Remove dedupditto functionality

If dedup is in use, the `dedupditto` property can be set, causing ZFS to
keep an extra copy of data that is referenced many times (>100x).  The
idea was that this data is more important than other data and thus we
want to be really sure that it is not lost if the disk experiences a
small amount of random corruption.

ZFS (and system administrators) rely on the pool-level redundancy to
protect their data (e.g. mirroring or RAIDZ).  Since the user/sysadmin
doesn't have control over what data will be offered extra redundancy by
dedupditto, this extra redundancy is not very useful.  The bulk of the
data is still vulnerable to loss based on the pool-level redundancy.
For example, if particle strikes corrupt 0.1% of blocks, you will either
be saved by mirror/raidz, or you will be sad.  This is true even if
dedupditto saved another 0.01% of blocks from being corrupted.

Therefore, the dedupditto functionality is rarely enabled (i.e. the
property is rarely set), and it fulfills its promise of increased
redundancy even more rarely.

Additionally, this feature does not work as advertised (on existing
releases), because scrub/resilver did not repair the extra (dedupditto)
copy (see #8270).

In summary, this seldom-used feature doesn't work, and even if it did it
wouldn't provide useful data protection.  It has a non-trivial
maintenance burden (again see #8270).

We should remove the dedupditto functionality.  For backwards
compatibility with the existing CLI, "zpool set dedupditto" will still
"succeed" (exit code zero), but won't have any effect.  For backwards
compatibility with existing pools that had dedupditto enabled at some
point, the code will still be able to understand dedupditto blocks and
free them when appropriate.  However, ZFS won't write any new dedupditto
blocks.

Author: ahrens

cmd/ztest/ztest.c

@@ -358,7 +358,6 @@ ztest_func_t ztest_dsl_prop_get_set;
 ztest_func_t ztest_spa_prop_get_set;
 ztest_func_t ztest_spa_create_destroy;
 ztest_func_t ztest_fault_inject;
-ztest_func_t ztest_ddt_repair;
 ztest_func_t ztest_dmu_snapshot_hold;
 ztest_func_t ztest_mmp_enable_disable;
 ztest_func_t ztest_scrub;
@@ -412,7 +411,6 @@ ztest_info_t ztest_info[] = {
 	ZTI_INIT(ztest_dmu_snapshot_create_destroy, 1, &zopt_sometimes),
 	ZTI_INIT(ztest_spa_create_destroy, 1, &zopt_sometimes),
 	ZTI_INIT(ztest_fault_inject, 1, &zopt_sometimes),
-	ZTI_INIT(ztest_ddt_repair, 1, &zopt_sometimes),
 	ZTI_INIT(ztest_dmu_snapshot_hold, 1, &zopt_sometimes),
 	ZTI_INIT(ztest_mmp_enable_disable, 1, &zopt_sometimes),
 	ZTI_INIT(ztest_reguid, 1, &zopt_rarely),
@@ -1281,18 +1279,18 @@ ztest_dsl_prop_set_uint64(char *osname, zfs_prop_t prop, uint64_t value,
 }
 
 static int
-ztest_spa_prop_set_uint64(zpool_prop_t prop, uint64_t value)
+ztest_spa_prop_set_string(zpool_prop_t prop, const char *value)
 {
 	spa_t *spa = ztest_spa;
 	nvlist_t *props = NULL;
 	int error;
 
-	VERIFY(nvlist_alloc(&props, NV_UNIQUE_NAME, 0) == 0);
-	VERIFY(nvlist_add_uint64(props, zpool_prop_to_name(prop), value) == 0);
+	props = fnvlist_alloc();
+	fnvlist_add_string(props, zpool_prop_to_name(prop), value);
 
 	error = spa_prop_set(spa, props);
 
-	nvlist_free(props);
+	fnvlist_free(props);
 
 	if (error == ENOSPC) {
 		ztest_record_enospc(FTAG);
@@ -1520,31 +1518,6 @@ ztest_tx_assign(dmu_tx_t *tx, uint64_t txg_how, const char *tag)
 	return (txg);
 }
 
-static void
-ztest_pattern_set(void *buf, uint64_t size, uint64_t value)
-{
-	uint64_t *ip = buf;
-	uint64_t *ip_end = (uint64_t *)((uintptr_t)buf + (uintptr_t)size);
-
-	while (ip < ip_end)
-		*ip++ = value;
-}
-
-#ifndef NDEBUG
-static boolean_t
-ztest_pattern_match(void *buf, uint64_t size, uint64_t value)
-{
-	uint64_t *ip = buf;
-	uint64_t *ip_end = (uint64_t *)((uintptr_t)buf + (uintptr_t)size);
-	uint64_t diff = 0;
-
-	while (ip < ip_end)
-		diff |= (value - *ip++);
-
-	return (diff == 0);
-}
-#endif
-
 static void
 ztest_bt_generate(ztest_block_tag_t *bt, objset_t *os, uint64_t object,
     uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg,
@@ -5564,8 +5537,10 @@ ztest_spa_prop_get_set(ztest_ds_t *zd, uint64_t id)
 
 	(void) pthread_rwlock_rdlock(&ztest_name_lock);
 
-	(void) ztest_spa_prop_set_uint64(ZPOOL_PROP_DEDUPDITTO,
-	    ZIO_DEDUPDITTO_MIN + ztest_random(ZIO_DEDUPDITTO_MIN));
+	char *propval = kmem_asprintf("test property value %u",
+	    (int)ztest_random(100));
+	(void) ztest_spa_prop_set_string(ZPOOL_PROP_COMMENT, propval);
+	strfree(propval);
 
 	VERIFY0(spa_prop_get(ztest_spa, &props));
 
@@ -6005,136 +5980,6 @@ ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
 	umem_free(pathrand, MAXPATHLEN);
 }
 
-/*
- * Verify that DDT repair works as expected.
- */
-void
-ztest_ddt_repair(ztest_ds_t *zd, uint64_t id)
-{
-	ztest_shared_t *zs = ztest_shared;
-	spa_t *spa = ztest_spa;
-	objset_t *os = zd->zd_os;
-	ztest_od_t *od;
-	uint64_t object, blocksize, txg, pattern;
-	enum zio_checksum checksum = spa_dedup_checksum(spa);
-	dmu_buf_t *db;
-	dmu_tx_t *tx;
-
-	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
-	ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0);
-
-	if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) {
-		umem_free(od, sizeof (ztest_od_t));
-		return;
-	}
-
-	/*
-	 * Take the name lock as writer to prevent anyone else from changing
-	 * the pool and dataset properties we need to maintain during this test.
-	 */
-	(void) pthread_rwlock_wrlock(&ztest_name_lock);
-
-	if (ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_DEDUP, checksum,
-	    B_FALSE) != 0 ||
-	    ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_COPIES, 1,
-	    B_FALSE) != 0) {
-		(void) pthread_rwlock_unlock(&ztest_name_lock);
-		umem_free(od, sizeof (ztest_od_t));
-		return;
-	}
-
-	dmu_objset_stats_t dds;
-	dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
-	dmu_objset_fast_stat(os, &dds);
-	dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
-
-	object = od[0].od_object;
-	blocksize = od[0].od_blocksize;
-	pattern = zs->zs_guid ^ dds.dds_guid;
-
-	/*
-	 * The numbers of copies written must always be greater than or
-	 * equal to the threshold set by the dedupditto property.  This
-	 * is initialized in ztest_run() and then randomly changed by
-	 * ztest_spa_prop_get_set(), these function will never set it
-	 * larger than 2 * ZIO_DEDUPDITTO_MIN.
-	 */
-	int copies = 2 * ZIO_DEDUPDITTO_MIN;
-
-	/*
-	 * The block size is limited by DMU_MAX_ACCESS (64MB) which
-	 * caps the maximum transaction size.  A block size of up to
-	 * SPA_OLD_MAXBLOCKSIZE is allowed which results in a maximum
-	 * transaction size of: 128K * 200 (copies) = ~25MB
-	 *
-	 * The actual block size is checked here, rather than requested
-	 * above, because the way ztest_od_init() is implemented it does
-	 * not guarantee the block size requested will be used.
-	 */
-	if (blocksize > SPA_OLD_MAXBLOCKSIZE) {
-		(void) pthread_rwlock_unlock(&ztest_name_lock);
-		umem_free(od, sizeof (ztest_od_t));
-		return;
-	}
-
-	ASSERT(object != 0);
-
-	tx = dmu_tx_create(os);
-	dmu_tx_hold_write(tx, object, 0, copies * blocksize);
-	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
-	if (txg == 0) {
-		(void) pthread_rwlock_unlock(&ztest_name_lock);
-		umem_free(od, sizeof (ztest_od_t));
-		return;
-	}
-
-	/*
-	 * Write all the copies of our block.
-	 */
-	for (int i = 0; i < copies; i++) {
-		uint64_t offset = i * blocksize;
-		int error = dmu_buf_hold(os, object, offset, FTAG, &db,
-		    DMU_READ_NO_PREFETCH);
-		if (error != 0) {
-			fatal(B_FALSE, "dmu_buf_hold(%p, %llu, %llu) = %u",
-			    os, (long long)object, (long long) offset, error);
-		}
-		ASSERT(db->db_offset == offset);
-		ASSERT(db->db_size == blocksize);
-		ASSERT(ztest_pattern_match(db->db_data, db->db_size, pattern) ||
-		    ztest_pattern_match(db->db_data, db->db_size, 0ULL));
-		dmu_buf_will_fill(db, tx);
-		ztest_pattern_set(db->db_data, db->db_size, pattern);
-		dmu_buf_rele(db, FTAG);
-	}
-
-	dmu_tx_commit(tx);
-	txg_wait_synced(spa_get_dsl(spa), txg);
-
-	/*
-	 * Find out what block we got.
-	 */
-	VERIFY0(dmu_buf_hold(os, object, 0, FTAG, &db, DMU_READ_NO_PREFETCH));
-	blkptr_t blk = *((dmu_buf_impl_t *)db)->db_blkptr;
-	dmu_buf_rele(db, FTAG);
-
-	/*
-	 * Damage the block.  Dedup-ditto will save us when we read it later.
-	 */
-	uint64_t psize = BP_GET_PSIZE(&blk);
-	abd_t *abd = abd_alloc_linear(psize, B_TRUE);
-	ztest_pattern_set(abd_to_buf(abd), psize, ~pattern);
-
-	(void) zio_wait(zio_rewrite(NULL, spa, 0, &blk,
-	    abd, psize, NULL, NULL, ZIO_PRIORITY_SYNC_WRITE,
-	    ZIO_FLAG_CANFAIL | ZIO_FLAG_INDUCE_DAMAGE, NULL));
-
-	abd_free(abd);
-
-	(void) pthread_rwlock_unlock(&ztest_name_lock);
-	umem_free(od, sizeof (ztest_od_t));
-}
-
 /*
  * By design ztest will never inject uncorrectable damage in to the pool.
  * Issue a scrub, wait for it to complete, and verify there is never any
@@ -7015,8 +6860,6 @@ ztest_run(ztest_shared_t *zs)
 	zs->zs_guid = dds.dds_guid;
 	dmu_objset_disown(os, B_TRUE, FTAG);
 
-	spa->spa_dedup_ditto = 2 * ZIO_DEDUPDITTO_MIN;
-
 	/*
 	 * Create a thread to periodically resume suspended I/O.
 	 */

include/sys/ddt.h

@@ -103,6 +103,10 @@ typedef struct ddt_phys {
 	uint64_t	ddp_phys_birth;
 } ddt_phys_t;
 
+/*
+ * Note, we no longer generate new DDT_PHYS_DITTO-type blocks.  However,
+ * we maintain the ability to free existing dedup-ditto blocks.
+ */
 enum ddt_phys_type {
 	DDT_PHYS_DITTO = 0,
 	DDT_PHYS_SINGLE = 1,
@@ -216,10 +220,6 @@ extern void ddt_get_dedup_stats(spa_t *spa, ddt_stat_t *dds_total);
 extern uint64_t ddt_get_dedup_dspace(spa_t *spa);
 extern uint64_t ddt_get_pool_dedup_ratio(spa_t *spa);
 
-extern int ddt_ditto_copies_needed(ddt_t *ddt, ddt_entry_t *dde,
-    ddt_phys_t *ddp_willref);
-extern int ddt_ditto_copies_present(ddt_entry_t *dde);
-
 extern size_t ddt_compress(void *src, uchar_t *dst, size_t s_len, size_t d_len);
 extern void ddt_decompress(uchar_t *src, void *dst, size_t s_len, size_t d_len);
 

include/sys/spa_impl.h

@@ -350,7 +350,6 @@ struct spa {
 	ddt_t		*spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */
 	uint64_t	spa_ddt_stat_object;	/* DDT statistics */
 	uint64_t	spa_dedup_dspace;	/* Cache get_dedup_dspace() */
-	uint64_t	spa_dedup_ditto;	/* dedup ditto threshold */
 	uint64_t	spa_dedup_checksum;	/* default dedup checksum */
 	uint64_t	spa_dspace;		/* dspace in normal class */
 	kmutex_t	spa_vdev_top_lock;	/* dueling offline/remove */

include/sys/zio.h

@@ -102,7 +102,6 @@ enum zio_checksum {
 #define	ZIO_CHECKSUM_VERIFY	(1 << 8)
 
 #define	ZIO_DEDUPCHECKSUM	ZIO_CHECKSUM_SHA256
-#define	ZIO_DEDUPDITTO_MIN	100
 
 /* supported encryption algorithms */
 enum zio_encrypt {

lib/libzfs/libzfs_pool.c

@@ -722,15 +722,8 @@ zpool_valid_proplist(libzfs_handle_t *hdl, const char *poolname,
 			}
 			break;
 		case ZPOOL_PROP_DEDUPDITTO:
-			if (intval < ZIO_DEDUPDITTO_MIN && intval != 0) {
-				zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
-				    "property '%s' value %d is invalid; only "
-				    "values of 0 or >= %" PRId32 " are allowed "
-				    "for this property."),
-				    propname, intval, ZIO_DEDUPDITTO_MIN);
-				(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
-				goto error;
-			}
+			printf("Note: property '%s' no longer has "
+			    "any effect\n", propname);
 			break;
 
 		default:

man/man8/zpool.8

@@ -768,14 +768,7 @@ such that it is available even if the pool becomes faulted.
 An administrator can provide additional information about a pool using this
 property.
 .It Sy dedupditto Ns = Ns Ar number
-Threshold for the number of block ditto copies.
-If the reference count for a deduplicated block increases above this number, a
-new ditto copy of this block is automatically stored.
-The default setting is
-.Sy 0
-which causes no ditto copies to be created for deduplicated blocks.
-The minimum legal nonzero setting is
-.Sy 100 .
+This property is deprecated and no longer has any effect.
 .It Sy delegation Ns = Ns Sy on Ns | Ns Sy off
 Controls whether a non-privileged user is granted access based on the dataset
 permissions defined on the dataset.

module/zcommon/zpool_prop.c

@@ -104,8 +104,6 @@ zpool_prop_init(void)
 	/* default number properties */
 	zprop_register_number(ZPOOL_PROP_VERSION, "version", SPA_VERSION,
 	    PROP_DEFAULT, ZFS_TYPE_POOL, "<version>", "VERSION");
-	zprop_register_number(ZPOOL_PROP_DEDUPDITTO, "dedupditto", 0,
-	    PROP_DEFAULT, ZFS_TYPE_POOL, "<threshold (min 100)>", "DEDUPDITTO");
 	zprop_register_number(ZPOOL_PROP_ASHIFT, "ashift", 0, PROP_DEFAULT,
 	    ZFS_TYPE_POOL, "<ashift, 9-16, or 0=default>", "ASHIFT");
 
@@ -140,6 +138,8 @@ zpool_prop_init(void)
 	    PROP_ONETIME, ZFS_TYPE_POOL, "TNAME");
 	zprop_register_hidden(ZPOOL_PROP_MAXDNODESIZE, "maxdnodesize",
 	    PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_POOL, "MAXDNODESIZE");
+	zprop_register_hidden(ZPOOL_PROP_DEDUPDITTO, "dedupditto",
+	    PROP_TYPE_NUMBER, PROP_DEFAULT, ZFS_TYPE_POOL, "DEDUPDITTO");
 }
 
 /*