Defer new resilvers until the current one ends

Currently, if a resilver is triggered for any reason while an
existing one is running, zfs will immediately restart the existing
resilver from the beginning to include the new drive. This causes
problems for system administrators when a drive fails while another
is already resilvering. In this case, the optimal thing to do to
reduce risk of data loss is to wait for the current resilver to end
before immediately replacing the second failed drive, which allows
the system to operate with two incomplete drives for the minimum
amount of time.

This patch introduces the resilver_defer feature that essentially
does this for the admin without forcing them to wait and monitor
the resilver manually. The change requires an on-disk feature
since we must mark drives that are part of a deferred resilver in
the vdev config to ensure that we do not assume they are done
resilvering when an existing resilver completes.

Reviewed-by: Matthew Ahrens <[email protected]>
Reviewed-by: John Kennedy <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed-by: @mmaybee 
Signed-off-by: Tom Caputi <[email protected]>
Closes #7732 

1 files changed, 52 insertions, 6 deletions

diff --git a/cmd/zpool/zpool_main.c b/cmd/zpool/zpool_main.c
index 5af626558..4845956e5 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);
 	}
@@ -6320,6 +6328,44 @@ zpool_do_scrub(int argc, char **argv)
 }
 
 /*
+ * 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.
  */
 static void