_resilver_disable_defer)) {
 		pool_scan_func_t func = POOL_SCAN_SCRUB;
 		dsl_scan_done(scn, B_FALSE, tx);
 		if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL))
@@ -4000,4 +4095,8 @@ MODULE_PARM_DESC(zfs_scan_strict_mem_lim,
 module_param(zfs_scan_fill_weight, int, 0644);
 MODULE_PARM_DESC(zfs_scan_fill_weight,
 	"Tunable to adjust bias towards more filled segments during scans");
+
+module_param(zfs_resilver_disable_defer, int, 0644);
+MODULE_PARM_DESC(zfs_resilver_disable_defer,
+	"Process all resilvers immediately");
 #endif
diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index fdce49c40c..3785981b76 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -6059,9 +6059,14 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 	/*
 	 * Schedule the resilver to restart in the future. We do this to
 	 * ensure that dmu_sync-ed blocks have been stitched into the
-	 * respective datasets.
+	 * respective datasets. We do not do this if resilvers have been
+	 * deferred.
 	 */
-	dsl_resilver_restart(spa->spa_dsl_pool, dtl_max_txg);
+	if (dsl_scan_resilvering(spa_get_dsl(spa)) &&
+	    spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER))
+		vdev_set_deferred_resilver(spa, newvd);
+	else
+		dsl_resilver_restart(spa->spa_dsl_pool, dtl_max_txg);
 
 	if (spa->spa_bootfs)
 		spa_event_notify(spa, newvd, NULL, ESC_ZFS_BOOTFS_VDEV_ATTACH);
@@ -6933,6 +6938,7 @@ static void
 spa_async_thread(void *arg)
 {
 	spa_t *spa = (spa_t *)arg;
+	dsl_pool_t *dp = spa->spa_dsl_pool;
 	int tasks;
 
 	ASSERT(spa->spa_sync_on);
@@ -7008,8 +7014,10 @@ spa_async_thread(void *arg)
 	/*
 	 * Kick off a resilver.
 	 */
-	if (tasks & SPA_ASYNC_RESILVER)
-		dsl_resilver_restart(spa->spa_dsl_pool, 0);
+	if (tasks & SPA_ASYNC_RESILVER &&
+	    (!dsl_scan_resilvering(dp) ||
+	    !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER)))
+		dsl_resilver_restart(dp, 0);
 
 	/*
 	 * Let the world know that we're done.
diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c
index 2c95626c4d..1521acc405 100644
--- a/module/zfs/vdev.c
+++ b/module/zfs/vdev.c
@@ -790,6 +790,9 @@ vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id,
 		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG,
 		    &vd->vdev_resilver_txg);
 
+		if (nvlist_exists(nv, ZPOOL_CONFIG_RESILVER_DEFER))
+			vdev_set_deferred_resilver(spa, vd);
+
 		/*
 		 * In general, when importing a pool we want to ignore the
 		 * persistent fault state, as the diagnosis made on another
@@ -1798,8 +1801,13 @@ vdev_open(vdev_t *vd)
 	 * since this would just restart the scrub we are already doing.
 	 */
 	if (vd->vdev_ops->vdev_op_leaf && !spa->spa_scrub_reopen &&
-	    vdev_resilver_needed(vd, NULL, NULL))
-		spa_async_request(spa, SPA_ASYNC_RESILVER);
+	    vdev_resilver_needed(vd, NULL, NULL)) {
+		if (dsl_scan_resilvering(spa->spa_dsl_pool) &&
+		    spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER))
+			vdev_set_deferred_resilver(spa, vd);
+		else
+			spa_async_request(spa, SPA_ASYNC_RESILVER);
+	}
 
 	return (0);
 }
@@ -2488,6 +2496,9 @@ vdev_dtl_should_excise(vdev_t *vd)
 	if (vd->vdev_state < VDEV_STATE_DEGRADED)
 		return (B_FALSE);
 
+	if (vd->vdev_resilver_deferred)
+		return (B_FALSE);
+
 	if (vd->vdev_resilver_txg == 0 ||
 	    range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]))
 		return (B_TRUE);
@@ -3618,8 +3629,14 @@ vdev_clear(spa_t *spa, vdev_t *vd)
 		if (vd != rvd && vdev_writeable(vd->vdev_top))
 			vdev_state_dirty(vd->vdev_top);
 
-		if (vd->vdev_aux == NULL && !vdev_is_dead(vd))
-			spa_async_request(spa, SPA_ASYNC_RESILVER);
+		if (vd->vdev_aux == NULL && !vdev_is_dead(vd)) {
+			if (dsl_scan_resilvering(spa->spa_dsl_pool) &&
+			    spa_feature_is_enabled(spa,
+			    SPA_FEATURE_RESILVER_DEFER))
+				vdev_set_deferred_resilver(spa, vd);
+			else
+				spa_async_request(spa, SPA_ASYNC_RESILVER);
+		}
 
 		spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_CLEAR);
 	}
@@ -3840,6 +3857,8 @@ vdev_get_stats_ex(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx)
 			vs->vs_fragmentation = (vd->vdev_mg != NULL) ?
 			    vd->vdev_mg->mg_fragmentation : 0;
 		}
+		if (vd->vdev_ops->vdev_op_leaf)
+			vs->vs_resilver_deferred = vd->vdev_resilver_deferred;
 	}
 
 	ASSERT(spa_config_held(vd->vdev_spa, SCL_ALL, RW_READER) != 0);
@@ -4578,6 +4597,14 @@ vdev_deadman(vdev_t *vd, char *tag)
 	}
 }
 
+void
+vdev_set_deferred_resilver(spa_t *spa, vdev_t *vd)
+{
+	ASSERT(vd->vdev_ops->vdev_op_leaf);
+	vd->vdev_resilver_deferred = B_TRUE;
+	spa->spa_resilver_deferred = B_TRUE;
+}
+
 #if defined(_KERNEL)
 EXPORT_SYMBOL(vdev_fault);
 EXPORT_SYMBOL(vdev_degrade);
diff --git a/module/zfs/vdev_label.c b/module/zfs/vdev_label.c
index 439ab7438d..7e86e3a8b3 100644
--- a/module/zfs/vdev_label.c
+++ b/module/zfs/vdev_label.c
@@ -524,6 +524,12 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 			fnvlist_add_uint64(nv, ZPOOL_CONFIG_VDEV_TOP_ZAP,
 			    vd->vdev_top_zap);
 		}
+
+		if (vd->vdev_resilver_deferred) {
+			ASSERT(vd->vdev_ops->vdev_op_leaf);
+			ASSERT(spa->spa_resilver_deferred);
+			fnvlist_add_boolean(nv, ZPOOL_CONFIG_RESILVER_DEFER);
+		}
 	}
 
 	if (getstats) {
diff --git a/tests/runfiles/linux.run b/tests/runfiles/linux.run
index 613660df95..4f556acde3 100644
--- a/tests/runfiles/linux.run
+++ b/tests/runfiles/linux.run
@@ -421,6 +421,10 @@ tags = ['functional', 'cli_root', 'zpool_reopen']
 tests = ['zpool_replace_001_neg', 'replace-o_ashift', 'replace_prop_ashift']
 tags = ['functional', 'cli_root', 'zpool_replace']
 
+[tests/functional/cli_root/zpool_resilver]
+tests = ['zpool_resilver_bad_args', 'zpool_resilver_restart']
+tags = ['functional', 'cli_root', 'zpool_resilver']
+
 [tests/functional/cli_root/zpool_scrub]
 tests = ['zpool_scrub_001_neg', 'zpool_scrub_002_pos', 'zpool_scrub_003_pos',
     'zpool_scrub_004_pos', 'zpool_scrub_005_pos',
diff --git a/tests/zfs-tests/tests/functional/cli_root/Makefile.am b/tests/zfs-tests/tests/functional/cli_root/Makefile.am
index 7a765a1603..13ff889d81 100644
--- a/tests/zfs-tests/tests/functional/cli_root/Makefile.am
+++ b/tests/zfs-tests/tests/functional/cli_root/Makefile.am
@@ -52,6 +52,7 @@ SUBDIRS = \
 	zpool_remove \
 	zpool_reopen \
 	zpool_replace \
+	zpool_resilver \
 	zpool_scrub \
 	zpool_set \
 	zpool_split \
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg b/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg
index 8907533c9e..48a32174fa 100644
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg
@@ -87,5 +87,6 @@ if is_linux; then
 	    "feature@encryption"
 	    "feature@project_quota"
 	    "feature@allocation_classes"
+	    "feature@resilver_defer"
 	)
 fi
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib
index 82860deb34..075ad85e9f 100755
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib
@@ -115,3 +115,10 @@ function is_scan_restarted #pool
 	zpool history -i $pool | grep -q "scan aborted, restarting"
 	return $?
 }
+
+function is_deferred_scan_started #pool
+{
+	typeset pool=$1
+	zpool history -i $pool | grep -q "starting deferred resilver"
+	return $?
+}
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh
index 30c389ce84..956ceebafb 100755
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh
@@ -29,7 +29,7 @@
 # 4. Execute scrub.
 # 5. "Plug back" disk.
 # 6. Reopen a pool with an -n flag.
-# 7. Check if scrub scan is NOT replaced by resilver.
+# 7. Check if resilver was deferred.
 # 8. Check if trying to put device to offline fails because of no valid
 #    replicas.
 #
@@ -75,11 +75,12 @@ log_must check_state $TESTPOOL "$REMOVED_DISK_ID" "online"
 log_must zinject -c all
 # 7. Check if scrub scan is NOT replaced by resilver.
 log_must wait_for_scrub_end $TESTPOOL $MAXTIMEOUT
-log_mustnot is_scan_restarted $TESTPOOL
+log_must is_deferred_scan_started $TESTPOOL
 
 # 8. Check if trying to put device to offline fails because of no valid
 #    replicas.
-log_mustnot zpool offline $TESTPOOL $DISK2
+log_must wait_for_resilver_end $TESTPOOL $MAXTIMEOUT
+log_must zpool offline $TESTPOOL $DISK2
 
 # clean up
 log_must zpool destroy $TESTPOOL
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_005_pos.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_005_pos.ksh
index 95029a8b6d..fc298d0106 100755
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_005_pos.ksh
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_005_pos.ksh
@@ -72,13 +72,13 @@ log_must zinject -d $REMOVED_DISK_ID -D25:1 $TESTPOOL
 log_must wait_for_resilver_start $TESTPOOL $MAXTIMEOUT
 
 # 6. Reopen a pool again with -n flag.
-zpool reopen -n $TESTPOOL
+log_must zpool reopen -n $TESTPOOL
 
 # 7. Wait until resilvering is finished and check if it was restarted.
 log_must wait_for_resilver_end $TESTPOOL $MAXTIMEOUT
 # remove delay from disk
 log_must zinject -c all
-log_must is_scan_restarted $TESTPOOL
+log_mustnot is_scan_restarted $TESTPOOL
 
 # clean up
 log_must zpool destroy $TESTPOOL
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/Makefile.am b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/Makefile.am
new file mode 100644
index 0000000000..2cec5335fb
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/Makefile.am
@@ -0,0 +1,9 @@
+pkgdatadir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cli_root/zpool_resilver
+dist_pkgdata_SCRIPTS = \
+	setup.ksh \
+	cleanup.ksh \
+	zpool_resilver_bad_args.ksh \
+	zpool_resilver_restart.ksh
+
+dist_pkgdata_DATA = \
+	zpool_resilver.cfg
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/cleanup.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/cleanup.ksh
new file mode 100755
index 0000000000..c74e23919c
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/cleanup.ksh
@@ -0,0 +1,33 @@
+#!/bin/ksh -p
+#
+# CDDL HEADER START
+#
+# The contents of this file are subject to the terms of the
+# Common Development and Distribution License (the "License").
+# You may not use this file except in compliance with the License.
+#
+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+# or http://www.opensolaris.org/os/licensing.
+# See the License for the specific language governing permissions
+# and limitations under the License.
+#
+# When distributing Covered Code, include this CDDL HEADER in each
+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+# If applicable, add the following below this CDDL HEADER, with the
+# fields enclosed by brackets "[]" replaced with your own identifying
+# information: Portions Copyright [yyyy] [name of copyright owner]
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
+# Use is subject to license terms.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_scrub/zpool_scrub.cfg
+
+verify_runnable "global"
+
+destroy_mirrors
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/setup.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/setup.ksh
new file mode 100755
index 0000000000..48ceecdf9e
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/setup.ksh
@@ -0,0 +1,39 @@
+#!/bin/ksh -p
+#
+# CDDL HEADER START
+#
+# The contents of this file are subject to the terms of the
+# Common Development and Distribution License (the "License").
+# You may not use this file except in compliance with the License.
+#
+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+# or http://www.opensolaris.org/os/licensing.
+# See the License for the specific language governing permissions
+# and limitations under the License.
+#
+# When distributing Covered Code, include this CDDL HEADER in each
+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+# If applicable, add the following below this CDDL HEADER, with the
+# fields enclosed by brackets "[]" replaced with your own identifying
+# information: Portions Copyright [yyyy] [name of copyright owner]
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright (c) 2018 by Datto. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_resilver/zpool_resilver.cfg
+
+verify_runnable "global"
+verify_disk_count "$DISKS" 3
+
+default_mirror_setup_noexit $DISK1 $DISK2 $DISK3
+
+mntpnt=$(get_prop mountpoint $TESTPOOL/$TESTFS)
+
+# Create 256M of data
+log_must file_write -b 1048576 -c 256 -o create -d 0 -f $mntpnt/bigfile
+log_pass
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver.cfg b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver.cfg
new file mode 100644
index 0000000000..7d92984d6a
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver.cfg
@@ -0,0 +1,33 @@
+#
+# CDDL HEADER START
+#
+# The contents of this file are subject to the terms of the
+# Common Development and Distribution License (the "License").
+# You may not use this file except in compliance with the License.
+#
+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+# or http://www.opensolaris.org/os/licensing.
+# See the License for the specific language governing permissions
+# and limitations under the License.
+#
+# When distributing Covered Code, include this CDDL HEADER in each
+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+# If applicable, add the following below this CDDL HEADER, with the
+# fields enclosed by brackets "[]" replaced with your own identifying
+# information: Portions Copyright [yyyy] [name of copyright owner]
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright (c) 2018 by Datto. All rights reserved.
+#
+
+export DISK1=$(echo $DISKS | nawk '{print $1}')
+export DISK2=$(echo $DISKS | nawk '{print $2}')
+export DISK3=$(echo $DISKS | nawk '{print $3}')
+
+export ZFS_SCAN_VDEV_LIMIT_SLOW=$((128*1024))
+export ZFS_SCAN_VDEV_LIMIT_DEFAULT=$((4*1024*1024))
+
+export MAXTIMEOUT=80
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_bad_args.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_bad_args.ksh
new file mode 100755
index 0000000000..9d973bec78
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_bad_args.ksh
@@ -0,0 +1,58 @@
+#!/bin/ksh -p
+#
+# CDDL HEADER START
+#
+# The contents of this file are subject to the terms of the
+# Common Development and Distribution License (the "License").
+# You may not use this file except in compliance with the License.
+#
+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+# or http://www.opensolaris.org/os/licensing.
+# See the License for the specific language governing permissions
+# and limitations under the License.
+#
+# When distributing Covered Code, include this CDDL HEADER in each
+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+# If applicable, add the following below this CDDL HEADER, with the
+# fields enclosed by brackets "[]" replaced with your own identifying
+# information: Portions Copyright [yyyy] [name of copyright owner]
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright (c) 2018 by Datto. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+
+#
+# DESCRIPTION:
+# A badly formed parameter passed to 'zpool resilver' should
+# return an error.
+#
+# STRATEGY:
+# 1. Create an array containing bad 'zpool reilver' parameters.
+# 2. For each element, execute the sub-command.
+# 3. Verify it returns an error.
+#
+
+verify_runnable "global"
+
+set -A args "" "-?" "blah blah" "-%" "--?" "-*" "-=" \
+    "-a" "-b" "-c" "-d" "-e" "-f" "-g" "-h" "-i" "-j" "-k" "-l" \
+    "-m" "-n" "-o" "-p" "-q" "-r" "-s" "-t" "-u" "-v" "-w" "-x" "-y" "-z" \
+    "-A" "-B" "-C" "-D" "-E" "-F" "-G" "-H" "-I" "-J" "-K" "-L" \
+    "-M" "-N" "-O" "-P" "-Q" "-R" "-S" "-T" "-U" "-V" "-W" "-X" "-W" "-Z"
+
+
+log_assert "Execute 'zpool resilver' using invalid parameters."
+
+typeset -i i=0
+while [[ $i -lt ${#args[*]} ]]; do
+	log_mustnot zpool resilver ${args[i]}
+
+	((i = i + 1))
+done
+
+log_pass "Badly formed 'zpool resilver' parameters fail as expected."
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_restart.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_restart.ksh
new file mode 100755
index 0000000000..de9e5ecdf4
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_resilver/zpool_resilver_restart.ksh
@@ -0,0 +1,95 @@
+#!/bin/ksh -p
+#
+# CDDL HEADER START
+#
+# The contents of this file are subject to the terms of the
+# Common Development and Distribution License (the "License").
+# You may not use this file except in compliance with the License.
+#
+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+# or http://www.opensolaris.org/os/licensing.
+# See the License for the specific language governing permissions
+# and limitations under the License.
+#
+# When distributing Covered Code, include this CDDL HEADER in each
+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+# If applicable, add the following below this CDDL HEADER, with the
+# fields enclosed by brackets "[]" replaced with your own identifying
+# information: Portions Copyright [yyyy] [name of copyright owner]
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright (c) 2018 Datto Inc.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_resilver/zpool_resilver.cfg
+
+#
+# DESCRIPTION:
+#	"Verify 'zpool resilver' restarts in-progress resilvers"
+#
+# STRATEGY:
+#	1. Write some data and detatch the first drive so it has resilver
+#	   work to do
+#	2. Repeat the process with a second disk
+#	3. Reattach the drives, causing the second drive's resilver to be
+#	   deferred
+#	4. Manually restart the resilver with all drives
+#
+# NOTES:
+#	Artificially limit the scrub speed by setting the zfs_scan_vdev_limit
+#	low and adding a 50ms zio delay in order to ensure that the resilver
+#	does not complete early.
+#
+
+verify_runnable "global"
+
+function cleanup
+{
+	log_must zinject -c all
+	log_must set_tunable64 zfs_scan_vdev_limit $ZFS_SCAN_VDEV_LIMIT_DEFAULT
+	log_must rm -f $mntpnt/biggerfile1
+	log_must rm -f $mntpnt/biggerfile2
+}
+
+log_onexit cleanup
+
+log_assert "Verify 'zpool resilver' restarts in-progress resilvers"
+
+mntpnt=$(get_prop mountpoint $TESTPOOL/$TESTFS)
+
+# 1. Write some data and detatch the first drive so it has resilver work to do
+log_must file_write -b 524288 -c 1024 -o create -d 0 -f $mntpnt/biggerfile1
+log_must sync
+log_must zpool detach $TESTPOOL $DISK2
+
+# 2. Repeat the process with a second disk
+log_must file_write -b 524288 -c 1024 -o create -d 0 -f $mntpnt/biggerfile2
+log_must sync
+log_must zpool detach $TESTPOOL $DISK3
+
+# 3. Reattach the drives, causing the second drive's resilver to be deferred
+log_must set_tunable64 zfs_scan_vdev_limit $ZFS_SCAN_VDEV_LIMIT_SLOW
+
+log_must zpool attach $TESTPOOL $DISK1 $DISK2
+log_must zinject -d $DISK2 -D50:1 $TESTPOOL
+log_must is_pool_resilvering $TESTPOOL true
+
+log_must zpool attach $TESTPOOL $DISK1 $DISK3
+log_must zinject -d $DISK3 -D50:1 $TESTPOOL
+log_must is_pool_resilvering $TESTPOOL true
+
+# 4. Manually restart the resilver with all drives
+log_must zpool resilver $TESTPOOL
+log_must zinject -c all
+log_must set_tunable64 zfs_scan_vdev_limit $ZFS_SCAN_VDEV_LIMIT_DEFAULT
+log_must wait_for_resilver_end $TESTPOOL $MAXTIMEOUT
+log_must is_deferred_scan_started $TESTPOOL
+log_must check_state $TESTPOOL "$DISK2" "online"
+log_must check_state $TESTPOOL "$DISK3" "online"
+
+log_pass "Verified 'zpool resilver' restarts in-progress resilvers"
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh
index 3bc798d1a9..fdf315dea5 100755
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh
@@ -25,6 +25,7 @@
 #
 
 . $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_reopen/zpool_reopen.shlib
 . $STF_SUITE/tests/functional/cli_root/zpool_scrub/zpool_scrub.cfg
 
 #
@@ -95,6 +96,7 @@ DISK1="$TEST_BASE_DIR/zpool_disk1.dat"
 DISK2="$TEST_BASE_DIR/zpool_disk2.dat"
 DISK3="$TEST_BASE_DIR/zpool_disk3.dat"
 DISK4="$TEST_BASE_DIR/zpool_disk4.dat"
+RESILVER_TIMEOUT=40
 
 # 1. Create the pool
 log_must truncate -s $DEVSIZE $DISK1
@@ -117,6 +119,7 @@ zpool_scrub_sync $TESTPOOL
 # 5. Online the first device and offline the second device
 zpool_do_sync 'online' $TESTPOOL $DISK1
 zpool_do_sync 'offline' $TESTPOOL $DISK2
+log_must wait_for_resilver_end $TESTPOOL $RESILVER_TIMEOUT
 
 # 6. Scrub the pool again
 zpool_scrub_sync $TESTPOOL