mirror_zfs/include/sys/metaslab_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

414 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 2009 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
/*
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
*/
#ifndef _SYS_METASLAB_IMPL_H
#define _SYS_METASLAB_IMPL_H
#include <sys/metaslab.h>
#include <sys/space_map.h>
#include <sys/range_tree.h>
#include <sys/vdev.h>
#include <sys/txg.h>
#include <sys/avl.h>
#ifdef __cplusplus
extern "C" {
#endif
/*
* Metaslab allocation tracing record.
*/
typedef struct metaslab_alloc_trace {
list_node_t mat_list_node;
metaslab_group_t *mat_mg;
metaslab_t *mat_msp;
uint64_t mat_size;
uint64_t mat_weight;
uint32_t mat_dva_id;
uint64_t mat_offset;
int mat_allocator;
} metaslab_alloc_trace_t;
/*
* Used by the metaslab allocation tracing facility to indicate
* error conditions. These errors are stored to the offset member
* of the metaslab_alloc_trace_t record and displayed by mdb.
*/
typedef enum trace_alloc_type {
TRACE_ALLOC_FAILURE = -1ULL,
TRACE_TOO_SMALL = -2ULL,
TRACE_FORCE_GANG = -3ULL,
TRACE_NOT_ALLOCATABLE = -4ULL,
TRACE_GROUP_FAILURE = -5ULL,
TRACE_ENOSPC = -6ULL,
TRACE_CONDENSING = -7ULL,
TRACE_VDEV_ERROR = -8ULL
} trace_alloc_type_t;
#define METASLAB_WEIGHT_PRIMARY (1ULL << 63)
#define METASLAB_WEIGHT_SECONDARY (1ULL << 62)
#define METASLAB_WEIGHT_CLAIM (1ULL << 61)
#define METASLAB_WEIGHT_TYPE (1ULL << 60)
#define METASLAB_ACTIVE_MASK \
(METASLAB_WEIGHT_PRIMARY | METASLAB_WEIGHT_SECONDARY | \
METASLAB_WEIGHT_CLAIM)
/*
* The metaslab weight is used to encode the amount of free space in a
* metaslab, such that the "best" metaslab appears first when sorting the
* metaslabs by weight. The weight (and therefore the "best" metaslab) can
* be determined in two different ways: by computing a weighted sum of all
* the free space in the metaslab (a space based weight) or by counting only
* the free segments of the largest size (a segment based weight). We prefer
* the segment based weight because it reflects how the free space is
* comprised, but we cannot always use it -- legacy pools do not have the
* space map histogram information necessary to determine the largest
* contiguous regions. Pools that have the space map histogram determine
* the segment weight by looking at each bucket in the histogram and
* determining the free space whose size in bytes is in the range:
* [2^i, 2^(i+1))
* We then encode the largest index, i, that contains regions into the
* segment-weighted value.
*
* Space-based weight:
*
* 64 56 48 40 32 24 16 8 0
* +-------+-------+-------+-------+-------+-------+-------+-------+
* |PSC1| weighted-free space |
* +-------+-------+-------+-------+-------+-------+-------+-------+
*
* PS - indicates primary and secondary activation
* C - indicates activation for claimed block zio
* space - the fragmentation-weighted space
*
* Segment-based weight:
*
* 64 56 48 40 32 24 16 8 0
* +-------+-------+-------+-------+-------+-------+-------+-------+
* |PSC0| idx| count of segments in region |
* +-------+-------+-------+-------+-------+-------+-------+-------+
*
* PS - indicates primary and secondary activation
* C - indicates activation for claimed block zio
* idx - index for the highest bucket in the histogram
* count - number of segments in the specified bucket
*/
#define WEIGHT_GET_ACTIVE(weight) BF64_GET((weight), 61, 3)
#define WEIGHT_SET_ACTIVE(weight, x) BF64_SET((weight), 61, 3, x)
#define WEIGHT_IS_SPACEBASED(weight) \
((weight) == 0 || BF64_GET((weight), 60, 1))
#define WEIGHT_SET_SPACEBASED(weight) BF64_SET((weight), 60, 1, 1)
/*
* These macros are only applicable to segment-based weighting.
*/
#define WEIGHT_GET_INDEX(weight) BF64_GET((weight), 54, 6)
#define WEIGHT_SET_INDEX(weight, x) BF64_SET((weight), 54, 6, x)
#define WEIGHT_GET_COUNT(weight) BF64_GET((weight), 0, 54)
#define WEIGHT_SET_COUNT(weight, x) BF64_SET((weight), 0, 54, x)
/*
* A metaslab class encompasses a category of allocatable top-level vdevs.
* Each top-level vdev is associated with a metaslab group which defines
* the allocatable region for that vdev. Examples of these categories include
* "normal" for data block allocations (i.e. main pool allocations) or "log"
* for allocations designated for intent log devices (i.e. slog devices).
* When a block allocation is requested from the SPA it is associated with a
* metaslab_class_t, and only top-level vdevs (i.e. metaslab groups) belonging
* to the class can be used to satisfy that request. Allocations are done
* by traversing the metaslab groups that are linked off of the mc_rotor field.
* This rotor points to the next metaslab group where allocations will be
* attempted. Allocating a block is a 3 step process -- select the metaslab
* group, select the metaslab, and then allocate the block. The metaslab
* class defines the low-level block allocator that will be used as the
* final step in allocation. These allocators are pluggable allowing each class
* to use a block allocator that best suits that class.
*/
struct metaslab_class {
kmutex_t mc_lock;
spa_t *mc_spa;
metaslab_group_t *mc_rotor;
metaslab_ops_t *mc_ops;
uint64_t mc_aliquot;
/*
* Track the number of metaslab groups that have been initialized
* and can accept allocations. An initialized metaslab group is
* one has been completely added to the config (i.e. we have
* updated the MOS config and the space has been added to the pool).
*/
uint64_t mc_groups;
/*
* Toggle to enable/disable the allocation throttle.
*/
boolean_t mc_alloc_throttle_enabled;
/*
* The allocation throttle works on a reservation system. Whenever
* an asynchronous zio wants to perform an allocation it must
* first reserve the number of blocks that it wants to allocate.
* If there aren't sufficient slots available for the pending zio
* then that I/O is throttled until more slots free up. The current
* number of reserved allocations is maintained by the mc_alloc_slots
* refcount. The mc_alloc_max_slots value determines the maximum
* number of allocations that the system allows. Gang blocks are
* allowed to reserve slots even if we've reached the maximum
* number of allocations allowed.
*/
uint64_t *mc_alloc_max_slots;
refcount_t *mc_alloc_slots;
uint64_t mc_alloc_groups; /* # of allocatable groups */
uint64_t mc_alloc; /* total allocated space */
uint64_t mc_deferred; /* total deferred frees */
uint64_t mc_space; /* total space (alloc + free) */
uint64_t mc_dspace; /* total deflated space */
uint64_t mc_histogram[RANGE_TREE_HISTOGRAM_SIZE];
};
/*
* Metaslab groups encapsulate all the allocatable regions (i.e. metaslabs)
* of a top-level vdev. They are linked together to form a circular linked
* list and can belong to only one metaslab class. Metaslab groups may become
* ineligible for allocations for a number of reasons such as limited free
* space, fragmentation, or going offline. When this happens the allocator will
* simply find the next metaslab group in the linked list and attempt
* to allocate from that group instead.
*/
struct metaslab_group {
kmutex_t mg_lock;
metaslab_t **mg_primaries;
metaslab_t **mg_secondaries;
avl_tree_t mg_metaslab_tree;
uint64_t mg_aliquot;
boolean_t mg_allocatable; /* can we allocate? */
uint64_t mg_ms_ready;
/*
* A metaslab group is considered to be initialized only after
* we have updated the MOS config and added the space to the pool.
* We only allow allocation attempts to a metaslab group if it
* has been initialized.
*/
boolean_t mg_initialized;
uint64_t mg_free_capacity; /* percentage free */
int64_t mg_bias;
int64_t mg_activation_count;
metaslab_class_t *mg_class;
vdev_t *mg_vd;
taskq_t *mg_taskq;
metaslab_group_t *mg_prev;
metaslab_group_t *mg_next;
/*
* In order for the allocation throttle to function properly, we cannot
* have too many IOs going to each disk by default; the throttle
* operates by allocating more work to disks that finish quickly, so
* allocating larger chunks to each disk reduces its effectiveness.
* However, if the number of IOs going to each allocator is too small,
* we will not perform proper aggregation at the vdev_queue layer,
* also resulting in decreased performance. Therefore, we will use a
* ramp-up strategy.
*
* Each allocator in each metaslab group has a current queue depth
* (mg_alloc_queue_depth[allocator]) and a current max queue depth
* (mg_cur_max_alloc_queue_depth[allocator]), and each metaslab group
* has an absolute max queue depth (mg_max_alloc_queue_depth). We
* add IOs to an allocator until the mg_alloc_queue_depth for that
* allocator hits the cur_max. Every time an IO completes for a given
* allocator on a given metaslab group, we increment its cur_max until
* it reaches mg_max_alloc_queue_depth. The cur_max resets every txg to
* help protect against disks that decrease in performance over time.
*
* It's possible for an allocator to handle more allocations than
* its max. This can occur when gang blocks are required or when other
* groups are unable to handle their share of allocations.
*/
uint64_t mg_max_alloc_queue_depth;
uint64_t *mg_cur_max_alloc_queue_depth;
refcount_t *mg_alloc_queue_depth;
int mg_allocators;
/*
* A metalab group that can no longer allocate the minimum block
* size will set mg_no_free_space. Once a metaslab group is out
* of space then its share of work must be distributed to other
* groups.
*/
boolean_t mg_no_free_space;
uint64_t mg_allocations;
uint64_t mg_failed_allocations;
uint64_t mg_fragmentation;
uint64_t mg_histogram[RANGE_TREE_HISTOGRAM_SIZE];
};
/*
* This value defines the number of elements in the ms_lbas array. The value
* of 64 was chosen as it covers all power of 2 buckets up to UINT64_MAX.
* This is the equivalent of highbit(UINT64_MAX).
*/
#define MAX_LBAS 64
/*
* Each metaslab maintains a set of in-core trees to track metaslab
* operations. The in-core free tree (ms_allocatable) contains the list of
* free segments which are eligible for allocation. As blocks are
* allocated, the allocated segment are removed from the ms_allocatable and
* added to a per txg allocation tree (ms_allocating). As blocks are
* freed, they are added to the free tree (ms_freeing). These trees
* allow us to process all allocations and frees in syncing context
* where it is safe to update the on-disk space maps. An additional set
* of in-core trees is maintained to track deferred frees
* (ms_defer). Once a block is freed it will move from the
* ms_freed to the ms_defer tree. A deferred free means that a block
* has been freed but cannot be used by the pool until TXG_DEFER_SIZE
* transactions groups later. For example, a block that is freed in txg
* 50 will not be available for reallocation until txg 52 (50 +
* TXG_DEFER_SIZE). This provides a safety net for uberblock rollback.
* A pool could be safely rolled back TXG_DEFERS_SIZE transactions
* groups and ensure that no block has been reallocated.
*
* The simplified transition diagram looks like this:
*
*
* ALLOCATE
* |
* V
* free segment (ms_allocatable) -> ms_allocating[4] -> (write to space map)
* ^
* | ms_freeing <--- FREE
* | |
* | v
* | ms_freed
* | |
* +-------- ms_defer[2] <-------+-------> (write to space map)
*
*
* Each metaslab's space is tracked in a single space map in the MOS,
* which is only updated in syncing context. Each time we sync a txg,
* we append the allocs and frees from that txg to the space map. The
* pool space is only updated once all metaslabs have finished syncing.
*
* To load the in-core free tree we read the space map from disk. This
* object contains a series of alloc and free records that are combined
* to make up the list of all free segments in this metaslab. These
* segments are represented in-core by the ms_allocatable and are stored
* in an AVL tree.
*
* As the space map grows (as a result of the appends) it will
* eventually become space-inefficient. When the metaslab's in-core
* free tree is zfs_condense_pct/100 times the size of the minimal
* on-disk representation, we rewrite it in its minimized form. If a
* metaslab needs to condense then we must set the ms_condensing flag to
* ensure that allocations are not performed on the metaslab that is
* being written.
*/
struct metaslab {
kmutex_t ms_lock;
kmutex_t ms_sync_lock;
kcondvar_t ms_load_cv;
space_map_t *ms_sm;
uint64_t ms_id;
uint64_t ms_start;
uint64_t ms_size;
uint64_t ms_fragmentation;
range_tree_t *ms_allocating[TXG_SIZE];
range_tree_t *ms_allocatable;
/*
* The following range trees are accessed only from syncing context.
* ms_free*tree only have entries while syncing, and are empty
* between syncs.
*/
range_tree_t *ms_freeing; /* to free this syncing txg */
range_tree_t *ms_freed; /* already freed this syncing txg */
range_tree_t *ms_defer[TXG_DEFER_SIZE];
range_tree_t *ms_checkpointing; /* to add to the checkpoint */
boolean_t ms_condensing; /* condensing? */
boolean_t ms_condense_wanted;
uint64_t ms_condense_checked_txg;
/*
* We must hold both ms_lock and ms_group->mg_lock in order to
* modify ms_loaded.
*/
boolean_t ms_loaded;
boolean_t ms_loading;
int64_t ms_deferspace; /* sum of ms_defermap[] space */
uint64_t ms_weight; /* weight vs. others in group */
uint64_t ms_activation_weight; /* activation weight */
/*
* Track of whenever a metaslab is selected for loading or allocation.
* We use this value to determine how long the metaslab should
* stay cached.
*/
uint64_t ms_selected_txg;
uint64_t ms_alloc_txg; /* last successful alloc (debug only) */
uint64_t ms_max_size; /* maximum allocatable size */
/*
* -1 if it's not active in an allocator, otherwise set to the allocator
* this metaslab is active for.
*/
int ms_allocator;
boolean_t ms_primary; /* Only valid if ms_allocator is not -1 */
/*
* The metaslab block allocators can optionally use a size-ordered
* range tree and/or an array of LBAs. Not all allocators use
* this functionality. The ms_allocatable_by_size should always
* contain the same number of segments as the ms_allocatable. The
* only difference is that the ms_allocatable_by_size is ordered by
* segment sizes.
*/
avl_tree_t ms_allocatable_by_size;
uint64_t ms_lbas[MAX_LBAS];
metaslab_group_t *ms_group; /* metaslab group */
avl_node_t ms_group_node; /* node in metaslab group tree */
txg_node_t ms_txg_node; /* per-txg dirty metaslab links */
boolean_t ms_new;
};
#ifdef __cplusplus
}
#endif
#endif /* _SYS_METASLAB_IMPL_H */