mirror_zfs/include/sys/spa_impl.h
Paul Dagnelie 492f64e941 OpenZFS 9112 - Improve allocation performance on high-end systems
Overview
========

We parallelize the allocation process by creating the concept of
"allocators". There are a certain number of allocators per metaslab
group, defined by the value of a tunable at pool open time.  Each
allocator for a given metaslab group has up to 2 active metaslabs; one
"primary", and one "secondary". The primary and secondary weight mean
the same thing they did in in the pre-allocator world; primary metaslabs
are used for most allocations, secondary metaslabs are used for ditto
blocks being allocated in the same metaslab group.  There is also the
CLAIM weight, which has been separated out from the other weights, but
that is less important to understanding the patch.  The active metaslabs
for each allocator are moved from their normal place in the metaslab
tree for the group to the back of the tree. This way, they will not be
selected for use by other allocators searching for new metaslabs unless
all the passive metaslabs are unsuitable for allocations.  If that does
happen, the allocators will "steal" from each other to ensure that IOs
don't fail until there is truly no space left to perform allocations.

In addition, the alloc queue for each metaslab group has been broken
into a separate queue for each allocator. We don't want to dramatically
increase the number of inflight IOs on low-end systems, because it can
significantly increase txg times. On the other hand, we want to ensure
that there are enough IOs for each allocator to allow for good
coalescing before sending the IOs to the disk.  As a result, we take a
compromise path; each allocator's alloc queue max depth starts at a
certain value for every txg. Every time an IO completes, we increase the
max depth. This should hopefully provide a good balance between the two
failure modes, while not dramatically increasing complexity.

We also parallelize the spa_alloc_tree and spa_alloc_lock, which cause
very similar contention when selecting IOs to allocate. This
parallelization uses the same allocator scheme as metaslab selection.

Performance Results
===================

Performance improvements from this change can vary significantly based
on the number of CPUs in the system, whether or not the system has a
NUMA architecture, the speed of the drives, the values for the various
tunables, and the workload being performed. For an fio async sequential
write workload on a 24 core NUMA system with 256 GB of RAM and 8 128 GB
SSDs, there is a roughly 25% performance improvement.

Future Work
===========

Analysis of the performance of the system with this patch applied shows
that a significant new bottleneck is the vdev disk queues, which also
need to be parallelized.  Prototyping of this change has occurred, and
there was a performance improvement, but more work needs to be done
before its stability has been verified and it is ready to be upstreamed.

Authored by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
Reviewed by: Alexander Motin <mav@FreeBSD.org>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Gordon Ross <gwr@nexenta.com>
Ported-by: Paul Dagnelie <pcd@delphix.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>

Porting Notes:
* Fix reservation test failures by increasing tolerance.

OpenZFS-issue: https://illumos.org/issues/9112
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/3f3cc3c3
Closes #7682
2018-07-31 10:52:33 -07:00

415 lines
16 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 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) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2017 by Delphix. All rights reserved.
* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright 2013 Saso Kiselkov. All rights reserved.
* Copyright (c) 2016 Actifio, Inc. All rights reserved.
* Copyright (c) 2017 Datto Inc.
*/
#ifndef _SYS_SPA_IMPL_H
#define _SYS_SPA_IMPL_H
#include <sys/spa.h>
#include <sys/spa_checkpoint.h>
#include <sys/vdev.h>
#include <sys/vdev_removal.h>
#include <sys/metaslab.h>
#include <sys/dmu.h>
#include <sys/dsl_pool.h>
#include <sys/uberblock_impl.h>
#include <sys/zfs_context.h>
#include <sys/avl.h>
#include <sys/refcount.h>
#include <sys/bplist.h>
#include <sys/bpobj.h>
#include <sys/dsl_crypt.h>
#include <sys/zfeature.h>
#include <sys/zthr.h>
#include <zfeature_common.h>
#ifdef __cplusplus
extern "C" {
#endif
typedef struct spa_error_entry {
zbookmark_phys_t se_bookmark;
char *se_name;
avl_node_t se_avl;
} spa_error_entry_t;
typedef struct spa_history_phys {
uint64_t sh_pool_create_len; /* ending offset of zpool create */
uint64_t sh_phys_max_off; /* physical EOF */
uint64_t sh_bof; /* logical BOF */
uint64_t sh_eof; /* logical EOF */
uint64_t sh_records_lost; /* num of records overwritten */
} spa_history_phys_t;
/*
* All members must be uint64_t, for byteswap purposes.
*/
typedef struct spa_removing_phys {
uint64_t sr_state; /* dsl_scan_state_t */
/*
* The vdev ID that we most recently attempted to remove,
* or -1 if no removal has been attempted.
*/
uint64_t sr_removing_vdev;
/*
* The vdev ID that we most recently successfully removed,
* or -1 if no devices have been removed.
*/
uint64_t sr_prev_indirect_vdev;
uint64_t sr_start_time;
uint64_t sr_end_time;
/*
* Note that we can not use the space map's or indirect mapping's
* accounting as a substitute for these values, because we need to
* count frees of not-yet-copied data as though it did the copy.
* Otherwise, we could get into a situation where copied > to_copy,
* or we complete before copied == to_copy.
*/
uint64_t sr_to_copy; /* bytes that need to be copied */
uint64_t sr_copied; /* bytes that have been copied or freed */
} spa_removing_phys_t;
/*
* This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
* (with key DMU_POOL_CONDENSING_INDIRECT). It is present if a condense
* of an indirect vdev's mapping object is in progress.
*/
typedef struct spa_condensing_indirect_phys {
/*
* The vdev ID of the indirect vdev whose indirect mapping is
* being condensed.
*/
uint64_t scip_vdev;
/*
* The vdev's old obsolete spacemap. This spacemap's contents are
* being integrated into the new mapping.
*/
uint64_t scip_prev_obsolete_sm_object;
/*
* The new mapping object that is being created.
*/
uint64_t scip_next_mapping_object;
} spa_condensing_indirect_phys_t;
struct spa_aux_vdev {
uint64_t sav_object; /* MOS object for device list */
nvlist_t *sav_config; /* cached device config */
vdev_t **sav_vdevs; /* devices */
int sav_count; /* number devices */
boolean_t sav_sync; /* sync the device list */
nvlist_t **sav_pending; /* pending device additions */
uint_t sav_npending; /* # pending devices */
};
typedef struct spa_config_lock {
kmutex_t scl_lock;
kthread_t *scl_writer;
int scl_write_wanted;
kcondvar_t scl_cv;
refcount_t scl_count;
} spa_config_lock_t;
typedef struct spa_config_dirent {
list_node_t scd_link;
char *scd_path;
} spa_config_dirent_t;
typedef enum zio_taskq_type {
ZIO_TASKQ_ISSUE = 0,
ZIO_TASKQ_ISSUE_HIGH,
ZIO_TASKQ_INTERRUPT,
ZIO_TASKQ_INTERRUPT_HIGH,
ZIO_TASKQ_TYPES
} zio_taskq_type_t;
/*
* State machine for the zpool-poolname process. The states transitions
* are done as follows:
*
* From To Routine
* PROC_NONE -> PROC_CREATED spa_activate()
* PROC_CREATED -> PROC_ACTIVE spa_thread()
* PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate()
* PROC_DEACTIVATE -> PROC_GONE spa_thread()
* PROC_GONE -> PROC_NONE spa_deactivate()
*/
typedef enum spa_proc_state {
SPA_PROC_NONE, /* spa_proc = &p0, no process created */
SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */
SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */
SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */
SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */
} spa_proc_state_t;
typedef struct spa_taskqs {
uint_t stqs_count;
taskq_t **stqs_taskq;
} spa_taskqs_t;
typedef enum spa_all_vdev_zap_action {
AVZ_ACTION_NONE = 0,
AVZ_ACTION_DESTROY, /* Destroy all per-vdev ZAPs and the AVZ. */
AVZ_ACTION_REBUILD, /* Populate the new AVZ, see spa_avz_rebuild */
AVZ_ACTION_INITIALIZE
} spa_avz_action_t;
typedef enum spa_config_source {
SPA_CONFIG_SRC_NONE = 0,
SPA_CONFIG_SRC_SCAN, /* scan of path (default: /dev/dsk) */
SPA_CONFIG_SRC_CACHEFILE, /* any cachefile */
SPA_CONFIG_SRC_TRYIMPORT, /* returned from call to tryimport */
SPA_CONFIG_SRC_SPLIT, /* new pool in a pool split */
SPA_CONFIG_SRC_MOS /* MOS, but not always from right txg */
} spa_config_source_t;
struct spa {
/*
* Fields protected by spa_namespace_lock.
*/
char spa_name[ZFS_MAX_DATASET_NAME_LEN]; /* pool name */
char *spa_comment; /* comment */
avl_node_t spa_avl; /* node in spa_namespace_avl */
nvlist_t *spa_config; /* last synced config */
nvlist_t *spa_config_syncing; /* currently syncing config */
nvlist_t *spa_config_splitting; /* config for splitting */
nvlist_t *spa_load_info; /* info and errors from load */
uint64_t spa_config_txg; /* txg of last config change */
int spa_sync_pass; /* iterate-to-convergence */
pool_state_t spa_state; /* pool state */
int spa_inject_ref; /* injection references */
uint8_t spa_sync_on; /* sync threads are running */
spa_load_state_t spa_load_state; /* current load operation */
boolean_t spa_indirect_vdevs_loaded; /* mappings loaded? */
boolean_t spa_trust_config; /* do we trust vdev tree? */
spa_config_source_t spa_config_source; /* where config comes from? */
uint64_t spa_import_flags; /* import specific flags */
spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
dsl_pool_t *spa_dsl_pool;
boolean_t spa_is_initializing; /* true while opening pool */
metaslab_class_t *spa_normal_class; /* normal data class */
metaslab_class_t *spa_log_class; /* intent log data class */
uint64_t spa_first_txg; /* first txg after spa_open() */
uint64_t spa_final_txg; /* txg of export/destroy */
uint64_t spa_freeze_txg; /* freeze pool at this txg */
uint64_t spa_load_max_txg; /* best initial ub_txg */
uint64_t spa_claim_max_txg; /* highest claimed birth txg */
inode_timespec_t spa_loaded_ts; /* 1st successful open time */
objset_t *spa_meta_objset; /* copy of dp->dp_meta_objset */
kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */
list_t spa_evicting_os_list; /* Objsets being evicted. */
kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */
txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */
vdev_t *spa_root_vdev; /* top-level vdev container */
int spa_min_ashift; /* of vdevs in normal class */
int spa_max_ashift; /* of vdevs in normal class */
uint64_t spa_config_guid; /* config pool guid */
uint64_t spa_load_guid; /* spa_load initialized guid */
uint64_t spa_last_synced_guid; /* last synced guid */
list_t spa_config_dirty_list; /* vdevs with dirty config */
list_t spa_state_dirty_list; /* vdevs with dirty state */
/*
* spa_alloc_locks and spa_alloc_trees are arrays, whose lengths are
* stored in spa_alloc_count. There is one tree and one lock for each
* allocator, to help improve allocation performance in write-heavy
* workloads.
*/
kmutex_t *spa_alloc_locks;
avl_tree_t *spa_alloc_trees;
int spa_alloc_count;
spa_aux_vdev_t spa_spares; /* hot spares */
spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */
nvlist_t *spa_label_features; /* Features for reading MOS */
uint64_t spa_config_object; /* MOS object for pool config */
uint64_t spa_config_generation; /* config generation number */
uint64_t spa_syncing_txg; /* txg currently syncing */
bpobj_t spa_deferred_bpobj; /* deferred-free bplist */
bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */
/* checksum context templates */
kmutex_t spa_cksum_tmpls_lock;
void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
uberblock_t spa_ubsync; /* last synced uberblock */
uberblock_t spa_uberblock; /* current uberblock */
boolean_t spa_extreme_rewind; /* rewind past deferred frees */
kmutex_t spa_scrub_lock; /* resilver/scrub lock */
uint64_t spa_scrub_inflight; /* in-flight scrub bytes */
uint64_t spa_load_verify_ios; /* in-flight verification IOs */
kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */
uint8_t spa_scrub_active; /* active or suspended? */
uint8_t spa_scrub_type; /* type of scrub we're doing */
uint8_t spa_scrub_finished; /* indicator to rotate logs */
uint8_t spa_scrub_started; /* started since last boot */
uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */
uint64_t spa_scan_pass_start; /* start time per pass/reboot */
uint64_t spa_scan_pass_scrub_pause; /* scrub pause time */
uint64_t spa_scan_pass_scrub_spent_paused; /* total paused */
uint64_t spa_scan_pass_exam; /* examined bytes per pass */
uint64_t spa_scan_pass_issued; /* issued bytes per pass */
kmutex_t spa_async_lock; /* protect async state */
kthread_t *spa_async_thread; /* thread doing async task */
int spa_async_suspended; /* async tasks suspended */
kcondvar_t spa_async_cv; /* wait for thread_exit() */
uint16_t spa_async_tasks; /* async task mask */
uint64_t spa_missing_tvds; /* unopenable tvds on load */
uint64_t spa_missing_tvds_allowed; /* allow loading spa? */
spa_removing_phys_t spa_removing_phys;
spa_vdev_removal_t *spa_vdev_removal;
spa_condensing_indirect_phys_t spa_condensing_indirect_phys;
spa_condensing_indirect_t *spa_condensing_indirect;
zthr_t *spa_condense_zthr; /* zthr doing condense. */
uint64_t spa_checkpoint_txg; /* the txg of the checkpoint */
spa_checkpoint_info_t spa_checkpoint_info; /* checkpoint accounting */
zthr_t *spa_checkpoint_discard_zthr;
char *spa_root; /* alternate root directory */
uint64_t spa_ena; /* spa-wide ereport ENA */
int spa_last_open_failed; /* error if last open failed */
uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */
uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */
uint64_t spa_load_txg; /* ub txg that loaded */
uint64_t spa_load_txg_ts; /* timestamp from that ub */
uint64_t spa_load_meta_errors; /* verify metadata err count */
uint64_t spa_load_data_errors; /* verify data err count */
uint64_t spa_verify_min_txg; /* start txg of verify scrub */
kmutex_t spa_errlog_lock; /* error log lock */
uint64_t spa_errlog_last; /* last error log object */
uint64_t spa_errlog_scrub; /* scrub error log object */
kmutex_t spa_errlist_lock; /* error list/ereport lock */
avl_tree_t spa_errlist_last; /* last error list */
avl_tree_t spa_errlist_scrub; /* scrub error list */
uint64_t spa_deflate; /* should we deflate? */
uint64_t spa_history; /* history object */
kmutex_t spa_history_lock; /* history lock */
vdev_t *spa_pending_vdev; /* pending vdev additions */
kmutex_t spa_props_lock; /* property lock */
uint64_t spa_pool_props_object; /* object for properties */
uint64_t spa_bootfs; /* default boot filesystem */
uint64_t spa_failmode; /* failure mode for the pool */
uint64_t spa_deadman_failmode; /* failure mode for deadman */
uint64_t spa_delegation; /* delegation on/off */
list_t spa_config_list; /* previous cache file(s) */
/* per-CPU array of root of async I/O: */
zio_t **spa_async_zio_root;
zio_t *spa_suspend_zio_root; /* root of all suspended I/O */
zio_t *spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */
kmutex_t spa_suspend_lock; /* protects suspend_zio_root */
kcondvar_t spa_suspend_cv; /* notification of resume */
zio_suspend_reason_t spa_suspended; /* pool is suspended */
uint8_t spa_claiming; /* pool is doing zil_claim() */
boolean_t spa_is_root; /* pool is root */
int spa_minref; /* num refs when first opened */
int spa_mode; /* FREAD | FWRITE */
spa_log_state_t spa_log_state; /* log state */
uint64_t spa_autoexpand; /* lun expansion on/off */
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 */
kmutex_t spa_proc_lock; /* protects spa_proc* */
kcondvar_t spa_proc_cv; /* spa_proc_state transitions */
spa_proc_state_t spa_proc_state; /* see definition */
proc_t *spa_proc; /* "zpool-poolname" process */
uint64_t spa_did; /* if procp != p0, did of t1 */
boolean_t spa_autoreplace; /* autoreplace set in open */
int spa_vdev_locks; /* locks grabbed */
uint64_t spa_creation_version; /* version at pool creation */
uint64_t spa_prev_software_version; /* See ub_software_version */
uint64_t spa_feat_for_write_obj; /* required to write to pool */
uint64_t spa_feat_for_read_obj; /* required to read from pool */
uint64_t spa_feat_desc_obj; /* Feature descriptions */
uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */
kmutex_t spa_feat_stats_lock; /* protects spa_feat_stats */
nvlist_t *spa_feat_stats; /* Cache of enabled features */
/* cache feature refcounts */
uint64_t spa_feat_refcount_cache[SPA_FEATURES];
taskqid_t spa_deadman_tqid; /* Task id */
uint64_t spa_deadman_calls; /* number of deadman calls */
hrtime_t spa_sync_starttime; /* starting time of spa_sync */
uint64_t spa_deadman_synctime; /* deadman sync expiration */
uint64_t spa_deadman_ziotime; /* deadman zio expiration */
uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */
spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */
uint64_t spa_errata; /* errata issues detected */
spa_stats_t spa_stats; /* assorted spa statistics */
spa_keystore_t spa_keystore; /* loaded crypto keys */
/* arc_memory_throttle() parameters during low memory condition */
uint64_t spa_lowmem_page_load; /* memory load during txg */
uint64_t spa_lowmem_last_txg; /* txg window start */
hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */
taskq_t *spa_zvol_taskq; /* Taskq for minor management */
taskq_t *spa_prefetch_taskq; /* Taskq for prefetch threads */
uint64_t spa_multihost; /* multihost aware (mmp) */
mmp_thread_t spa_mmp; /* multihost mmp thread */
/*
* spa_refcount & spa_config_lock must be the last elements
* because refcount_t changes size based on compilation options.
* In order for the MDB module to function correctly, the other
* fields must remain in the same location.
*/
spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
refcount_t spa_refcount; /* number of opens */
taskq_t *spa_upgrade_taskq; /* taskq for upgrade jobs */
};
extern char *spa_config_path;
extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent);
extern void spa_taskq_dispatch_sync(spa_t *, zio_type_t t, zio_taskq_type_t q,
task_func_t *func, void *arg, uint_t flags);
extern void spa_load_spares(spa_t *spa);
extern void spa_load_l2cache(spa_t *spa);
extern sysevent_t *spa_event_create(spa_t *spa, vdev_t *vd, nvlist_t *hist_nvl,
const char *name);
extern void spa_event_post(sysevent_t *ev);
#ifdef __cplusplus
}
#endif
#endif /* _SYS_SPA_IMPL_H */