Allow zfs to send replication streams with missing snapshots

A tentative implementation and discussion was done in openzfs#5285.
According to it a send --skip-missing|-s flag has been added.
In a replication stream, when there are snapshots missing in
the hierarchy, if -s is provided print a warning and ignore
dataset (and its children) instead of throwing an error

Reviewed-by: Paul Dagnelie <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Pablo Correa Gómez <[email protected]>
Closes openzfs#11710

Author: pabloyoyoista

cmd/zfs/zfs_main.c

@@ -4376,6 +4376,7 @@ zfs_do_send(int argc, char **argv)
 
 	struct option long_options[] = {
 		{"replicate",	no_argument,		NULL, 'R'},
+		{"skip-missing",	no_argument,		NULL, 's'},
 		{"redact",	required_argument,	NULL, 'd'},
 		{"props",	no_argument,		NULL, 'p'},
 		{"parsable",	no_argument,		NULL, 'P'},
@@ -4394,7 +4395,7 @@ zfs_do_send(int argc, char **argv)
 	};
 
 	/* check options */
-	while ((c = getopt_long(argc, argv, ":i:I:RDpvnPLeht:cwbd:S",
+	while ((c = getopt_long(argc, argv, ":i:I:RsDpvnPLeht:cwbd:S",
 	    long_options, NULL)) != -1) {
 		switch (c) {
 		case 'i':
@@ -4411,6 +4412,9 @@ zfs_do_send(int argc, char **argv)
 		case 'R':
 			flags.replicate = B_TRUE;
 			break;
+		case 's':
+			flags.skipmissing = B_TRUE;
+			break;
 		case 'd':
 			redactbook = optarg;
 			break;
@@ -4575,6 +4579,13 @@ zfs_do_send(int argc, char **argv)
 		    resume_token));
 	}
 
+	if (flags.skipmissing && !flags.replicate) {
+		(void) fprintf(stderr,
+		    gettext("skip-missing flag can only be used in "
+		    "conjunction with replicate\n"));
+		usage(B_FALSE);
+	}
+
 	/*
 	 * For everything except -R and -I, use the new, cleaner code path.
 	 */

include/libzfs.h

@@ -666,6 +666,9 @@ typedef struct sendflags {
 	/* recursive send  (ie, -R) */
 	boolean_t replicate;
 
+	/* for recursive send, skip sending missing snapshots */
+	boolean_t skipmissing;
+
 	/* for incrementals, do all intermediate snapshots */
 	boolean_t doall;
 

lib/libzfs/libzfs_sendrecv.c

@@ -247,6 +247,7 @@ typedef struct send_data {
 	boolean_t raw;
 	boolean_t doall;
 	boolean_t replicate;
+	boolean_t skipmissing;
 	boolean_t verbose;
 	boolean_t backup;
 	boolean_t seenfrom;
@@ -497,7 +498,8 @@ send_iterate_fs(zfs_handle_t *zhp, void *arg)
 	 * - skip sending the current dataset if it was created later than
 	 *   the parent tosnap
 	 * - return error if the current dataset was created earlier than
-	 *   the parent tosnap
+	 *   the parent tosnap, unless --skip-missing specified. Then
+	 *   just print a warning
 	 */
 	if (sd->tosnap != NULL && tosnap_txg == 0) {
 		if (sd->tosnap_txg != 0 && txg > sd->tosnap_txg) {
@@ -506,6 +508,11 @@ send_iterate_fs(zfs_handle_t *zhp, void *arg)
 				    "skipping dataset %s: snapshot %s does "
 				    "not exist\n"), zhp->zfs_name, sd->tosnap);
 			}
+		} else if (sd->skipmissing) {
+			(void) fprintf(stderr, dgettext(TEXT_DOMAIN,
+			    "WARNING: skipping dataset %s and its children:"
+			    " snapshot %s does not exist\n"),
+			    zhp->zfs_name, sd->tosnap);
 		} else {
 			(void) fprintf(stderr, dgettext(TEXT_DOMAIN,
 			    "cannot send %s@%s%s: snapshot %s@%s does not "
@@ -649,8 +656,9 @@ send_iterate_fs(zfs_handle_t *zhp, void *arg)
 static int
 gather_nvlist(libzfs_handle_t *hdl, const char *fsname, const char *fromsnap,
     const char *tosnap, boolean_t recursive, boolean_t raw, boolean_t doall,
-    boolean_t replicate, boolean_t verbose, boolean_t backup, boolean_t holds,
-    boolean_t props, nvlist_t **nvlp, avl_tree_t **avlp)
+    boolean_t replicate, boolean_t skipmissing, boolean_t verbose,
+    boolean_t backup, boolean_t holds, boolean_t props, nvlist_t **nvlp,
+    avl_tree_t **avlp)
 {
 	zfs_handle_t *zhp;
 	send_data_t sd = { 0 };
@@ -668,6 +676,7 @@ gather_nvlist(libzfs_handle_t *hdl, const char *fsname, const char *fromsnap,
 	sd.raw = raw;
 	sd.doall = doall;
 	sd.replicate = replicate;
+	sd.skipmissing = skipmissing;
 	sd.verbose = verbose;
 	sd.backup = backup;
 	sd.holds = holds;
@@ -1975,8 +1984,8 @@ send_conclusion_record(int fd, zio_cksum_t *zc)
 static int
 send_prelim_records(zfs_handle_t *zhp, const char *from, int fd,
     boolean_t gather_props, boolean_t recursive, boolean_t verbose,
-    boolean_t dryrun, boolean_t raw, boolean_t replicate, boolean_t backup,
-    boolean_t holds, boolean_t props, boolean_t doall,
+    boolean_t dryrun, boolean_t raw, boolean_t replicate, boolean_t skipmissing,
+    boolean_t backup, boolean_t holds, boolean_t props, boolean_t doall,
     nvlist_t **fssp, avl_tree_t **fsavlp)
 {
 	int err = 0;
@@ -2022,8 +2031,8 @@ send_prelim_records(zfs_handle_t *zhp, const char *from, int fd,
 		}
 
 		if ((err = gather_nvlist(zhp->zfs_hdl, tofs,
-		    from, tosnap, recursive, raw, doall, replicate, verbose,
-		    backup, holds, props, &fss, fsavlp)) != 0) {
+		    from, tosnap, recursive, raw, doall, replicate, skipmissing,
+		    verbose, backup, holds, props, &fss, fsavlp)) != 0) {
 			return (zfs_error(zhp->zfs_hdl, EZFS_BADBACKUP,
 			    errbuf));
 		}
@@ -2160,8 +2169,9 @@ zfs_send(zfs_handle_t *zhp, const char *fromsnap, const char *tosnap,
 		err = send_prelim_records(tosnap, fromsnap, outfd,
 		    flags->replicate || flags->props || flags->holds,
 		    flags->replicate, flags->verbosity > 0, flags->dryrun,
-		    flags->raw, flags->replicate, flags->backup, flags->holds,
-		    flags->props, flags->doall, &fss, &fsavl);
+		    flags->raw, flags->replicate, flags->skipmissing,
+		    flags->backup, flags->holds, flags->props, flags->doall,
+		    &fss, &fsavl);
 		zfs_close(tosnap);
 		if (err != 0)
 			goto err_out;
@@ -2464,7 +2474,7 @@ zfs_send_one(zfs_handle_t *zhp, const char *from, int fd, sendflags_t *flags,
 		 */
 		err = send_prelim_records(zhp, NULL, fd, B_TRUE, B_FALSE,
 		    flags->verbosity > 0, flags->dryrun, flags->raw,
-		    flags->replicate, flags->backup, flags->holds,
+		    flags->replicate, B_FALSE, flags->backup, flags->holds,
 		    flags->props, flags->doall, NULL, NULL);
 		if (err != 0)
 			return (err);
@@ -3236,7 +3246,7 @@ recv_incremental_replication(libzfs_handle_t *hdl, const char *tofs,
 	deleted = fnvlist_alloc();
 
 	if ((error = gather_nvlist(hdl, tofs, fromsnap, NULL,
-	    recursive, B_TRUE, B_FALSE, recursive, B_FALSE, B_FALSE,
+	    recursive, B_TRUE, B_FALSE, recursive, B_FALSE, B_FALSE, B_FALSE,
 	    B_FALSE, B_TRUE, &local_nv, &local_avl)) != 0)
 		return (error);
 
@@ -4729,8 +4739,8 @@ zfs_receive_one(libzfs_handle_t *hdl, int infd, const char *tosnap,
 		 */
 		*cp = '\0';
 		if (gather_nvlist(hdl, destsnap, NULL, NULL, B_FALSE, B_TRUE,
-		    B_FALSE, B_FALSE, B_FALSE, B_FALSE, B_FALSE, B_TRUE,
-		    &local_nv, &local_avl) == 0) {
+		    B_FALSE, B_FALSE, B_FALSE, B_FALSE, B_FALSE, B_FALSE,
+		    B_TRUE, &local_nv, &local_avl) == 0) {
 			*cp = '@';
 			fs = fsavl_find(local_avl, drrb->drr_toguid, NULL);
 			fsavl_destroy(local_avl);

man/man8/zfs-send.8

@@ -39,12 +39,12 @@
 .Sh SYNOPSIS
 .Nm zfs
 .Cm send
-.Op Fl DLPRbcehnpvw
+.Op Fl DLPRsbcehnpvw
 .Op Oo Fl I Ns | Ns Fl i Oc Ar snapshot
 .Ar snapshot
 .Nm zfs
 .Cm send
-.Op Fl DLPRcenpvw
+.Op Fl DLPRscenpvw
 .Op Fl i Ar snapshot Ns | Ns Ar bookmark
 .Ar filesystem Ns | Ns Ar volume Ns | Ns Ar snapshot
 .Nm zfs
@@ -139,6 +139,12 @@ do not exist on the sending side are destroyed. If the
 flag is used to send encrypted datasets, then
 .Fl w
 must also be specified.
+.It Fl s, -skip-missing
+Allows sending a replication stream even when there are snapshots missing in the
+hierarchy. When a snapshot is missing, instead of throwing an error and aborting
+the send, a warning is printed to STDERR and the dataset to which it belongs
+and its descendents are skipped. This flag can only be used in conjunction with
+.Fl R .
 .It Fl e, -embed
 Generate a more compact stream by using
 .Sy WRITE_EMBEDDED

tests/runfiles/common.run

@@ -258,7 +258,7 @@ tags = ['functional', 'cli_root', 'zfs_rollback']
 tests = ['zfs_send_001_pos', 'zfs_send_002_pos', 'zfs_send_003_pos',
     'zfs_send_004_neg', 'zfs_send_005_pos', 'zfs_send_006_pos',
     'zfs_send_007_pos', 'zfs_send_encrypted', 'zfs_send_raw',
-    'zfs_send_sparse', 'zfs_send-b']
+    'zfs_send_sparse', 'zfs_send-b', 'zfs_send_skip_missing']
 tags = ['functional', 'cli_root', 'zfs_send']
 
 [tests/functional/cli_root/zfs_set]

tests/zfs-tests/tests/functional/cli_root/zfs_send/Makefile.am

@@ -13,7 +13,8 @@ dist_pkgdata_SCRIPTS = \
 	zfs_send_encrypted_unloaded.ksh \
 	zfs_send_raw.ksh \
 	zfs_send_sparse.ksh \
-	zfs_send-b.ksh
+	zfs_send-b.ksh \
+	zfs_send_skip_missing.ksh
 
 dist_pkgdata_DATA = \
 	zfs_send.cfg

tests/zfs-tests/tests/functional/cli_root/zfs_send/zfs_send_skip_missing.ksh

@@ -0,0 +1,77 @@
+#!/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) 2016, loli10K. All rights reserved.
+# Copyright (c) 2021, Pablo Correa Gómez. All rights reserved.
+#
+
+. $STF_SUITE/tests/functional/cli_root/cli_common.kshlib
+. $STF_SUITE/tests/functional/cli_root/zfs_send/zfs_send.cfg
+
+#
+# DESCRIPTION:
+#	Verify 'zfs send' will avoid sending replication send
+#	streams when we're missing snapshots in the dataset
+#	hierarchy, unless -s|--skip-missing provided
+#
+# STRATEGY:
+#	1. Create a parent and child fs and then only snapshot the parent
+#	2. Verify sending with replication will fail
+#	3. Verify sending with skip-missing will print a warning but succeed
+#
+
+verify_runnable "both"
+
+function cleanup
+{
+	snapexists $SNAP && log_must zfs destroy -f $SNAP
+
+	datasetexists $PARENT && log_must zfs destroy -rf $PARENT
+
+	[[ -e $WARNF ]] && log_must rm -f $WARNF
+	rm -f $TEST_BASE_DIR/devnull
+}
+
+log_assert "Verify 'zfs send -Rs' works as expected."
+log_onexit cleanup
+
+PARENT=$TESTPOOL/parent
+CHILD=$PARENT/child
+SNAP=$PARENT@snap
+WARNF=$TEST_BASE_DIR/warn.2
+
+log_note "Verify 'zfs send -R' fails to generate replication stream"\
+	 " for datasets created before"
+
+log_must zfs create $PARENT
+log_must zfs create $CHILD
+log_must zfs snapshot $SNAP
+log_mustnot eval "zfs send -R $SNAP >$TEST_BASE_DIR/devnull"
+
+log_note "Verify 'zfs send -Rs' warns about missing snapshots, "\
+	 "but still succeeds"
+
+log_must eval "zfs send -Rs $SNAP 2> $WARNF >$TEST_BASE_DIR/devnull"
+log_must eval "[[ -s $WARNF ]]"
+
+log_pass "Verify 'zfs send -Rs' works as expected."