mirror of
https://git.proxmox.com/git/mirror_zfs.git
synced 2024-12-26 19:19:32 +03:00
f4aeb23f52
Traditional dedup keeps a separate ddt_phys_t "type" for each possible count of DVAs (that is, copies=) parameter. Each of these are tracked independently of each other, and have their own set of DVAs. This leads to an (admittedly rare) situation where you can create as many as six copies of the data, by changing the copies= parameter between copying. This is both a waste of storage on disk, but also a waste of space in the stored DDT entries, since there never needs to be more than three DVAs to handle all possible values of copies=. This commit adds a new FDT feature, DDT_FLAG_FLAT. When active, only the first ddt_phys_t is used. Each time a block is written with the dedup bit set, this single phys is checked to see if it has enough DVAs to fulfill the request. If it does, the block is filled with the saved DVAs as normal. If not, an adjusted write is issued to create as many extra copies as are needed to fulfill the request, which are then saved into the entry too. Because a single phys is no longer an all-or-nothing, but can be transitioning from fewer to more DVAs, the write path now has to keep a copy of the previous "known good" DVA set so we can revert to it in case an error occurs. zio_ddt_write() has been restructured and heavily commented to make it much easier to see what's happening. Backwards compatibility is maintained simply by allocating four ddt_phys_t when the DDT_FLAG_FLAT flag is not set, and updating the phys selection macros to check the flag. In the old arrangement, each number of copies gets a whole phys, so it will always have either zero or all necessary DVAs filled, with no in-between, so the old behaviour naturally falls out of the new code. Reviewed-by: Alexander Motin <mav@FreeBSD.org> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Rob Norris <rob.norris@klarasystems.com> Co-authored-by: Don Brady <don.brady@klarasystems.com> Sponsored-by: Klara, Inc. Sponsored-by: iXsystems, Inc. Closes #15893
222 lines
8.3 KiB
C
222 lines
8.3 KiB
C
/*
|
|
* 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 https://opensource.org/licenses/CDDL-1.0.
|
|
* 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) 2010, Oracle and/or its affiliates. All rights reserved.
|
|
* Copyright (c) 2012, 2017 by Delphix. All rights reserved.
|
|
* Copyright (c) 2017, 2019, Datto Inc. All rights reserved.
|
|
*/
|
|
|
|
#ifndef _SYS_DSL_SCAN_H
|
|
#define _SYS_DSL_SCAN_H
|
|
|
|
#include <sys/zfs_context.h>
|
|
#include <sys/zio.h>
|
|
#include <sys/zap.h>
|
|
#include <sys/ddt.h>
|
|
#include <sys/bplist.h>
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
struct objset;
|
|
struct dsl_dir;
|
|
struct dsl_dataset;
|
|
struct dsl_pool;
|
|
struct dmu_tx;
|
|
|
|
extern int zfs_scan_suspend_progress;
|
|
|
|
/*
|
|
* All members of this structure must be uint64_t, for byteswap
|
|
* purposes.
|
|
*/
|
|
typedef struct dsl_scan_phys {
|
|
uint64_t scn_func; /* pool_scan_func_t */
|
|
uint64_t scn_state; /* dsl_scan_state_t */
|
|
uint64_t scn_queue_obj;
|
|
uint64_t scn_min_txg;
|
|
uint64_t scn_max_txg;
|
|
uint64_t scn_cur_min_txg;
|
|
uint64_t scn_cur_max_txg;
|
|
uint64_t scn_start_time;
|
|
uint64_t scn_end_time;
|
|
uint64_t scn_to_examine; /* total bytes to be scanned */
|
|
uint64_t scn_examined; /* bytes scanned so far */
|
|
uint64_t scn_skipped; /* bytes skipped by scanner */
|
|
uint64_t scn_processed;
|
|
uint64_t scn_errors; /* scan I/O error count */
|
|
uint64_t scn_ddt_class_max;
|
|
ddt_bookmark_t scn_ddt_bookmark;
|
|
zbookmark_phys_t scn_bookmark;
|
|
uint64_t scn_flags; /* dsl_scan_flags_t */
|
|
} dsl_scan_phys_t;
|
|
|
|
#define SCAN_PHYS_NUMINTS (sizeof (dsl_scan_phys_t) / sizeof (uint64_t))
|
|
|
|
typedef enum dsl_scan_flags {
|
|
DSF_VISIT_DS_AGAIN = 1<<0,
|
|
DSF_SCRUB_PAUSED = 1<<1,
|
|
} dsl_scan_flags_t;
|
|
|
|
#define DSL_SCAN_FLAGS_MASK (DSF_VISIT_DS_AGAIN)
|
|
|
|
typedef struct dsl_errorscrub_phys {
|
|
uint64_t dep_func; /* pool_scan_func_t */
|
|
uint64_t dep_state; /* dsl_scan_state_t */
|
|
uint64_t dep_cursor; /* serialized zap cursor for tracing progress */
|
|
uint64_t dep_start_time; /* error scrub start time, unix timestamp */
|
|
uint64_t dep_end_time; /* error scrub end time, unix timestamp */
|
|
uint64_t dep_to_examine; /* total error blocks to be scrubbed */
|
|
uint64_t dep_examined; /* blocks scrubbed so far */
|
|
uint64_t dep_errors; /* error scrub I/O error count */
|
|
uint64_t dep_paused_flags; /* flag for paused */
|
|
} dsl_errorscrub_phys_t;
|
|
|
|
#define ERRORSCRUB_PHYS_NUMINTS (sizeof (dsl_errorscrub_phys_t) \
|
|
/ sizeof (uint64_t))
|
|
|
|
/*
|
|
* Every pool will have one dsl_scan_t and this structure will contain
|
|
* in-memory information about the scan and a pointer to the on-disk
|
|
* representation (i.e. dsl_scan_phys_t). Most of the state of the scan
|
|
* is contained on-disk to allow the scan to resume in the event of a reboot
|
|
* or panic. This structure maintains information about the behavior of a
|
|
* running scan, some caching information, and how it should traverse the pool.
|
|
*
|
|
* The following members of this structure direct the behavior of the scan:
|
|
*
|
|
* scn_suspending - a scan that cannot be completed in a single txg or
|
|
* has exceeded its allotted time will need to suspend.
|
|
* When this flag is set the scanner will stop traversing
|
|
* the pool and write out the current state to disk.
|
|
*
|
|
* scn_restart_txg - directs the scanner to either restart or start a
|
|
* a scan at the specified txg value.
|
|
*
|
|
* scn_done_txg - when a scan completes its traversal it will set
|
|
* the completion txg to the next txg. This is necessary
|
|
* to ensure that any blocks that were freed during
|
|
* the scan but have not yet been processed (i.e deferred
|
|
* frees) are accounted for.
|
|
*
|
|
* This structure also maintains information about deferred frees which are
|
|
* a special kind of traversal. Deferred free can exist in either a bptree or
|
|
* a bpobj structure. The scn_is_bptree flag will indicate the type of
|
|
* deferred free that is in progress. If the deferred free is part of an
|
|
* asynchronous destroy then the scn_async_destroying flag will be set.
|
|
*/
|
|
typedef struct dsl_scan {
|
|
struct dsl_pool *scn_dp;
|
|
uint64_t scn_restart_txg;
|
|
uint64_t scn_done_txg;
|
|
uint64_t scn_sync_start_time;
|
|
uint64_t scn_issued_before_pass;
|
|
|
|
/* for freeing blocks */
|
|
boolean_t scn_is_bptree;
|
|
boolean_t scn_async_destroying;
|
|
boolean_t scn_async_stalled;
|
|
uint64_t scn_async_block_min_time_ms;
|
|
|
|
/* flags and stats for controlling scan state */
|
|
boolean_t scn_is_sorted; /* doing sequential scan */
|
|
boolean_t scn_clearing; /* scan is issuing sequential extents */
|
|
boolean_t scn_checkpointing; /* scan is issuing all queued extents */
|
|
boolean_t scn_suspending; /* scan is suspending until next txg */
|
|
uint64_t scn_last_checkpoint; /* time of last checkpoint */
|
|
|
|
/* members for thread synchronization */
|
|
zio_t *scn_zio_root; /* root zio for waiting on IO */
|
|
taskq_t *scn_taskq; /* task queue for issuing extents */
|
|
|
|
/* for controlling scan prefetch, protected by spa_scrub_lock */
|
|
boolean_t scn_prefetch_stop; /* prefetch should stop */
|
|
zbookmark_phys_t scn_prefetch_bookmark; /* prefetch start bookmark */
|
|
avl_tree_t scn_prefetch_queue; /* priority queue of prefetch IOs */
|
|
uint64_t scn_maxinflight_bytes; /* max bytes in flight for pool */
|
|
|
|
/* per txg statistics */
|
|
uint64_t scn_visited_this_txg; /* total bps visited this txg */
|
|
uint64_t scn_dedup_frees_this_txg; /* dedup bps freed this txg */
|
|
uint64_t scn_holes_this_txg;
|
|
uint64_t scn_lt_min_this_txg;
|
|
uint64_t scn_gt_max_this_txg;
|
|
uint64_t scn_ddt_contained_this_txg;
|
|
uint64_t scn_objsets_visited_this_txg;
|
|
uint64_t scn_avg_seg_size_this_txg;
|
|
uint64_t scn_segs_this_txg;
|
|
uint64_t scn_avg_zio_size_this_txg;
|
|
uint64_t scn_zios_this_txg;
|
|
|
|
/* zap cursor for tracing error scrub progress */
|
|
zap_cursor_t errorscrub_cursor;
|
|
/* members needed for syncing scan status to disk */
|
|
dsl_scan_phys_t scn_phys; /* on disk representation of scan */
|
|
dsl_scan_phys_t scn_phys_cached;
|
|
avl_tree_t scn_queue; /* queue of datasets to scan */
|
|
kmutex_t scn_queue_lock; /* serializes scn_queue inserts */
|
|
uint64_t scn_queues_pending; /* outstanding data to issue */
|
|
/* members needed for syncing error scrub status to disk */
|
|
dsl_errorscrub_phys_t errorscrub_phys;
|
|
} dsl_scan_t;
|
|
|
|
typedef struct dsl_scan_io_queue dsl_scan_io_queue_t;
|
|
|
|
void scan_init(void);
|
|
void scan_fini(void);
|
|
int dsl_scan_init(struct dsl_pool *dp, uint64_t txg);
|
|
int dsl_scan_setup_check(void *, dmu_tx_t *);
|
|
void dsl_scan_setup_sync(void *, dmu_tx_t *);
|
|
void dsl_scan_fini(struct dsl_pool *dp);
|
|
void dsl_scan_sync(struct dsl_pool *, dmu_tx_t *);
|
|
int dsl_scan_cancel(struct dsl_pool *);
|
|
int dsl_scan(struct dsl_pool *, pool_scan_func_t);
|
|
void dsl_scan_assess_vdev(struct dsl_pool *dp, vdev_t *vd);
|
|
boolean_t dsl_scan_scrubbing(const struct dsl_pool *dp);
|
|
boolean_t dsl_errorscrubbing(const struct dsl_pool *dp);
|
|
boolean_t dsl_errorscrub_active(dsl_scan_t *scn);
|
|
void dsl_scan_restart_resilver(struct dsl_pool *, uint64_t txg);
|
|
int dsl_scrub_set_pause_resume(const struct dsl_pool *dp,
|
|
pool_scrub_cmd_t cmd);
|
|
void dsl_errorscrub_sync(struct dsl_pool *, dmu_tx_t *);
|
|
boolean_t dsl_scan_resilvering(struct dsl_pool *dp);
|
|
boolean_t dsl_scan_resilver_scheduled(struct dsl_pool *dp);
|
|
boolean_t dsl_dataset_unstable(struct dsl_dataset *ds);
|
|
void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum,
|
|
ddt_t *ddt, ddt_lightweight_entry_t *ddlwe, dmu_tx_t *tx);
|
|
void dsl_scan_ds_destroyed(struct dsl_dataset *ds, struct dmu_tx *tx);
|
|
void dsl_scan_ds_snapshotted(struct dsl_dataset *ds, struct dmu_tx *tx);
|
|
void dsl_scan_ds_clone_swapped(struct dsl_dataset *ds1, struct dsl_dataset *ds2,
|
|
struct dmu_tx *tx);
|
|
boolean_t dsl_scan_active(dsl_scan_t *scn);
|
|
boolean_t dsl_scan_is_paused_scrub(const dsl_scan_t *scn);
|
|
boolean_t dsl_errorscrub_is_paused(const dsl_scan_t *scn);
|
|
void dsl_scan_freed(spa_t *spa, const blkptr_t *bp);
|
|
void dsl_scan_io_queue_destroy(dsl_scan_io_queue_t *queue);
|
|
void dsl_scan_io_queue_vdev_xfer(vdev_t *svd, vdev_t *tvd);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|
|
|
|
#endif /* _SYS_DSL_SCAN_H */
|