Implement Redacted Send/Receive

Redacted send/receive allows users to send subsets of their data to a target system. One possible use case for this feature is to not transmit sensitive information to a data warehousing, test/dev, or analytics environment. Another is to save space by not replicating unimportant data within a given dataset, for example in backup tools like zrepl. Redacted send/receive is a three-stage process. First, a clone (or clones) is made of the snapshot to be sent to the target. In this clone (or clones), all unnecessary or unwanted data is removed or modified. This clone is then snapshotted to create the "redaction snapshot" (or snapshots). Second, the new zfs redact command is used to create a redaction bookmark. The redaction bookmark stores the list of blocks in a snapshot that were modified by the redaction snapshot(s). Finally, the redaction bookmark is passed as a parameter to zfs send. When sending to the snapshot that was redacted, the redaction bookmark is used to filter out blocks that contain sensitive or unwanted information, and those blocks are not included in the send stream. When sending from the redaction bookmark, the blocks it contains are considered as candidate blocks in addition to those blocks in the destination snapshot that were modified since the creation_txg of the redaction bookmark. This step is necessary to allow the target to rehydrate data in the case where some blocks are accidentally or unnecessarily modified in the redaction snapshot. The changes to bookmarks to enable fast space estimation involve adding deadlists to bookmarks. There is also logic to manage the life cycles of these deadlists. The new size estimation process operates in cases where previously an accurate estimate could not be provided. In those cases, a send is performed where no data blocks are read, reducing the runtime significantly and providing a byte-accurate size estimate. Reviewed-by: Dan Kimmel <dan.kimmel@delphix.com> Reviewed-by: Matt Ahrens <mahrens@delphix.com> Reviewed-by: Prashanth Sreenivasa <pks@delphix.com> Reviewed-by: John Kennedy <john.kennedy@delphix.com> Reviewed-by: George Wilson <george.wilson@delphix.com> Reviewed-by: Chris Williamson <chris.williamson@delphix.com> Reviewed-by: Pavel Zhakarov <pavel.zakharov@delphix.com> Reviewed-by: Sebastien Roy <sebastien.roy@delphix.com> Reviewed-by: Prakash Surya <prakash.surya@delphix.com> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Paul Dagnelie <pcd@delphix.com> Closes #7958
2026-05-22 02:27:36 +03:00 · 2019-06-19 09:48:13 -07:00
parent c1b5801bb5
commit 30af21b025
103 changed files with 11513 additions and 2668 deletions
@@ -33,6 +33,7 @@

 #include <assert.h>
 #include <ctype.h>
+#include <sys/debug.h>
 #include <errno.h>
 #include <getopt.h>
 #include <libgen.h>
@@ -119,6 +120,7 @@ static int zfs_do_unload_key(int argc, char **argv);
 static int zfs_do_change_key(int argc, char **argv);
 static int zfs_do_project(int argc, char **argv);
 static int zfs_do_version(int argc, char **argv);
+static int zfs_do_redact(int argc, char **argv);

 /*
 * Enable a reasonable set of defaults for libumem debugging on DEBUG builds.
@@ -173,7 +175,8 @@ typedef enum {
 	HELP_LOAD_KEY,
 	HELP_UNLOAD_KEY,
 	HELP_CHANGE_KEY,
-	HELP_VERSION
+	HELP_VERSION,
+	HELP_REDACT,
 } zfs_help_t;

 typedef struct zfs_command {
@@ -238,6 +241,7 @@ static zfs_command_t command_table[] = {
 	{ "load-key",	zfs_do_load_key,	HELP_LOAD_KEY		},
 	{ "unload-key",	zfs_do_unload_key,	HELP_UNLOAD_KEY		},
 	{ "change-key",	zfs_do_change_key,	HELP_CHANGE_KEY		},
+	{ "redact",	zfs_do_redact,		HELP_REDACT		},
 };

 #define	NCOMMAND	(sizeof (command_table) / sizeof (command_table[0]))
@@ -279,7 +283,7 @@ get_usage(zfs_help_t idx)
 		    "[filesystem|volume|snapshot] ...\n"));
 	case HELP_MOUNT:
 		return (gettext("\tmount\n"
-		    "\tmount [-lvO] [-o opts] <-a | filesystem>\n"));
+		    "\tmount [-flvO] [-o opts] <-a | filesystem>\n"));
 	case HELP_PROMOTE:
 		return (gettext("\tpromote <clone-filesystem>\n"));
 	case HELP_RECEIVE:
@@ -302,6 +306,9 @@ get_usage(zfs_help_t idx)
 		    "<snapshot>\n"
 		    "\tsend [-nvPLecw] [-i snapshot|bookmark] "
 		    "<filesystem|volume|snapshot>\n"
+		    "[-i bookmark] <snapshot> <bookmark_name>\n"
+		    "\tsend [-DnPpvLecr] [-i bookmark|snapshot] "
+		    "--redact <bookmark> <snapshot>\n"
 		    "\tsend [-nvPe] -t <receive_resume_token>\n"));
 	case HELP_SET:
 		return (gettext("\tset <property=value> ... "
@@ -386,6 +393,9 @@ get_usage(zfs_help_t idx)
 		    "\tchange-key -i [-l] <filesystem|volume>\n"));
 	case HELP_VERSION:
 		return (gettext("\tversion\n"));
+	case HELP_REDACT:
+		return (gettext("\tredact <snapshot> <bookmark> "
+		    "<redaction_snapshot> ..."));
 	}

 	abort();
@@ -543,6 +553,8 @@ usage(boolean_t requested)
 		(void) fprintf(fp, "YES       NO   <size> | none\n");
 		(void) fprintf(fp, "\t%-15s ", "written@<snap>");
 		(void) fprintf(fp, " NO       NO   <size>\n");
+		(void) fprintf(fp, "\t%-15s ", "written#<bookmark>");
+		(void) fprintf(fp, " NO       NO   <size>\n");

 		(void) fprintf(fp, gettext("\nSizes are specified in bytes "
 		    "with standard units such as K, M, G, etc.\n"));
@@ -1501,6 +1513,13 @@ zfs_do_destroy(int argc, char **argv)
 			return (-1);
 		}

+		/*
+		 * Unfortunately, zfs_bookmark() doesn't honor the
+		 * casesensitivity setting.  However, we can't simply
+		 * remove this check, because lzc_destroy_bookmarks()
+		 * ignores non-existent bookmarks, so this is necessary
+		 * to get a proper error message.
+		 */
 		if (!zfs_bookmark_exists(argv[0])) {
 			(void) fprintf(stderr, gettext("bookmark '%s' "
 			    "does not exist.\n"), argv[0]);
@@ -3595,6 +3614,73 @@ zfs_do_promote(int argc, char **argv)
 	return (ret);
 }

+static int
+zfs_do_redact(int argc, char **argv)
+{
+	char *snap = NULL;
+	char *bookname = NULL;
+	char **rsnaps = NULL;
+	int numrsnaps = 0;
+	argv++;
+	argc--;
+	if (argc < 3) {
+		(void) fprintf(stderr, gettext("too few arguments"));
+		usage(B_FALSE);
+	}
+
+	snap = argv[0];
+	bookname = argv[1];
+	rsnaps = argv + 2;
+	numrsnaps = argc - 2;
+
+	nvlist_t *rsnapnv = fnvlist_alloc();
+
+	for (int i = 0; i < numrsnaps; i++) {
+		fnvlist_add_boolean(rsnapnv, rsnaps[i]);
+	}
+
+	int err = lzc_redact(snap, bookname, rsnapnv);
+	fnvlist_free(rsnapnv);
+
+	switch (err) {
+	case 0:
+		break;
+	case ENOENT:
+		(void) fprintf(stderr,
+		    gettext("provided snapshot %s does not exist"), snap);
+		break;
+	case EEXIST:
+		(void) fprintf(stderr, gettext("specified redaction bookmark "
+		    "(%s) provided already exists"), bookname);
+		break;
+	case ENAMETOOLONG:
+		(void) fprintf(stderr, gettext("provided bookmark name cannot "
+		    "be used, final name would be too long"));
+		break;
+	case E2BIG:
+		(void) fprintf(stderr, gettext("too many redaction snapshots "
+		    "specified"));
+		break;
+	case EINVAL:
+		(void) fprintf(stderr, gettext("redaction snapshot must be "
+		    "descendent of snapshot being redacted"));
+		break;
+	case EALREADY:
+		(void) fprintf(stderr, gettext("attempted to redact redacted "
+		    "dataset or with respect to redacted dataset"));
+		break;
+	case ENOTSUP:
+		(void) fprintf(stderr, gettext("redaction bookmarks feature "
+		    "not enabled"));
+		break;
+	default:
+		(void) fprintf(stderr, gettext("internal error: %s"),
+		    strerror(errno));
+	}
+
+	return (err);
+}
+
 /*
 * zfs rollback [-rRf] <snapshot>
 *
@@ -4006,6 +4092,7 @@ usage:
 	return (-1);
 }

+
 /*
 * Send a backup stream to stdout.
 */
@@ -4020,10 +4107,11 @@ zfs_do_send(int argc, char **argv)
 	sendflags_t flags = { 0 };
 	int c, err;
 	nvlist_t *dbgnv = NULL;
-	boolean_t extraverbose = B_FALSE;
+	char *redactbook = NULL;

 	struct option long_options[] = {
 		{"replicate",	no_argument,		NULL, 'R'},
+		{"redact",	required_argument,	NULL, 'd'},
 		{"props",	no_argument,		NULL, 'p'},
 		{"parsable",	no_argument,		NULL, 'P'},
 		{"dedup",	no_argument,		NULL, 'D'},
@@ -4040,8 +4128,8 @@ zfs_do_send(int argc, char **argv)
 	};

 	/* check options */
-	while ((c = getopt_long(argc, argv, ":i:I:RDpvnPLeht:cwb", long_options,
-	    NULL)) != -1) {
+	while ((c = getopt_long(argc, argv, ":i:I:RDpvnPLeht:cwbd:",
+	    long_options, NULL)) != -1) {
 		switch (c) {
 		case 'i':
 			if (fromname)
@@ -4057,6 +4145,9 @@ zfs_do_send(int argc, char **argv)
 		case 'R':
 			flags.replicate = B_TRUE;
 			break;
+		case 'd':
+			redactbook = optarg;
+			break;
 		case 'p':
 			flags.props = B_TRUE;
 			break;
@@ -4068,12 +4159,9 @@ zfs_do_send(int argc, char **argv)
 			break;
 		case 'P':
 			flags.parsable = B_TRUE;
-			flags.verbose = B_TRUE;
 			break;
 		case 'v':
-			if (flags.verbose)
-				extraverbose = B_TRUE;
-			flags.verbose = B_TRUE;
+			flags.verbosity++;
 			flags.progress = B_TRUE;
 			break;
 		case 'D':
@@ -4141,19 +4229,22 @@ zfs_do_send(int argc, char **argv)
 		}
 	}

+	if (flags.parsable && flags.verbosity == 0)
+		flags.verbosity = 1;
+
 	argc -= optind;
 	argv += optind;

 	if (resume_token != NULL) {
 		if (fromname != NULL || flags.replicate || flags.props ||
-		    flags.backup || flags.dedup) {
+		    flags.backup || flags.dedup || flags.holds ||
+		    redactbook != NULL) {
 			(void) fprintf(stderr,
 			    gettext("invalid flags combined with -t\n"));
 			usage(B_FALSE);
 		}
-		if (argc != 0) {
-			(void) fprintf(stderr, gettext("no additional "
-			    "arguments are permitted with -t\n"));
+		if (argc > 0) {
+			(void) fprintf(stderr, gettext("too many arguments\n"));
 			usage(B_FALSE);
 		}
 	} else {
@@ -4168,6 +4259,12 @@ zfs_do_send(int argc, char **argv)
 		}
 	}

+	if (flags.raw && redactbook != NULL) {
+		(void) fprintf(stderr,
+		    gettext("Error: raw sends may not be redacted.\n"));
+		return (1);
+	}
+
 	if (!flags.dryrun && isatty(STDOUT_FILENO)) {
 		(void) fprintf(stderr,
 		    gettext("Error: Stream can not be written to a terminal.\n"
@@ -4181,43 +4278,70 @@ zfs_do_send(int argc, char **argv)
 	}

 	/*
-	 * Special case sending a filesystem, or from a bookmark.
+	 * For everything except -R and -I, use the new, cleaner code path.
 	 */
-	if (strchr(argv[0], '@') == NULL ||
-	    (fromname && strchr(fromname, '#') != NULL)) {
+	if (!(flags.replicate || flags.doall)) {
 		char frombuf[ZFS_MAX_DATASET_NAME_LEN];

-		if (flags.replicate || flags.doall || flags.props ||
-		    flags.backup || flags.dedup || flags.holds ||
-		    (strchr(argv[0], '@') == NULL &&
-		    (flags.dryrun || flags.verbose || flags.progress))) {
-			(void) fprintf(stderr, gettext("Error: "
-			    "Unsupported flag with filesystem or bookmark.\n"));
-			return (1);
+		if (redactbook != NULL) {
+			if (strchr(argv[0], '@') == NULL) {
+				(void) fprintf(stderr, gettext("Error: Cannot "
+				    "do a redacted send to a filesystem.\n"));
+				return (1);
+			}
 		}

 		zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET);
 		if (zhp == NULL)
 			return (1);

+		if (fromname != NULL && (strchr(fromname, '#') == NULL &&
+		    strchr(fromname, '@') == NULL)) {
+			/*
+			 * Neither bookmark or snapshot was specified.  Print a
+			 * warning, and assume snapshot.
+			 */
+			(void) fprintf(stderr, "Warning: incremental source "
+			    "didn't specify type, assuming snapshot. Use '@' "
+			    "or '#' prefix to avoid ambiguity.\n");
+			(void) snprintf(frombuf, sizeof (frombuf), "@%s",
+			    fromname);
+			fromname = frombuf;
+		}
 		if (fromname != NULL &&
 		    (fromname[0] == '#' || fromname[0] == '@')) {
 			/*
 			 * Incremental source name begins with # or @.
 			 * Default to same fs as target.
 			 */
+			char tmpbuf[ZFS_MAX_DATASET_NAME_LEN];
+			(void) strlcpy(tmpbuf, fromname, sizeof (tmpbuf));
 			(void) strlcpy(frombuf, argv[0], sizeof (frombuf));
 			cp = strchr(frombuf, '@');
 			if (cp != NULL)
 				*cp = '\0';
-			(void) strlcat(frombuf, fromname, sizeof (frombuf));
+			(void) strlcat(frombuf, tmpbuf, sizeof (frombuf));
 			fromname = frombuf;
 		}
-		err = zfs_send_one(zhp, fromname, STDOUT_FILENO, flags);
+		err = zfs_send_one(zhp, fromname, STDOUT_FILENO, &flags,
+		    redactbook);
 		zfs_close(zhp);
 		return (err != 0);
 	}

+	if (fromname != NULL && strchr(fromname, '#')) {
+		(void) fprintf(stderr,
+		    gettext("Error: multiple snapshots cannot be "
+		    "sent from a bookmark.\n"));
+		return (1);
+	}
+
+	if (redactbook != NULL) {
+		(void) fprintf(stderr, gettext("Error: multiple snapshots "
+		    "cannot be sent redacted.\n"));
+		return (1);
+	}
+
 	cp = strchr(argv[0], '@');
 	*cp = '\0';
 	toname = cp + 1;
@@ -4261,9 +4385,9 @@ zfs_do_send(int argc, char **argv)
 		flags.doall = B_TRUE;

 	err = zfs_send(zhp, fromname, toname, &flags, STDOUT_FILENO, NULL, 0,
-	    extraverbose ? &dbgnv : NULL);
+	    flags.verbosity >= 3 ? &dbgnv : NULL);

-	if (extraverbose && dbgnv != NULL) {
+	if (flags.verbosity >= 3 && dbgnv != NULL) {
 		/*
 		 * dump_nvlist prints to stdout, but that's been
 		 * redirected to a file.  Make it print to stderr
@@ -6379,6 +6503,17 @@ share_mount_one(zfs_handle_t *zhp, int op, int flags, char *protocol,
 		return (1);
 	}

+	if (zfs_prop_get_int(zhp, ZFS_PROP_REDACTED) && !(flags & MS_FORCE)) {
+		if (!explicit)
+			return (0);
+
+		(void) fprintf(stderr, gettext("cannot %s '%s': "
+		    "Dataset is not complete, was created by receiving "
+		    "a redacted zfs send stream.\n"), cmdname,
+		    zfs_get_name(zhp));
+		return (1);
+	}
+
 	/*
 	 * At this point, we have verified that the mountpoint and/or
 	 * shareopts are appropriate for auto management. If the
@@ -6537,7 +6672,7 @@ share_mount(int op, int argc, char **argv)
 	int flags = 0;

 	/* check options */
-	while ((c = getopt(argc, argv, op == OP_MOUNT ? ":alvo:O" : "al"))
+	while ((c = getopt(argc, argv, op == OP_MOUNT ? ":alvo:Of" : "al"))
 	    != -1) {
 		switch (c) {
 		case 'a':
@@ -6565,6 +6700,9 @@ share_mount(int op, int argc, char **argv)
 		case 'O':
 			flags |= MS_OVERLAY;
 			break;
+		case 'f':
+			flags |= MS_FORCE;
+			break;
 		case ':':
 			(void) fprintf(stderr, gettext("missing argument for "
 			    "'%c' option\n"), optopt);