"
The main goal of this series is to improve efficiency of reads from large partitions by
reducing amount of I/O needed to read the sstable index. This is achieved by caching
index file pages and partition index entries in memory.
Currently, the pages are cached by individual reads only for the duration of the read.
This was done to facilitate binary search in the promoted index (intra-partition index).
After this series, all reads share the index file page cache, which stays around even after reads stop.
The page cache is subject to eviction. It uses the same region as the current row cache and shares
the LRU with row cache entries. This means that LRU objects need to be virtualized. This series takes
an easy approach and does this by introducing a virtual base class. This adds an overhead to row cache
entry to store the vtable pointer.
SStable indexes have a hierarchy. There is a summary, which is a sparse partition key index into the
full partition index. This one is already kept in memory. The partition index is divided by the summary
into pages. Each entry in the partition index contains promoted index, which is a sparse index into atoms
identified by the clustering key (rows, tombstones).
In order to read the promoted index, the reader needs to read the partition index entry first.
To speed this up, this series also adds caching of partition index entries. This cache survives
reads and is subject to eviction, just like the index file page cache. The unit of caching is
the partition index page. Without this cache, each access to promoted index would have to be
preceded with the parsing of the partition index page containing the partition key.
Performance testing results follow.
1) scylla-bench large partition reads
Populated with:
perf_fast_forward --run-tests=large-partition-skips --datasets=sb-large-part-ds1 \
-c1 -m1G --populate --value-size=1024 --rows=10000000
Single partition, 9G data file, 4MB index file
Test execution:
build/release/scylla -c1 -m4G
scylla-bench -workload uniform -mode read -limit 1 -concurrency 100 -partition-count 1 \
-clustering-row-count 10000000 -duration 60m
TL;DR: after: 2x throughput, 0.5 median latency
Before (c1daf2bb24):
Results
Time (avg): 5m21.033180213s
Total ops: 966951
Total rows: 966951
Operations/s: 3011.997048812112
Rows/s: 3011.997048812112
Latency:
max: 74.055679ms
99.9th: 63.569919ms
99th: 41.320447ms
95th: 38.076415ms
90th: 37.158911ms
median: 34.537471ms
mean: 33.195994ms
After:
Results
Time (avg): 5m14.706669345s
Total ops: 2042831
Total rows: 2042831
Operations/s: 6491.22243800942
Rows/s: 6491.22243800942
Latency:
max: 60.096511ms
99.9th: 35.520511ms
99th: 27.000831ms
95th: 23.986175ms
90th: 21.659647ms
median: 15.040511ms
mean: 15.402076ms
2) scylla-bench small partitions
I tested several scenarios with a varying data set size, e.g. data fully fitting in memory,
half fitting, and being much larger. The improvement varied a bit but in all cases the "after"
code performed slightly better.
Below is a representative run over data set which does not fit in memory.
scylla -c1 -m4G
scylla-bench -workload uniform -mode read -concurrency 400 -partition-count 10000000 \
-clustering-row-count 1 -duration 60m -no-lower-bound
Before:
Time (avg): 51.072411913s
Total ops: 3165885
Total rows: 3165885
Operations/s: 61988.164024260645
Rows/s: 61988.164024260645
Latency:
max: 34.045951ms
99.9th: 25.985023ms
99th: 23.298047ms
95th: 19.070975ms
90th: 17.530879ms
median: 3.899391ms
mean: 6.450616ms
After:
Time (avg): 50.232410679s
Total ops: 3778863
Total rows: 3778863
Operations/s: 75227.58014424688
Rows/s: 75227.58014424688
Latency:
max: 37.027839ms
99.9th: 24.805375ms
99th: 18.219007ms
95th: 14.090239ms
90th: 12.124159ms
median: 4.030463ms
mean: 5.315111ms
The results include the warmup phase which populates the partition index cache, so the hot-cache effect
is dampened in the statistics. See the 99th percentile. Latency gets better after the cache warms up which
moves it lower.
3) perf_fast_forward --run-tests=large-partition-skips
Caching is not used here, included to show there are no regressions for the cold cache case.
TL;DR: No significant change
perf_fast_forward --run-tests=large-partition-skips --datasets=large-part-ds1 -c1 -m1G
Config: rows: 10000000, value size: 2000
Before:
read skip time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk cpu
1 0 36.429822 4 10000000 274500 62 274521 274429 153889.2 153883 19696986 153853 0 0 0 0 0 0 0 22.5%
1 1 36.856236 4 5000000 135662 7 135670 135650 155652.0 155652 19704117 139326 1 0 1 1 0 0 0 38.1%
1 8 36.347667 4 1111112 30569 0 30570 30569 155652.0 155652 19704117 139071 1 0 1 1 0 0 0 19.5%
1 16 36.278866 4 588236 16214 1 16215 16213 155652.0 155652 19704117 139073 1 0 1 1 0 0 0 16.6%
1 32 36.174784 4 303031 8377 0 8377 8376 155652.0 155652 19704117 139056 1 0 1 1 0 0 0 12.3%
1 64 36.147104 4 153847 4256 0 4256 4256 155652.0 155652 19704117 139109 1 0 1 1 0 0 0 11.1%
1 256 9.895288 4 38911 3932 1 3933 3930 100869.2 100868 3178298 59944 38912 0 1 1 0 0 0 14.3%
1 1024 2.599921 4 9757 3753 0 3753 3753 26604.0 26604 801850 15071 9758 0 1 1 0 0 0 14.6%
1 4096 0.784568 4 2441 3111 1 3111 3109 7982.0 7982 205946 3772 2442 0 1 1 0 0 0 13.8%
64 1 36.553975 4 9846154 269359 10 269369 269337 155663.8 155652 19704117 139230 1 0 1 1 0 0 0 28.2%
64 8 36.509694 4 8888896 243467 8 243475 243449 155652.0 155652 19704117 139120 1 0 1 1 0 0 0 26.5%
64 16 36.466282 4 8000000 219381 4 219385 219374 155652.0 155652 19704117 139232 1 0 1 1 0 0 0 24.8%
64 32 36.395926 4 6666688 183171 6 183180 183165 155652.0 155652 19704117 139158 1 0 1 1 0 0 0 21.8%
64 64 36.296856 4 5000000 137753 4 137757 137737 155652.0 155652 19704117 139105 1 0 1 1 0 0 0 17.7%
64 256 20.590392 4 2000000 97133 18 97151 94996 135248.8 131395 7877402 98335 31282 0 1 1 0 0 0 15.7%
64 1024 6.225773 4 588288 94492 1436 95434 88748 46066.5 41321 2324378 30360 9193 0 1 1 0 0 0 15.8%
64 4096 1.856069 4 153856 82893 54 82948 82721 16115.0 16043 583674 11574 2675 0 1 1 0 0 0 16.3%
After:
read skip time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk cpu
1 0 36.429240 4 10000000 274505 38 274515 274417 153887.8 153883 19696986 153849 0 0 0 0 0 0 0 22.4%
1 1 36.933806 4 5000000 135377 15 135385 135354 155658.0 155658 19704085 139398 1 0 1 1 0 0 0 40.0%
1 8 36.419187 4 1111112 30509 2 30510 30507 155658.0 155658 19704085 139233 1 0 1 1 0 0 0 22.0%
1 16 36.353475 4 588236 16181 0 16182 16181 155658.0 155658 19704085 139183 1 0 1 1 0 0 0 19.2%
1 32 36.251356 4 303031 8359 0 8359 8359 155658.0 155658 19704085 139120 1 0 1 1 0 0 0 14.8%
1 64 36.203692 4 153847 4249 0 4250 4249 155658.0 155658 19704085 139071 1 0 1 1 0 0 0 13.0%
1 256 9.965876 4 38911 3904 0 3906 3904 100875.2 100874 3178266 60108 38912 0 1 1 0 0 0 17.9%
1 1024 2.637501 4 9757 3699 1 3700 3697 26610.0 26610 801818 15071 9758 0 1 1 0 0 0 19.5%
1 4096 0.806745 4 2441 3026 1 3027 3024 7988.0 7988 205914 3773 2442 0 1 1 0 0 0 18.3%
64 1 36.611243 4 9846154 268938 5 268942 268921 155669.8 155705 19704085 139330 2 0 1 1 0 0 0 29.9%
64 8 36.559471 4 8888896 243135 11 243156 243124 155658.0 155658 19704085 139261 1 0 1 1 0 0 0 28.1%
64 16 36.510319 4 8000000 219116 15 219126 219101 155658.0 155658 19704085 139173 1 0 1 1 0 0 0 26.3%
64 32 36.439069 4 6666688 182954 9 182964 182943 155658.0 155658 19704085 139274 1 0 1 1 0 0 0 23.2%
64 64 36.334808 4 5000000 137609 11 137612 137596 155658.0 155658 19704085 139258 2 0 1 1 0 0 0 19.1%
64 256 20.624759 4 2000000 96971 88 97059 92717 138296.0 131401 7877370 98332 31282 0 1 1 0 0 0 17.2%
64 1024 6.260598 4 588288 93967 1429 94905 88051 45939.5 41327 2324346 30361 9193 0 1 1 0 0 0 17.8%
64 4096 1.881338 4 153856 81780 140 81920 81520 16109.8 16092 582714 11617 2678 0 1 1 0 0 0 18.2%
4) perf_fast_forward --run-tests=large-partition-slicing
Caching enabled, each line shows the median run from many iterations
TL;DR: We can observe reduction in IO which translates to reduction in execution time,
especially for slicing in the middle of partition.
perf_fast_forward --run-tests=large-partition-slicing --datasets=large-part-ds1 -c1 -m1G --keep-cache-across-test-cases
Config: rows: 10000000, value size: 2000
Before:
offset read time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu
0 1 0.000491 127 1 2037 24 2109 127 4.0 4 128 2 2 0 1 1 0 0 0 157 80 3058208 15.0%
0 32 0.000561 1740 32 56995 410 60031 47208 5.0 5 160 3 2 0 1 1 0 0 0 386 111 113353 17.5%
0 256 0.002052 488 256 124736 7111 144762 89053 16.6 17 672 14 2 0 1 1 0 0 0 2113 446 52669 18.6%
0 4096 0.016437 61 4096 249199 692 252389 244995 69.4 69 8640 57 5 0 1 1 0 0 0 26638 1717 23321 22.4%
5000000 1 0.002171 221 1 461 2 466 221 25.0 25 268 3 3 0 1 1 0 0 0 638 376 14311524 10.2%
5000000 32 0.002392 404 32 13376 48 13528 13015 27.0 27 332 5 3 0 1 1 0 0 0 931 432 489691 11.9%
5000000 256 0.003659 279 256 69967 764 73130 52563 39.5 41 780 19 3 0 1 1 0 0 0 2689 825 93756 15.8%
5000000 4096 0.018592 55 4096 220313 433 234214 218803 94.2 94 9484 62 9 0 1 1 0 0 0 27349 2213 26562 21.0%
After:
offset read time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu
0 1 0.000229 115 1 4371 85 4585 115 2.1 2 64 1 1 1 0 0 0 0 0 90 31 1314749 22.2%
0 32 0.000277 2174 32 115674 1015 128109 14144 3.0 3 96 2 1 1 0 0 0 0 0 319 62 52508 26.1%
0 256 0.001786 576 256 143298 5534 179142 113715 14.7 17 544 15 1 1 0 0 0 0 0 2110 453 45419 21.4%
0 4096 0.015498 61 4096 264289 2006 268850 259342 67.4 67 8576 59 4 1 0 0 0 0 0 26657 1738 22897 23.7%
5000000 1 0.000415 233 1 2411 15 2456 234 4.1 4 128 2 2 1 0 0 0 0 0 199 72 2644719 16.8%
5000000 32 0.000635 1413 32 50398 349 51149 46439 6.0 6 192 4 2 1 0 0 0 0 0 458 128 125893 18.6%
5000000 256 0.002028 486 256 126228 3024 146327 82559 17.8 18 1024 13 4 1 0 0 0 0 0 2123 385 51787 19.6%
5000000 4096 0.016836 61 4096 243294 814 263434 241660 73.0 73 9344 62 8 1 0 0 0 0 0 26922 1920 24389 22.4%
Future work:
- Check the impact on non-uniform workloads. Caching sstable indexes takes space away from the row cache
which may reduce the hit ratio.
- Reduce memory footprint of partition index cache. Currently, about 8x bloat over the on-disk size.
- Disable cache population for "bypass cache" reads
- Add a switch to disable sstable index caching, per-node, maybe per-table
- Better sstable index format. Current format leads to inefficiency in caching since only some elements of the cached
page can be hot. A B-tree index would be more efficient. Same applies to the partition index. Only some elements in
the partition index page can be hot.
- Add heuristic for reducing index file IO size when large partitions are anticipated. If we're bound by disk's
bandwidth it's wasteful to read the front of promoted index using 32K IO, better use 4K which should cover the
partition entry and then let binary search read the rest.
In V2:
- Fixed perf_fast_forward regression in the number of IOs used to read partition index page
The reader uses 32K reads, which were split by page cache into 4K reads
Fix by propagating IO size hints to page cache and using single IO to populate it.
New patch: "cached_file: Issue single I/O for the whole read range on miss"
- Avoid large allocations to store partition index page entries (due to managed_vector storage).
There is a unit test which detects this and fails.
Fixed by implementing chunked_managed_vector, based on chunked_vector.
- fixed bug in cached_file::evict_gently() where the wrong allocation strategy was used to free btree chunks
- Simplify region_impl::free_buf() according to Avi's suggestions
- Fit segment_kind in segment_descriptor::_free_space and lift requirement that _buf_pointers emptiness determines the kind
- Workaround sigsegv which was most likely due to coroutine miscompilation. Worked around by manipulating local object scope.
- Wire up system/drop_sstable_caches RESTful API
- Fix use-after-move on permit for the old scanning ka/la index reader
- Fixed more cases of double open_data() in tests leading to assert failure
- Adjusted cached_file class doc to account for changes in behavior.
- Rebased
Fixes#7079.
Refs #363.
"
* tag 'sstable-index-caching-v2' of github.com:tgrabiec/scylla: (39 commits)
api: Drop sstable index caches on system/drop_sstable_caches
cached_file: Issue single I/O for the whole read range on miss
row_cache: cache_tracker: Do not register metrics when constructed for tests
sstables, cached_file: Evict cache gently when sstable is destroyed
sstables: Hide partition_index_cache implementation away from sstables.hh
sstables: Drop shared_index_lists alias
sstables: Destroy partition index cache gently
sstables: Cache partition index pages in LSA and link to LRU
utils: Introduce lsa::weak_ptr<>
sstables: Rename index_list to partition_index_page and shared_index_lists to partition_index_cache
sstables, cached_file: Avoid copying buffers from cache when parsing promoted index
cached_file: Introduce get_page_units()
sstables: read: Document that primitive_consumer::read_32() is alloc-free
sstables: read: Count partition index page evictions
sstables: Drop the _use_binary_search flag from index entries
sstables: index_reader: Keep index objects under LSA
lsa: chunked_managed_vector: Adapt more to managed_vector
utils: lsa: chunked_managed_vector: Make LSA-aware
test: chunked_managed_vector_test: Make exception_safe_class standard layout
lsa: Copy chunked_vector to chunked_managed_vector
...
logalloc has a nice leak/double-free sanitizer, with the nice
feature of capturing backtraces to make error reports easy to
track down. But capturing backtraces is itself very expensive.
This patch makes backtrace capture optional, reducing database_test
runtime from 30 minutes to 20 minutes on my machine.
Closes#8978
Currently, reading a page range would issue I/O for each missing
page. This is inefficient, better to issue a single I/O for the whole
range and populate cache from that.
As an optimization, issue a single I/O if the first page is missing.
This is important for index reads which optimistically try to read
32KB of index file to read the partition index page.
As part of this change, the container for partition index pages was
changed from utils::loading_shared_values to intrusive_btree. This is
to avoid reactor stalls which the former induces with a large number
of elements (pages) due to its use of a hashtable under the hood,
which reallocates contiguous storage.
Simplifies managing non-owning references to LSA-managed objects. The
lsa::weak_ptr is a smart pointer which is not invalidated by LSA and
can be used safely in any allocator context. Dereferenced will always
give a valid reference.
This can be used as a building block for implementing cursors into
LSA-based caches.
Example simple use:
// LSA-managed
struct X : public lsa::weakly_referencable<X> {
int value;
};
lsa::weak_ptr<X> x_ptr = with_allocator(region(), [] {
X* x = current_allocator().construct<X>();
return x->weak_from_this();
});
std::cout << x_ptr->value;
Will be needed later for reading a page view which cannot use
make_tracked_temporary_buffer(). Standardize on get_page_units(),
converting existing code to wrap the units in a deleter.
Will be needed by index reader to ensure that destructor doesn't
invoke the allocator so that all is destroyed in the desried
allocation context before the object is destroyed.
After this patch, there is a singe index file page cache per
sstable, shared by index readers. The cache survives reads,
which reduces amount of I/O on subsequent reads.
As part of this, cached_file needed to be adjusted in the following ways.
The page cache may occupy a significant portion of memory. Keeping the
pages in the standard allocator could cause memory fragmentation
problems. To avoid them, the cache_file is changed to keep buffers in LSA
using lsa_buffer allocation method.
When a page is needed by the seastar I/O layer, it needs to be copied
to a temporary_buffer which is stable, so must be allocated in the
standard allocator space. We copy the page on-demand. Concurrent
requests for the same page will share the temporary_buffer. When page
is not used, it only lives in the LSA space.
In the subsequent patches cached_file::stream will be adjusted to also support
access via cached_page::ptr_type directly, to avoid materializating a
temporary_buffer.
While a page is used, it is not linked in the LRU so that it is not
freed. This ensures that the storage which is actively consumed
remains stable, either via temporary_buffer (kept alive by its
deleter), or by cached_page::ptr_type directly.
lsa_buffer is similar in spirit to std::unique_ptr<char[]>. It owns
buffers allocated inside LSA segments. It uses an alternative
allocation method which differs from regular LSA allocations in the
following ways:
1) LSA segments only hold buffers, they don't hold metadata. They
also don't mix with standard allocations. So a 128K segment can
hold 32 4K buffers.
2) objects' life time is managed by lsa_buffer, an owning smart
pointer, which is automatically updated when buffers are migrated
to another segment. This makes LSA allocations easier to use and
off-loads metadata management to the client (which can keep the
lsa_buffer wherever he wants).
The metadata is kept inside segment_descriptor, in a vector. Each
allocated buffer will have an entangled object there (8 bytes), which
is paired with an entabled object inside lsa_buffer.
The reason to have an alternative allocation method is to efficiently
pack buffers inside LSA segments.
It's an adpator between seastar::file and cached_file. It gives a
seastar::file which will serve reads using a given cached_file as a
read-through cache.
We want buffers to be accounted only when they are used outside
cached_file. Cached pages should not be accounted because they will
stay around for longer than the read after subsequent commits.
In preparation for tracking different kinds of objects, not just
rows_entry, in the LRU, switch to the LRU implementation form
utils/lru.hh which can hold arbitrary element type.
The LRU can link objects of different types, which is achieved by
having a virtual base class called "evictable" from which the linked
objects should inherit. Whe the object is removed from the LRU,
evictable::on_evicted() is called.
The container is non-owning.
clear_gently of the foreign_ptr needs to run on the owning
shard, so provide a specialization from the SmartPointer
implementation.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Define a bunch of clear_gently methods that asynchronously
clear the contents of containers and allow yielding.
This replaces clear_gently(std::list<T>&) used by row level
repair by a more generic template implementation.
Note that we do not use coroutines in this patch
to facilitate backporting to releases that do not support coroutines
and since a miscompilation bug was hit with clang++ 11 when attempting
to coroutinize this patch (see
https://bugs.llvm.org/show_bug.cgi?id=50345).
Test: stall_free_test(debug)
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
std::copy_if runs without yielding.
See https://github.com/scylladb/scylla/issues/8897#issuecomment-867522480
Also, eliminate extraneous loop on merge
first1 will point to the inserted value which is a copy of *first2.
Since list2 is sorted in ascending order, the next item from list2
will never be less than the one we've just inserted,
so we waste an iteration to merely increment first1 again.
Fixes#8897
Test: unit(dev), stall_free_test(debug)
DTest: repair_additional_test.py:RepairAdditionalTest.{repair_same_row_diff_value_3nodes_diff_shard_count_test,repair_disjoint_row_3nodes_diff_shard_count_test} (dev)
Closes#8925
* github.com:scylladb/scylla:
utils: merge_to_gently: eliminate extraneous loop on merge
utils: merge_to_gently: prevent stall in std::copy_if
first1 will point to the inserted value which is a copy of *first2.
Since list2 is sorted in ascending order, the next item from list2
will never be less than the one we've just inserted,
so we waste an iteration to merely increment first1 again.
Note that the standard states that no iterators or references are invalidated
on insert so we can safely keep looking at `first1` after inserting a copy of
`*first2` before it.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
The helper is used to walk the tree key-by-key destroying it
in the mean time. Current implementation of this method just
uses the "regular" erasing code which actually rebalances the
tree despite the name.
The biggest problem with removing the rebalancing is that at
some point non-balanced tree may have the left-most key on an
inner node, so to make 100% rebalance-less unlink every other
method of the tree would have to be prepared for that. However,
there's an option to make "light rebalance" (as it's called in
this patch) that only maintains this crucial property of the
tree -- the left-most key is on the leaf.
Some more tech details. Current rebalancer starts when the
node population falls below 1/2 of its capacity and tries to
- grab a key from one of the siblings if it's balanced
- merge two siblings together if they are small enough
The light rebalance is lighter in two ways. First, it leaves
the node unbalanced until it becomes empty. And then it goes
ahead and replaces it with the next sibling.
This change removes ~60% of the keys movements on random test.
Keys still move when the sibling replace happens because in
this case the separation key needs to be placed at the right
sibling 0 position which means shifting all its keys right.
tests: unit(debug)
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
Message-Id: <20210623083836.27491-1-xemul@scylladb.com>
"
The LSA small objects allocation latency is greatly affected by
the way this allocator encodes the object descriptor in front of
each allocated slot.
Nowadays it's one of VLE variants implemented with the help of a
loop. Re-implementing this piece with less instructions and without
a loop allows greatly reducing the allocation latency.
The speed-up mostly comes from loop-less code that doesn't confuse
branch predictor. Also the express encoder seems to benefit from
writing 8 bytes of the encoded value in one go, rather than byte-
-by-byte.
Perf measurements:
1. (new) logallog test shows ~40% smaller times
2. perf_mutation in release mode shows ~2% increase in tps
3. the encoder itself is 2 - 4 times faster on x86_64 and
1.05 - 3 times faster on aarch64. The speed-up depends on
the 'encoded length', old encoder has linear time, the
new one is constant
tests: unit(dev), perf(release), just encoder on Aarch64
"
* 'br-lsa-alloc-latency-4' of https://github.com/xemul/scylla:
lsa: Use express encoder
uleb64: Add express encoding
lsa: Extract uleb64 code into header
test: LSA allocation perf test
To make it possible to use the express encoder, lsa needs to
make sure that the value is below express supreme value and
provide the size of the gap after the encoded value.
Both requirements can be satisfied when encoding the migrator
index on object allocation.
On free the encoded value can be larger, so the extended
express encoder will need more instructions and will not be
that efficient, so the old encoder is used there.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
Standard encoding is compiled into a loop that puts values
into memory byte-by-byte. This works slowly, but reliably.
When allocating an object LSA uses ubel64 encoder with 2
features that allow to optimize the encoder:
1. the value is migrator.index() which is small enough
to fit 2 bytes when encoded
2. After the descriptor there usually comes an object
which is of 8+ bytes in size
Feature #1 makes it possible to encode the value with just
a few instructions. In O3 level clang makes it like
mov %esi,%ecx
and $0xfc0,%ecx
and $0x3f,%esi
lea (%rsi,%rcx,4),%ecx
add $0x40,%ecx
Next, the encoder needs to put the value into a gap whose
size depends on the alignment of previous and current objects,
so the classical algo loops through this size. Feature #2
makes it possible to put the encoded value and the needed
amount of zeros by using 2 64-bit movs. In this case the
encoded value gets off the needed size and overwrites some
memory after. That's OK, as this overwritten memory is where
the allocated object _will_ be, so the contents there is not
of any interest.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The LSA code encodes an object descriptor before the object
itself. The descriptor is 32-bit value and to put it in an
efficient manner it's encoded into unsigned little-endian
base-64 sequence.
The encoding code is going to be optimized, so put it into a
dedicated header in advance.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The one was added to smothly switch tri-comparing stuff from int
to strong-ordering. As for today only tests still need it and the
conversion is pretty simple, plus operator<<(ostream&) for the
std::strong_ordering type.
* xemul/br-remove-int-or-strong-ordering-2:
util: Drop int_or_strong_ordering concept
tests: Switch total-order-check onto strong_ordering
to_string: Add formatter for strong_ordering
tests: Return strong-ordering from tri-comparators
Add an implicit converter of the enum_option to the underyling enum
it is holding. This is needed for using switch() on an enum_option.
Signed-off-by: Nadav Har'El <nyh@scylladb.com>
Currently, advance_and_wait() allocates a new gate
which might fail. Rather than returning this failure
as an exceptional future - which will require its callers
to handle that failure, keep the function as noexcept and
let an exception from make_lw_shared<gate>() terminate the program.
This makes the function "fail-free" to its callers,
in particular, when called from the table::stop() path where
we can't do much about these failures and we require close/stop
functions to always succeed.
The alternative of make the allocation of a new gate optional
and covering from it in start() is possible but was deemed not
worth it as it will add complexity and cost to start() that's
called on the common, hot, path.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Message-Id: <20210525055336.1190029-1-bhalevy@scylladb.com>
config_file.cc instantiates std::istream& std::operator>>(std::istream&,
std::unordered_map<seastar::sstring, seastar::sstring>&), but that
instantiation is ignored since config_file_impl.hh specializes
that signature. -Winstantiation-after-specialization warns about it,
so re-enable it now that the code base is clean.
Also remove the matching "extern template" declaration, which has no
definition any more.
Closes#8696
"
The patch set is an assorted collection of header cleanups, e.g:
* Reduce number of boost includes in header files
* Switch to forward declarations in some places
A quick measurement was performed to see if these changes
provide any improvement in build times (ccache cleaned and
existing build products wiped out).
The results are posted below (`/usr/bin/time -v ninja dev-build`)
for 24 cores/48 threads CPU setup (AMD Threadripper 2970WX).
Before:
Command being timed: "ninja dev-build"
User time (seconds): 28262.47
System time (seconds): 824.85
Percent of CPU this job got: 3979%
Elapsed (wall clock) time (h:mm:ss or m:ss): 12:10.97
Average shared text size (kbytes): 0
Average unshared data size (kbytes): 0
Average stack size (kbytes): 0
Average total size (kbytes): 0
Maximum resident set size (kbytes): 2129888
Average resident set size (kbytes): 0
Major (requiring I/O) page faults: 1402838
Minor (reclaiming a frame) page faults: 124265412
Voluntary context switches: 1879279
Involuntary context switches: 1159999
Swaps: 0
File system inputs: 0
File system outputs: 11806272
Socket messages sent: 0
Socket messages received: 0
Signals delivered: 0
Page size (bytes): 4096
Exit status: 0
After:
Command being timed: "ninja dev-build"
User time (seconds): 26270.81
System time (seconds): 767.01
Percent of CPU this job got: 3905%
Elapsed (wall clock) time (h:mm:ss or m:ss): 11:32.36
Average shared text size (kbytes): 0
Average unshared data size (kbytes): 0
Average stack size (kbytes): 0
Average total size (kbytes): 0
Maximum resident set size (kbytes): 2117608
Average resident set size (kbytes): 0
Major (requiring I/O) page faults: 1400189
Minor (reclaiming a frame) page faults: 117570335
Voluntary context switches: 1870631
Involuntary context switches: 1154535
Swaps: 0
File system inputs: 0
File system outputs: 11777280
Socket messages sent: 0
Socket messages received: 0
Signals delivered: 0
Page size (bytes): 4096
Exit status: 0
The observed improvement is about 5% of total wall clock time
for `dev-build` target.
Also, all commits make sure that headers stay self-sufficient,
which would help to further improve the situation in the future.
"
* 'feature/header_cleanups_v1' of https://github.com/ManManson/scylla:
transport: remove extraneous `qos/service_level_controller` includes from headers
treewide: remove evidently unneded storage_proxy includes from some places
service_level_controller: remove extraneous `service/storage_service.hh` include
sstables/writer: remove extraneous `service/storage_service.hh` include
treewide: remove extraneous database.hh includes from headers
treewide: reduce boost headers usage in scylla header files
cql3: remove extraneous includes from some headers
cql3: various forward declaration cleanups
utils: add missing <limits> header in `extremum_tracking.hh`
Yet another patch preventing potentially large allocations.
Currently, collection_mutation{_view,}_description linearize each collection
value during deserialization. It's not unthinkable that a user adds a
large element to a list or a map, so let's avoid that.
This patch removes the dependency on linearizing_input_stream, which does not
provide a way to read fragmented subbuffers, and replaces it with a new
helper, which does. (Extending linearizing_input_stream is not viable without
rewriting it completely).
Only linearization of collection values is corrected in this patch.
Collection keys are still linearized. Storing them in managed_bytes is likely
to be more harmful than helpful, because large map keys are extremely unlikely,
and UUIDs, which are used as keys in lists, do not fit into manages_bytes's
small value optimization, so this would incure an extra allocation for every
list element.
Note: this patch leaves utils/linearizing_input_stream.hh unused.
Refs: #8120Closes#8690
As an optimization for optimistic cases, reclaim_from_evictable first evicts
the requested amount of memory before attempting to reclaim segments through
compactions. However, due to an oversight, it does this before every compaction
instead of once before all compactions.
Usually reclaim_from_evictable is called with small targets, or is preemptible,
and in those cases this issue is not visible. However, when the target is bigger
than one segment and the reclaim is not preemptible, which is he case when it's
called from allocating_section, this results in a quadratic explosion of
evictions, which can evict several hundred MiB to reclaim a few MiB.
Fix that by calculating the target of memory eviction only once, instead of
recalculating it after every compaction.
Fixes#8542.
Closes#8611
Following Nadav's advice, instead of ignoring the test
in sanitize/debug modes, the allocator simply has a special path
of failing sufficiently large allocation requests.
With that, a problem with the address sanitizer is bypassed
and other debug mode sanitizers can inspect and check
if there are no more problems related to wrapping the original
rapidjson allocator.
Closes#8539
The B+ tree is not intrusive and supports both kinds of objects --
dynamic (in sense of previous patch) and fixed-size. Respectively,
the nodes provide .storage_size() method and get the embedded object
storage size themselves. Thus, if a dynamic object is used with the
tree but it misses the .storage_size() itself this would come
unnoticed. Fortunately, dynamic objects use the .reconstruct()
method, so the method should be equipeed with the DybnamicObject
concept.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
Usually lsa allocation is performed with the construct() helper that
allocates a sizeof(T) slot and constructs it in place. Some rare
objects have dynamic size, so they are created by alloc()ating a
slot of some specific size and (!) must provide the correct overload
of size_for_allocation_strategy that reports back the relevant
storage size.
This "must provide" is not enforced, if missed a default sizer would
be instantiated, but won't work properly. This patch makes all users
of alloc() conform to DynamicObject concept which requires the
presense of .storage_size() method to tell that size.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
