Summary:
Background. One of the core risks of chosing HyperClockCache is ending up with degraded performance if estimated_entry_charge is very significantly wrong. Too low leads to under-utilized hash table, which wastes a bit of (tracked) memory and likely increases access times due to larger working set size (more TLB misses). Too high leads to fully populated hash table (at some limit with reasonable lookup performance) and not being able to cache as many objects as the memory limit would allow. In either case, performance degradation is graceful/continuous but can be quite significant. For example, cutting block size in half without updating estimated_entry_charge could lead to a large portion of configured block cache memory (up to roughly 1/3) going unused.
Fix. This change adds a mechanism through which the DB periodically probes the block cache(s) for "problems" to report, and adds diagnostics to the HyperClockCache for bad estimated_entry_charge. The periodic probing is currently done with DumpStats / stats_dump_period_sec, and diagnostics reported to info_log (normally LOG file).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10965
Test Plan:
unit test included. Doesn't cover all the implemented subtleties of reporting, but ensures basics of when to report or not.
Also manual testing with db_bench. Create db with
```
./db_bench --benchmarks=fillrandom,flush --num=3000000 --disable_wal=1
```
Use and check LOG file for HyperClockCache for various block sizes (used as estimated_entry_charge)
```
./db_bench --use_existing_db --benchmarks=readrandom --num=3000000 --duration=20 --stats_dump_period_sec=8 --cache_type=hyper_clock_cache -block_size=XXXX
```
Seeing warnings / errors or not as expected.
Reviewed By: anand1976
Differential Revision: D41406932
Pulled By: pdillinger
fbshipit-source-id: 4ca56162b73017e4b9cec2cad74466f49c27a0a7
Summary:
This was just a stepping stone to what eventually became HyperClockCache, and is now just more code to maintain.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10954
Test Plan: tests updated
Reviewed By: akankshamahajan15
Differential Revision: D41310123
Pulled By: pdillinger
fbshipit-source-id: 618ee148a1a0a29ee756ba8fe28359617b7cd67c
Summary:
Ran `find ./db/ -type f | xargs clang-format -i`. Excluded minor changes it tried to make on db/db_impl/. Everything else it changed was directly under db/ directory. Included minor manual touchups mentioned in PR commit history.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10910
Reviewed By: riversand963
Differential Revision: D40880683
Pulled By: ajkr
fbshipit-source-id: cfe26cda05b3fb9a72e3cb82c286e21d8c5c4174
Summary:
This new property allows users to trigger the background block cache stats collection mode through the `GetProperty()` and `GetMapProperty()` APIs. The background mode has much lower overhead at the expense of returning stale values in more cases.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10832
Test Plan: updated unit test
Reviewed By: pdillinger
Differential Revision: D40497883
Pulled By: ajkr
fbshipit-source-id: bdcc93402f426463abb2153756aad9e295447343
Summary:
This change establishes a distinctive name for the experimental new lock-free clock cache (originally developed by guidotag and revamped in PR https://github.com/facebook/rocksdb/issues/10626). A few reasons:
* We want to make it clear that this is a fundamentally different implementation vs. the old clock cache, to avoid people saying "I already tried clock cache."
* We want to highlight the key feature: it's fast (especially under parallel load)
* Because it requires an estimated charge per entry, it is not drop-in API compatible with old clock cache. This estimate might always be required for highest performance, and giving it a distinct name should reduce confusion about the distinct API requirements.
* We might develop a variant requiring the same estimate parameter but with LRU eviction. In that case, using the name HyperLRUCache should make things more clear. (FastLRUCache is just a prototype that might soon be removed.)
Some API detail:
* To reduce copy-pasting parameter lists, etc. as in LRUCache construction, I have a `MakeSharedCache()` function on `HyperClockCacheOptions` instead of `NewHyperClockCache()`.
* Changes -cache_type=clock_cache to -cache_type=hyper_clock_cache for applicable tools. I think this is more consistent / sustainable for reasons already stated.
For performance tests see https://github.com/facebook/rocksdb/pull/10626
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10684
Test Plan: no interesting functional changes; tests updated
Reviewed By: anand1976
Differential Revision: D39547800
Pulled By: pdillinger
fbshipit-source-id: 5c0fe1b5cf3cb680ab369b928c8569682b9795bf
Summary:
* Consolidates most metadata into a single word per slot so that more
can be accomplished with a single atomic update. In the common case,
Lookup was previously about 4 atomic updates, now just 1 atomic update.
Common case Release was previously 1 atomic read + 1 atomic update,
now just 1 atomic update.
* Eliminate spins / waits / yields, which likely threaten some "lock free"
benefits. Compare-exchange loops are only used in explicit Erase, and
strict_capacity_limit=true Insert. Eviction uses opportunistic compare-
exchange.
* Relaxes some aggressiveness and guarantees. For example,
* Duplicate Inserts will sometimes go undetected and the shadow duplicate
will age out with eviction.
* In many cases, the older Inserted value for a given cache key will be kept
(i.e. Insert does not support overwrite).
* Entries explicitly erased (rather than evicted) might not be freed
immediately in some rare cases.
* With strict_capacity_limit=false, capacity limit is not tracked/enforced as
precisely as LRUCache, but is self-correcting and should only deviate by a
very small number of extra or fewer entries.
* Use smaller "computed default" number of cache shards in many cases,
because benefits to larger usage tracking / eviction pools outweigh the small
cost of more lock-free atomic contention. The improvement in CPU and I/O
is dramatic in some limit-memory cases.
* Even without the sharding change, the eviction algorithm is likely more
effective than LRU overall because it's more stateful, even though the
"hot path" state tracking for it is essentially free with ref counting. It
is like a generalized CLOCK with aging (see code comments). I don't have
performance numbers showing a specific improvement, but in theory, for a
Poisson access pattern to each block, keeping some state allows better
estimation of time to next access (Poisson interval) than strict LRU. The
bounded randomness in CLOCK can also reduce "cliff" effect for repeated
range scans approaching and exceeding cache size.
## Hot path algorithm comparison
Rough descriptions, focusing on number and kind of atomic operations:
* Old `Lookup()` (2-5 atomic updates per probe):
```
Loop:
Increment internal ref count at slot
If possible hit:
Check flags atomic (and non-atomic fields)
If cache hit:
Three distinct updates to 'flags' atomic
Increment refs for internal-to-external
Return
Decrement internal ref count
while atomic read 'displacements' > 0
```
* New `Lookup()` (1-2 atomic updates per probe):
```
Loop:
Increment acquire counter in meta word (optimistic)
If visible entry (already read meta word):
If match (read non-atomic fields):
Return
Else:
Decrement acquire counter in meta word
Else if invisible entry (rare, already read meta word):
Decrement acquire counter in meta word
while atomic read 'displacements' > 0
```
* Old `Release()` (1 atomic update, conditional on atomic read, rarely more):
```
Read atomic ref count
If last reference and invisible (rare):
Use CAS etc. to remove
Return
Else:
Decrement ref count
```
* New `Release()` (1 unconditional atomic update, rarely more):
```
Increment release counter in meta word
If last reference and invisible (rare):
Use CAS etc. to remove
Return
```
## Performance test setup
Build DB with
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16
```
Test with
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom -readonly -num=30000000 -bloom_bits=16 -cache_index_and_filter_blocks=1 -cache_size=${CACHE_MB}000000 -duration 60 -threads=$THREADS -statistics
```
Numbers on a single socket Skylake Xeon system with 48 hardware threads, DEBUG_LEVEL=0 PORTABLE=0. Very similar story on a dual socket system with 80 hardware threads. Using (every 2nd) Fibonacci MB cache sizes to sample the territory between powers of two. Configurations:
base: LRUCache before this change, but with db_bench change to default cache_numshardbits=-1 (instead of fixed at 6)
folly: LRUCache before this change, with folly enabled (distributed mutex) but on an old compiler (sorry)
gt_clock: experimental ClockCache before this change
new_clock: experimental ClockCache with this change
## Performance test results
First test "hot path" read performance, with block cache large enough for whole DB:
4181MB 1thread base -> kops/s: 47.761
4181MB 1thread folly -> kops/s: 45.877
4181MB 1thread gt_clock -> kops/s: 51.092
4181MB 1thread new_clock -> kops/s: 53.944
4181MB 16thread base -> kops/s: 284.567
4181MB 16thread folly -> kops/s: 249.015
4181MB 16thread gt_clock -> kops/s: 743.762
4181MB 16thread new_clock -> kops/s: 861.821
4181MB 24thread base -> kops/s: 303.415
4181MB 24thread folly -> kops/s: 266.548
4181MB 24thread gt_clock -> kops/s: 975.706
4181MB 24thread new_clock -> kops/s: 1205.64 (~= 24 * 53.944)
4181MB 32thread base -> kops/s: 311.251
4181MB 32thread folly -> kops/s: 274.952
4181MB 32thread gt_clock -> kops/s: 1045.98
4181MB 32thread new_clock -> kops/s: 1370.38
4181MB 48thread base -> kops/s: 310.504
4181MB 48thread folly -> kops/s: 268.322
4181MB 48thread gt_clock -> kops/s: 1195.65
4181MB 48thread new_clock -> kops/s: 1604.85 (~= 24 * 1.25 * 53.944)
4181MB 64thread base -> kops/s: 307.839
4181MB 64thread folly -> kops/s: 272.172
4181MB 64thread gt_clock -> kops/s: 1204.47
4181MB 64thread new_clock -> kops/s: 1615.37
4181MB 128thread base -> kops/s: 310.934
4181MB 128thread folly -> kops/s: 267.468
4181MB 128thread gt_clock -> kops/s: 1188.75
4181MB 128thread new_clock -> kops/s: 1595.46
Whether we have just one thread on a quiet system or an overload of threads, the new version wins every time in thousand-ops per second, sometimes dramatically so. Mutex-based implementation quickly becomes contention-limited. New clock cache shows essentially perfect scaling up to number of physical cores (24), and then each hyperthreaded core adding about 1/4 the throughput of an additional physical core (see 48 thread case). Block cache miss rates (omitted above) are negligible across the board. With partitioned instead of full filters, the maximum speed-up vs. base is more like 2.5x rather than 5x.
Now test a large block cache with low miss ratio, but some eviction is required:
1597MB 1thread base -> kops/s: 46.603 io_bytes/op: 1584.63 miss_ratio: 0.0201066 max_rss_mb: 1589.23
1597MB 1thread folly -> kops/s: 45.079 io_bytes/op: 1530.03 miss_ratio: 0.019872 max_rss_mb: 1550.43
1597MB 1thread gt_clock -> kops/s: 48.711 io_bytes/op: 1566.63 miss_ratio: 0.0198923 max_rss_mb: 1691.4
1597MB 1thread new_clock -> kops/s: 51.531 io_bytes/op: 1589.07 miss_ratio: 0.0201969 max_rss_mb: 1583.56
1597MB 32thread base -> kops/s: 301.174 io_bytes/op: 1439.52 miss_ratio: 0.0184218 max_rss_mb: 1656.59
1597MB 32thread folly -> kops/s: 273.09 io_bytes/op: 1375.12 miss_ratio: 0.0180002 max_rss_mb: 1586.8
1597MB 32thread gt_clock -> kops/s: 904.497 io_bytes/op: 1411.29 miss_ratio: 0.0179934 max_rss_mb: 1775.89
1597MB 32thread new_clock -> kops/s: 1182.59 io_bytes/op: 1440.77 miss_ratio: 0.0185449 max_rss_mb: 1636.45
1597MB 128thread base -> kops/s: 309.91 io_bytes/op: 1438.25 miss_ratio: 0.018399 max_rss_mb: 1689.98
1597MB 128thread folly -> kops/s: 267.605 io_bytes/op: 1394.16 miss_ratio: 0.0180286 max_rss_mb: 1631.91
1597MB 128thread gt_clock -> kops/s: 691.518 io_bytes/op: 9056.73 miss_ratio: 0.0186572 max_rss_mb: 1982.26
1597MB 128thread new_clock -> kops/s: 1406.12 io_bytes/op: 1440.82 miss_ratio: 0.0185463 max_rss_mb: 1685.63
610MB 1thread base -> kops/s: 45.511 io_bytes/op: 2279.61 miss_ratio: 0.0290528 max_rss_mb: 615.137
610MB 1thread folly -> kops/s: 43.386 io_bytes/op: 2217.29 miss_ratio: 0.0289282 max_rss_mb: 600.996
610MB 1thread gt_clock -> kops/s: 46.207 io_bytes/op: 2275.51 miss_ratio: 0.0290057 max_rss_mb: 637.934
610MB 1thread new_clock -> kops/s: 48.879 io_bytes/op: 2283.1 miss_ratio: 0.0291253 max_rss_mb: 613.5
610MB 32thread base -> kops/s: 306.59 io_bytes/op: 2250 miss_ratio: 0.0288721 max_rss_mb: 683.402
610MB 32thread folly -> kops/s: 269.176 io_bytes/op: 2187.86 miss_ratio: 0.0286938 max_rss_mb: 628.742
610MB 32thread gt_clock -> kops/s: 855.097 io_bytes/op: 2279.26 miss_ratio: 0.0288009 max_rss_mb: 733.062
610MB 32thread new_clock -> kops/s: 1121.47 io_bytes/op: 2244.29 miss_ratio: 0.0289046 max_rss_mb: 666.453
610MB 128thread base -> kops/s: 305.079 io_bytes/op: 2252.43 miss_ratio: 0.0288884 max_rss_mb: 723.457
610MB 128thread folly -> kops/s: 269.583 io_bytes/op: 2204.58 miss_ratio: 0.0287001 max_rss_mb: 676.426
610MB 128thread gt_clock -> kops/s: 53.298 io_bytes/op: 8128.98 miss_ratio: 0.0292452 max_rss_mb: 956.273
610MB 128thread new_clock -> kops/s: 1301.09 io_bytes/op: 2246.04 miss_ratio: 0.0289171 max_rss_mb: 788.812
The new version is still winning every time, sometimes dramatically so, and we can tell from the maximum resident memory numbers (which contain some noise, by the way) that the new cache is not cheating on memory usage. IMPORTANT: The previous generation experimental clock cache appears to hit a serious bottleneck in the higher thread count configurations, presumably due to some of its waiting functionality. (The same bottleneck is not seen with partitioned index+filters.)
Now we consider even smaller cache sizes, with higher miss ratios, eviction work, etc.
233MB 1thread base -> kops/s: 10.557 io_bytes/op: 227040 miss_ratio: 0.0403105 max_rss_mb: 247.371
233MB 1thread folly -> kops/s: 15.348 io_bytes/op: 112007 miss_ratio: 0.0372238 max_rss_mb: 245.293
233MB 1thread gt_clock -> kops/s: 6.365 io_bytes/op: 244854 miss_ratio: 0.0413873 max_rss_mb: 259.844
233MB 1thread new_clock -> kops/s: 47.501 io_bytes/op: 2591.93 miss_ratio: 0.0330989 max_rss_mb: 242.461
233MB 32thread base -> kops/s: 96.498 io_bytes/op: 363379 miss_ratio: 0.0459966 max_rss_mb: 479.227
233MB 32thread folly -> kops/s: 109.95 io_bytes/op: 314799 miss_ratio: 0.0450032 max_rss_mb: 400.738
233MB 32thread gt_clock -> kops/s: 2.353 io_bytes/op: 385397 miss_ratio: 0.048445 max_rss_mb: 500.688
233MB 32thread new_clock -> kops/s: 1088.95 io_bytes/op: 2567.02 miss_ratio: 0.0330593 max_rss_mb: 303.402
233MB 128thread base -> kops/s: 84.302 io_bytes/op: 378020 miss_ratio: 0.0466558 max_rss_mb: 1051.84
233MB 128thread folly -> kops/s: 89.921 io_bytes/op: 338242 miss_ratio: 0.0460309 max_rss_mb: 812.785
233MB 128thread gt_clock -> kops/s: 2.588 io_bytes/op: 462833 miss_ratio: 0.0509158 max_rss_mb: 1109.94
233MB 128thread new_clock -> kops/s: 1299.26 io_bytes/op: 2565.94 miss_ratio: 0.0330531 max_rss_mb: 361.016
89MB 1thread base -> kops/s: 0.574 io_bytes/op: 5.35977e+06 miss_ratio: 0.274427 max_rss_mb: 91.3086
89MB 1thread folly -> kops/s: 0.578 io_bytes/op: 5.16549e+06 miss_ratio: 0.27276 max_rss_mb: 96.8984
89MB 1thread gt_clock -> kops/s: 0.512 io_bytes/op: 4.13111e+06 miss_ratio: 0.242817 max_rss_mb: 119.441
89MB 1thread new_clock -> kops/s: 48.172 io_bytes/op: 2709.76 miss_ratio: 0.0346162 max_rss_mb: 100.754
89MB 32thread base -> kops/s: 5.779 io_bytes/op: 6.14192e+06 miss_ratio: 0.320399 max_rss_mb: 311.812
89MB 32thread folly -> kops/s: 5.601 io_bytes/op: 5.83838e+06 miss_ratio: 0.313123 max_rss_mb: 252.418
89MB 32thread gt_clock -> kops/s: 0.77 io_bytes/op: 3.99236e+06 miss_ratio: 0.236296 max_rss_mb: 396.422
89MB 32thread new_clock -> kops/s: 1064.97 io_bytes/op: 2687.23 miss_ratio: 0.0346134 max_rss_mb: 155.293
89MB 128thread base -> kops/s: 4.959 io_bytes/op: 6.20297e+06 miss_ratio: 0.323945 max_rss_mb: 823.43
89MB 128thread folly -> kops/s: 4.962 io_bytes/op: 5.9601e+06 miss_ratio: 0.319857 max_rss_mb: 626.824
89MB 128thread gt_clock -> kops/s: 1.009 io_bytes/op: 4.1083e+06 miss_ratio: 0.242512 max_rss_mb: 1095.32
89MB 128thread new_clock -> kops/s: 1224.39 io_bytes/op: 2688.2 miss_ratio: 0.0346207 max_rss_mb: 218.223
^ Now something interesting has happened: the new clock cache has gained a dramatic lead in the single-threaded case, and this is because the cache is so small, and full filters are so big, that dividing the cache into 64 shards leads to significant (random) imbalances in cache shards and excessive churn in imbalanced shards. This new clock cache only uses two shards for this configuration, and that helps to ensure that entries are part of a sufficiently big pool that their eviction order resembles the single-shard order. (This effect is not seen with partitioned index+filters.)
Even smaller cache size:
34MB 1thread base -> kops/s: 0.198 io_bytes/op: 1.65342e+07 miss_ratio: 0.939466 max_rss_mb: 48.6914
34MB 1thread folly -> kops/s: 0.201 io_bytes/op: 1.63416e+07 miss_ratio: 0.939081 max_rss_mb: 45.3281
34MB 1thread gt_clock -> kops/s: 0.448 io_bytes/op: 4.43957e+06 miss_ratio: 0.266749 max_rss_mb: 100.523
34MB 1thread new_clock -> kops/s: 1.055 io_bytes/op: 1.85439e+06 miss_ratio: 0.107512 max_rss_mb: 75.3125
34MB 32thread base -> kops/s: 3.346 io_bytes/op: 1.64852e+07 miss_ratio: 0.93596 max_rss_mb: 180.48
34MB 32thread folly -> kops/s: 3.431 io_bytes/op: 1.62857e+07 miss_ratio: 0.935693 max_rss_mb: 137.531
34MB 32thread gt_clock -> kops/s: 1.47 io_bytes/op: 4.89704e+06 miss_ratio: 0.295081 max_rss_mb: 392.465
34MB 32thread new_clock -> kops/s: 8.19 io_bytes/op: 3.70456e+06 miss_ratio: 0.20826 max_rss_mb: 519.793
34MB 128thread base -> kops/s: 2.293 io_bytes/op: 1.64351e+07 miss_ratio: 0.931866 max_rss_mb: 449.484
34MB 128thread folly -> kops/s: 2.34 io_bytes/op: 1.6219e+07 miss_ratio: 0.932023 max_rss_mb: 396.457
34MB 128thread gt_clock -> kops/s: 1.798 io_bytes/op: 5.4241e+06 miss_ratio: 0.324881 max_rss_mb: 1104.41
34MB 128thread new_clock -> kops/s: 10.519 io_bytes/op: 2.39354e+06 miss_ratio: 0.136147 max_rss_mb: 1050.52
As the miss ratio gets higher (say, above 10%), the CPU time spent in eviction starts to erode the advantage of using fewer shards (13% miss rate much lower than 94%). LRU's O(1) eviction time can eventually pay off when there's enough block cache churn:
13MB 1thread base -> kops/s: 0.195 io_bytes/op: 1.65732e+07 miss_ratio: 0.946604 max_rss_mb: 45.6328
13MB 1thread folly -> kops/s: 0.197 io_bytes/op: 1.63793e+07 miss_ratio: 0.94661 max_rss_mb: 33.8633
13MB 1thread gt_clock -> kops/s: 0.519 io_bytes/op: 4.43316e+06 miss_ratio: 0.269379 max_rss_mb: 100.684
13MB 1thread new_clock -> kops/s: 0.176 io_bytes/op: 1.54148e+07 miss_ratio: 0.91545 max_rss_mb: 66.2383
13MB 32thread base -> kops/s: 3.266 io_bytes/op: 1.65544e+07 miss_ratio: 0.943386 max_rss_mb: 132.492
13MB 32thread folly -> kops/s: 3.396 io_bytes/op: 1.63142e+07 miss_ratio: 0.943243 max_rss_mb: 101.863
13MB 32thread gt_clock -> kops/s: 2.758 io_bytes/op: 5.13714e+06 miss_ratio: 0.310652 max_rss_mb: 396.121
13MB 32thread new_clock -> kops/s: 3.11 io_bytes/op: 1.23419e+07 miss_ratio: 0.708425 max_rss_mb: 321.758
13MB 128thread base -> kops/s: 2.31 io_bytes/op: 1.64823e+07 miss_ratio: 0.939543 max_rss_mb: 425.539
13MB 128thread folly -> kops/s: 2.339 io_bytes/op: 1.6242e+07 miss_ratio: 0.939966 max_rss_mb: 346.098
13MB 128thread gt_clock -> kops/s: 3.223 io_bytes/op: 5.76928e+06 miss_ratio: 0.345899 max_rss_mb: 1087.77
13MB 128thread new_clock -> kops/s: 2.984 io_bytes/op: 1.05341e+07 miss_ratio: 0.606198 max_rss_mb: 898.27
gt_clock is clearly blowing way past its memory budget for lower miss rates and best throughput. new_clock also seems to be exceeding budgets, and this warrants more investigation but is not the use case we are targeting with the new cache. With partitioned index+filter, the miss ratio is much better, and although still high enough that the eviction CPU time is definitely offsetting mutex contention:
13MB 1thread base -> kops/s: 16.326 io_bytes/op: 23743.9 miss_ratio: 0.205362 max_rss_mb: 65.2852
13MB 1thread folly -> kops/s: 15.574 io_bytes/op: 19415 miss_ratio: 0.184157 max_rss_mb: 56.3516
13MB 1thread gt_clock -> kops/s: 14.459 io_bytes/op: 22873 miss_ratio: 0.198355 max_rss_mb: 63.9688
13MB 1thread new_clock -> kops/s: 16.34 io_bytes/op: 24386.5 miss_ratio: 0.210512 max_rss_mb: 61.707
13MB 128thread base -> kops/s: 289.786 io_bytes/op: 23710.9 miss_ratio: 0.205056 max_rss_mb: 103.57
13MB 128thread folly -> kops/s: 185.282 io_bytes/op: 19433.1 miss_ratio: 0.184275 max_rss_mb: 116.219
13MB 128thread gt_clock -> kops/s: 354.451 io_bytes/op: 23150.6 miss_ratio: 0.200495 max_rss_mb: 102.871
13MB 128thread new_clock -> kops/s: 295.359 io_bytes/op: 24626.4 miss_ratio: 0.212452 max_rss_mb: 121.109
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10626
Test Plan: updated unit tests, stress/crash test runs including with TSAN, ASAN, UBSAN
Reviewed By: anand1976
Differential Revision: D39368406
Pulled By: pdillinger
fbshipit-source-id: 5afc44da4c656f8f751b44552bbf27bd3ca6fef9
Summary:
Although we've been tracking SST unique IDs in the DB manifest
unconditionally, checking has been opt-in and with an extra pass at DB::Open
time. This changes the behavior of `verify_sst_unique_id_in_manifest` to
check unique ID against manifest every time an SST file is opened through
table cache (normal DB operations), replacing the explicit pass over files
at DB::Open time. This change also enables the option by default and
removes the "EXPERIMENTAL" designation.
One possible criticism is that the option no longer ensures the integrity
of a DB at Open time. This is far from an all-or-nothing issue. Verifying
the IDs of all SST files hardly ensures all the data in the DB is readable.
(VerifyChecksum is supposed to do that.) Also, with
max_open_files=-1 (default, extremely common), all SST files are
opened at DB::Open time anyway.
Implementation details:
* `VerifySstUniqueIdInManifest()` functions are the extra/explicit pass
that is now removed.
* Unit tests that manipulate/corrupt table properties have to opt out of
this check, because that corrupts the "actual" unique id. (And even for
testing we don't currently have a mechanism to set "no unique id"
in the in-memory file metadata for new files.)
* A lot of other unit test churn relates to (a) default checking on, and
(b) checking on SST open even without DB::Open (e.g. on flush)
* Use `FileMetaData` for more `TableCache` operations (in place of
`FileDescriptor`) so that we have access to the unique_id whenever
we might need to open an SST file. **There is the possibility of
performance impact because we can no longer use the more
localized `fd` part of an `FdWithKeyRange` but instead follow the
`file_metadata` pointer. However, this change (possible regression)
is only done for `GetMemoryUsageByTableReaders`.**
* Removed a completely unnecessary constructor overload of
`TableReaderOptions`
Possible follow-up:
* Verification only happens when opening through table cache. Are there
more places where this should happen?
* Improve error message when there is a file size mismatch vs. manifest
(FIXME added in the appropriate place).
* I'm not sure there's a justification for `FileDescriptor` to be distinct from
`FileMetaData`.
* I'm skeptical that `FdWithKeyRange` really still makes sense for
optimizing some data locality by duplicating some data in memory, but I
could be wrong.
* An unnecessary overload of NewTableReader was recently added, in
the public API nonetheless (though unusable there). It should be cleaned
up to put most things under `TableReaderOptions`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10532
Test Plan:
updated unit tests
Performance test showing no significant difference (just noise I think):
`./db_bench -benchmarks=readwhilewriting[-X10] -num=3000000 -disable_wal=1 -bloom_bits=8 -write_buffer_size=1000000 -target_file_size_base=1000000`
Before: readwhilewriting [AVG 10 runs] : 68702 (± 6932) ops/sec
After: readwhilewriting [AVG 10 runs] : 68239 (± 7198) ops/sec
Reviewed By: jay-zhuang
Differential Revision: D38765551
Pulled By: pdillinger
fbshipit-source-id: a827a708155f12344ab2a5c16e7701c7636da4c2
Summary:
Previously, automatic compaction could be triggered prior to the test invoking CompactRange(). It could lead to the following flaky failure:
```
/root/project/db/db_block_cache_test.cc:753: Failure
Expected equality of these values:
1 + kNumBlocks
Which is: 11
options.statistics->getTickerCount(BLOCK_CACHE_INDEX_ADD)
Which is: 10
```
A sequence leading to this failure was:
* Automatic compaction
* files [1] [2] trivially moved
* files [3] [4] [5] [6] trivially moved
* CompactRange()
* files [7] [8] [9] trivially moved
* file [10] trivially moved
In such a case, the index/filter block adds that the test expected did not happen since there were no new files.
This PR just tweaks settings to ensure the `CompactRange()` produces one new file.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10635
Reviewed By: cbi42
Differential Revision: D39250869
Pulled By: ajkr
fbshipit-source-id: a3c94c49069e28c49c40b4b80dae0059739d19fd
Summary:
RocksDB's `Cache` abstraction currently supports two priority levels for items: high (used for frequently accessed/highly valuable SST metablocks like index/filter blocks) and low (used for SST data blocks). Blobs are typically lower-value targets for caching than data blocks, since 1) with BlobDB, data blocks containing blob references conceptually form an index structure which has to be consulted before we can read the blob value, and 2) cached blobs represent only a single key-value, while cached data blocks generally contain multiple KVs. Since we would like to make it possible to use the same backing cache for the block cache and the blob cache, it would make sense to add a new, lower-than-low cache priority level (bottom level) for blobs so data blocks are prioritized over them.
This task is a part of https://github.com/facebook/rocksdb/issues/10156
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10461
Reviewed By: siying
Differential Revision: D38672823
Pulled By: ltamasi
fbshipit-source-id: 90cf7362036563d79891f47be2cc24b827482743
Summary:
... so that cache keys can be derived from DB manifest data
before reading the file from storage--so that every part of the file
can potentially go in a persistent cache.
See updated comments in cache_key.cc for technical details. Importantly,
the new cache key encoding uses some fancy but efficient math to pack
data into the cache key without depending on the sizes of the various
pieces. This simplifies some existing code creating cache keys, like
cache warming before the file size is known.
This should provide us an essentially permanent mapping between SST
unique IDs and base cache keys, with the ability to "upgrade" SST
unique IDs (and thus cache keys) with new SST format_versions.
These cache keys are of similar, perhaps indistinguishable quality to
the previous generation. Before this change (see "corrected" days
between collision):
```
./cache_bench -stress_cache_key -sck_keep_bits=43
18 collisions after 2 x 90 days, est 10 days between (1.15292e+19 corrected)
```
After this change (keep 43 bits, up through 50, to validate "trajectory"
is ok on "corrected" days between collision):
```
19 collisions after 3 x 90 days, est 14.2105 days between (1.63836e+19 corrected)
16 collisions after 5 x 90 days, est 28.125 days between (1.6213e+19 corrected)
15 collisions after 7 x 90 days, est 42 days between (1.21057e+19 corrected)
15 collisions after 17 x 90 days, est 102 days between (1.46997e+19 corrected)
15 collisions after 49 x 90 days, est 294 days between (2.11849e+19 corrected)
15 collisions after 62 x 90 days, est 372 days between (1.34027e+19 corrected)
15 collisions after 53 x 90 days, est 318 days between (5.72858e+18 corrected)
15 collisions after 309 x 90 days, est 1854 days between (1.66994e+19 corrected)
```
However, the change does modify (probably weaken) the "guaranteed unique" promise from this
> SST files generated in a single process are guaranteed to have unique cache keys, unless/until number session ids * max file number = 2**86
to this (see https://github.com/facebook/rocksdb/issues/10388)
> With the DB id limitation, we only have nice guaranteed unique cache keys for files generated in a single process until biggest session_id_counter and offset_in_file reach combined 64 bits
I don't think this is a practical concern, though.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10394
Test Plan: unit tests updated, see simulation results above
Reviewed By: jay-zhuang
Differential Revision: D38667529
Pulled By: pdillinger
fbshipit-source-id: 49af3fe7f47e5b61162809a78b76c769fd519fba
Summary:
Moved linux builds to using docker to avoid CI instability caused by dependency installation site down.
Added the `Dockerfile` which is used to build the image.
The build time is also significantly reduced, because no dependencies installation and with using 2xlarge+ instance for slow build (like tsan test).
Also fixed a few issues detected while building this:
* `DestoryDB()` Status not checked for a few tests
* nullptr might be used in `inlineskiplist.cc`
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10496
Test Plan: CI
Reviewed By: ajkr
Differential Revision: D38554200
Pulled By: jay-zhuang
fbshipit-source-id: 16e8fb2bf07b9c84bb27fb18421c4d54f2f248fd
Summary:
A test in db_block_cache_test.cc was skipping ClockCache due to the 16-byte key length requirement. We fixed this. Along the way, we fixed a bug in ApplyToSomeEntries, which assumed the function being applied could modify handle metadata, and thus took an exclusive reference. This is incompatible with calls that need to inspect every element (including externally referenced ones) to gather stats.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10482
Test Plan: ``make -j24 check``
Reviewed By: anand1976
Differential Revision: D38553073
Pulled By: guidotag
fbshipit-source-id: 0ed63fed4d3b89e5056b35b7091fce579f5647ae
Summary:
RocksDB's `Cache` abstraction currently supports two priority levels for items: high (used for frequently accessed/highly valuable SST metablocks like index/filter blocks) and low (used for SST data blocks). Blobs are typically lower-value targets for caching than data blocks, since 1) with BlobDB, data blocks containing blob references conceptually form an index structure which has to be consulted before we can read the blob value, and 2) cached blobs represent only a single key-value, while cached data blocks generally contain multiple KVs. Since we would like to make it possible to use the same backing cache for the block cache and the blob cache, it would make sense to add a new, lower-than-low cache priority level (bottom level) for blobs so data blocks are prioritized over them.
This task is a part of https://github.com/facebook/rocksdb/issues/10156
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10309
Reviewed By: ltamasi
Differential Revision: D38211655
Pulled By: gangliao
fbshipit-source-id: 65ef33337db4d85277cc6f9782d67c421ad71dd5
Summary:
ClockCache is still in experimental stage, and currently fails some pre-release fbcode tests. See https://www.internalfb.com/diff/D37772011. API calls to construct ClockCache are done via the function NewClockCache. For now, NewClockCache calls will return an LRUCache (with appropriate arguments), which is stable.
The idea that NewClockCache returns nullptr was also floated, but this would be interpreted as unsupported cache, and a default LRUCache would be constructed instead, potentially causing a performance regression that is harder to identify.
A new version of the NewClockCache function was created for our internal tests.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10351
Test Plan: ``make -j24 check`` and re-run the pre-release tests.
Reviewed By: pdillinger
Differential Revision: D37802685
Pulled By: guidotag
fbshipit-source-id: 0a8d10612ff21e576f7360cb13e20bc36e244972
Summary:
I noticed it would clean up some things to have Cache::Insert()
return our MemoryLimit Status instead of Incomplete for the case in
which the capacity limit is reached. I suspect this fixes some existing but
unknown bugs where this Incomplete could be confused with other uses
of Incomplete, especially no_io cases. This is the most suspicious case I
noticed, but was not able to reproduce a bug, in part because the existing
code is not covered by unit tests (FIXME added): 57adbf0e91/table/get_context.cc (L397)
I audited all the existing uses of IsIncomplete and updated those that
seemed relevant.
HISTORY updated with a clear warning to users of strict_capacity_limit=true
to update uses of `IsIncomplete()`
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10262
Test Plan: updated unit tests
Reviewed By: hx235
Differential Revision: D37473155
Pulled By: pdillinger
fbshipit-source-id: 4bd9d9353ccddfe286b03ebd0652df8ce20f99cb
Summary:
In leveled compaction, try to trivial move more than one files if possible, up to 4 files or max_compaction_bytes. This is to allow higher write throughput for some use cases where data is loaded in sequential order, where appying compaction results is the bottleneck.
When pick up a file to compact and it doesn't have overlapping files in the next level, try to expand to the next file if there is still no overlapping.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10190
Test Plan:
Add some unit tests.
For performance, Try to run
./db_bench_multi_move --benchmarks=fillseq --compression_type=lz4 --write_buffer_size=5000000 --num=100000000 --value_size=1000 -level_compaction_dynamic_level_bytes
Together with https://github.com/facebook/rocksdb/pull/10188 , stalling will be eliminated in this benchmark.
Reviewed By: jay-zhuang
Differential Revision: D37230647
fbshipit-source-id: 42b260f545c46abc5d90335ac2bbfcd09602b549
Summary:
This is the initial step in the development of a lock-free clock cache. This PR includes the base hash table design (which we mostly ported over from FastLRUCache) and the clock eviction algorithm. Importantly, it's still _not_ lock-free---all operations use a shard lock. Besides the locking, there are other features left as future work:
- Remove keys from the handles. Instead, use 128-bit bijective hashes of them for handle comparisons, probing (we need two 32-bit hashes of the key for double hashing) and sharding (we need one 6-bit hash).
- Remove the clock_usage_ field, which is updated on every lookup. Even if it were atomically updated, it could cause memory invalidations across cores.
- Middle insertions into the clock list.
- A test that exercises the clock eviction policy.
- Update the Java API of ClockCache and Java calls to C++.
Along the way, we improved the code and comments quality of FastLRUCache. These changes are relatively minor.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10273
Test Plan: ``make -j24 check``
Reviewed By: pdillinger
Differential Revision: D37522461
Pulled By: guidotag
fbshipit-source-id: 3d70b737dbb70dcf662f00cef8c609750f083943
Summary:
In https://github.com/facebook/rocksdb/issues/9535, release 7.0, we hid the old block-based filter from being created using
the public API, because of its inefficiency. Although we normally maintain read compatibility
on old DBs forever, filters are not required for reading a DB, only for optimizing read
performance. Thus, it should be acceptable to remove this code and the substantial
maintenance burden it carries as useful features are developed and validated (such
as user timestamp).
This change completely removes the code for reading and writing the old block-based
filters, net removing about 1370 lines of code no longer needed. Options removed from
testing / benchmarking tools. The prior existence is only evident in a couple of places:
* `CacheEntryRole::kDeprecatedFilterBlock` - We can update this public API enum in
a major release to minimize source code incompatibilities.
* A warning is logged when an old table file is opened that used the old block-based
filter. This is provided as a courtesy, and would be a pain to unit test, so manual testing
should suffice. Unfortunately, sst_dump does not tell you whether a file uses
block-based filter, and the structure of the code makes it very difficult to fix.
* To detect that case, `kObsoleteFilterBlockPrefix` (renamed from `kFilterBlockPrefix`)
for metaindex is maintained (for now).
Other notes:
* In some cases where numbers are associated with filter configurations, we have had to
update the assigned numbers so that they all correspond to something that exists.
* Fixed potential stat counting bug by assuming `filter_checked = false` for cases
like `filter == nullptr` rather than assuming `filter_checked = true`
* Removed obsolete `block_offset` and `prefix_extractor` parameters from several
functions.
* Removed some unnecessary checks `if (!table_prefix_extractor() && !prefix_extractor)`
because the caller guarantees the prefix extractor exists and is compatible
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10184
Test Plan:
tests updated, manually test new warning in LOG using base version to
generate a DB
Reviewed By: riversand963
Differential Revision: D37212647
Pulled By: pdillinger
fbshipit-source-id: 06ee020d8de3b81260ffc36ad0c1202cbf463a80
Summary:
We make the size of the per-shard hash table fixed. The base level of the hash table is now preallocated with the required capacity. The user must provide an estimate of the size of the values.
Notice that even though the base level becomes fixed, the chains are still dynamic. Overall, the shard capacity mechanisms haven't changed, so we don't need to test this.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10154
Test Plan: `make -j24 check`
Reviewed By: pdillinger
Differential Revision: D37124451
Pulled By: guidotag
fbshipit-source-id: cba6ac76052fe0ec60b8ff4211b3de7650e80d0c
Summary:
FastLRUCache now only supports 16B keys. The tests have changed to reflect this.
Because the unit tests were designed for caches that accept any string as keys, some tests are no longer compatible with FastLRUCache. We have disabled those for runs with FastLRUCache. (We could potentially change all tests to use 16B keys, but we don't because the cache public API does not require this.)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10137
Test Plan: make -j24 check
Reviewed By: gitbw95
Differential Revision: D37083934
Pulled By: guidotag
fbshipit-source-id: be1719cf5f8364a9a32bc4555bce1a0de3833b0d
Summary:
ToString() is created as some platform doesn't support std::to_string(). However, we've already used std::to_string() by mistake for 16 months (in db/db_info_dumper.cc). This commit just remove ToString().
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9955
Test Plan: Watch CI tests
Reviewed By: riversand963
Differential Revision: D36176799
fbshipit-source-id: bdb6dcd0e3a3ab96a1ac810f5d0188f684064471
Summary:
To support a project to prototype and evaluate algorithmic
enhancments and alternatives to LRUCache, here I have separated out
LRUCache into internal-only "FastLRUCache" and cut it down to
essentials, so that details like secondary cache handling and
priorities do not interfere with prototyping. These can be
re-integrated later as needed, along with refactoring to minimize code
duplication (which would slow down prototyping for now).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9917
Test Plan:
unit tests updated to ensure basic functionality has (likely)
been preserved
Reviewed By: anand1976
Differential Revision: D35995554
Pulled By: pdillinger
fbshipit-source-id: d67b20b7ada3b5d3bfe56d897a73885894a1d9db
Summary:
This gives users the ability to examine the map populated by `GetMapProperty()` with property `kBlockCacheEntryStats`. It also sets us up for a possible future where cache reservations are configured according to `CacheEntryRole`s rather than flags coupled to roles.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9838
Test Plan:
- migrated test DBBlockCacheTest.CacheEntryRoleStats to use this API. That test verifies some of the contents are as expected
- added a DBPropertiesTest to verify the public map keys are present, and nothing else
Reviewed By: hx235
Differential Revision: D35629493
Pulled By: ajkr
fbshipit-source-id: 5c4356b8560e85d1f881fd32c44c15960b02fc68
Summary:
Follow-up to https://github.com/facebook/rocksdb/issues/9126
Added new unit tests to validate some of the claims of guaranteed uniqueness
within certain large bounds.
Also cleaned up the cache_bench -stress-cache-key tool with better comments
and description.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9329
Test Plan: no changes to production code
Reviewed By: mrambacher
Differential Revision: D33269328
Pulled By: pdillinger
fbshipit-source-id: 3a2b684a6b2b15f79dc872e563e3d16563be26de
Summary:
This change standardizes on a new 16-byte cache key format for
block cache (incl compressed and secondary) and persistent cache (but
not table cache and row cache).
The goal is a really fast cache key with practically ideal stability and
uniqueness properties without external dependencies (e.g. from FileSystem).
A fixed key size of 16 bytes should enable future optimizations to the
concurrent hash table for block cache, which is a heavy CPU user /
bottleneck, but there appears to be measurable performance improvement
even with no changes to LRUCache.
This change replaces a lot of disjointed and ugly code handling cache
keys with calls to a simple, clean new internal API (cache_key.h).
(Preserving the old cache key logic under an option would be very ugly
and likely negate the performance gain of the new approach. Complete
replacement carries some inherent risk, but I think that's acceptable
with sufficient analysis and testing.)
The scheme for encoding new cache keys is complicated but explained
in cache_key.cc.
Also: EndianSwapValue is moved to math.h to be next to other bit
operations. (Explains some new include "math.h".) ReverseBits operation
added and unit tests added to hash_test for both.
Fixes https://github.com/facebook/rocksdb/issues/7405 (presuming a root cause)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9126
Test Plan:
### Basic correctness
Several tests needed updates to work with the new functionality, mostly
because we are no longer relying on filesystem for stable cache keys
so table builders & readers need more context info to agree on cache
keys. This functionality is so core, a huge number of existing tests
exercise the cache key functionality.
### Performance
Create db with
`TEST_TMPDIR=/dev/shm ./db_bench -bloom_bits=10 -benchmarks=fillrandom -num=3000000 -partition_index_and_filters`
And test performance with
`TEST_TMPDIR=/dev/shm ./db_bench -readonly -use_existing_db -bloom_bits=10 -benchmarks=readrandom -num=3000000 -duration=30 -cache_index_and_filter_blocks -cache_size=250000 -threads=4`
using DEBUG_LEVEL=0 and simultaneous before & after runs.
Before ops/sec, avg over 100 runs: 121924
After ops/sec, avg over 100 runs: 125385 (+2.8%)
### Collision probability
I have built a tool, ./cache_bench -stress_cache_key to broadly simulate host-wide cache activity
over many months, by making some pessimistic simplifying assumptions:
* Every generated file has a cache entry for every byte offset in the file (contiguous range of cache keys)
* All of every file is cached for its entire lifetime
We use a simple table with skewed address assignment and replacement on address collision
to simulate files coming & going, with quite a variance (super-Poisson) in ages. Some output
with `./cache_bench -stress_cache_key -sck_keep_bits=40`:
```
Total cache or DBs size: 32TiB Writing 925.926 MiB/s or 76.2939TiB/day
Multiply by 9.22337e+18 to correct for simulation losses (but still assume whole file cached)
```
These come from default settings of 2.5M files per day of 32 MB each, and
`-sck_keep_bits=40` means that to represent a single file, we are only keeping 40 bits of
the 128-bit cache key. With file size of 2\*\*25 contiguous keys (pessimistic), our simulation
is about 2\*\*(128-40-25) or about 9 billion billion times more prone to collision than reality.
More default assumptions, relatively pessimistic:
* 100 DBs in same process (doesn't matter much)
* Re-open DB in same process (new session ID related to old session ID) on average
every 100 files generated
* Restart process (all new session IDs unrelated to old) 24 times per day
After enough data, we get a result at the end:
```
(keep 40 bits) 17 collisions after 2 x 90 days, est 10.5882 days between (9.76592e+19 corrected)
```
If we believe the (pessimistic) simulation and the mathematical generalization, we would need to run a billion machines all for 97 billion days to expect a cache key collision. To help verify that our generalization ("corrected") is robust, we can make our simulation more precise with `-sck_keep_bits=41` and `42`, which takes more running time to get enough data:
```
(keep 41 bits) 16 collisions after 4 x 90 days, est 22.5 days between (1.03763e+20 corrected)
(keep 42 bits) 19 collisions after 10 x 90 days, est 47.3684 days between (1.09224e+20 corrected)
```
The generalized prediction still holds. With the `-sck_randomize` option, we can see that we are beating "random" cache keys (except offsets still non-randomized) by a modest amount (roughly 20x less collision prone than random), which should make us reasonably comfortable even in "degenerate" cases:
```
197 collisions after 1 x 90 days, est 0.456853 days between (4.21372e+18 corrected)
```
I've run other tests to validate other conditions behave as expected, never behaving "worse than random" unless we start chopping off structured data.
Reviewed By: zhichao-cao
Differential Revision: D33171746
Pulled By: pdillinger
fbshipit-source-id: f16a57e369ed37be5e7e33525ace848d0537c88f
Summary:
I'm working on a new format_version=6 to support context
checksum (https://github.com/facebook/rocksdb/issues/9058) and this includes much of the refactoring and test
updates to support that change.
Test coverage data and manual inspection agree on dead code in
block_based_table_reader.cc (removed).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9240
Test Plan:
tests enhanced to cover more cases etc.
Extreme case performance testing indicates small % regression in fillseq (w/ compaction), though CPU profile etc. doesn't suggest any explanation. There is enhanced correctness checking in Footer::DecodeFrom, but this should be negligible.
TEST_TMPDIR=/dev/shm/ ./db_bench -benchmarks=fillseq -memtablerep=vector -allow_concurrent_memtable_write=false -num=30000000 -checksum_type=1 --disable_wal={false,true}
(Each is ops/s averaged over 50 runs, run simultaneously with competing configuration for load fairness)
Before w/ wal: 454512
After w/ wal: 444820 (-2.1%)
Before w/o wal: 1004560
After w/o wal: 998897 (-0.6%)
Since this doesn't modify WAL code, one would expect real effects to be larger in w/o wal case.
This regression will be corrected in a follow-up PR.
Reviewed By: ajkr
Differential Revision: D32813769
Pulled By: pdillinger
fbshipit-source-id: 444a244eabf3825cd329b7d1b150cddce320862f
Summary:
When table_options.prepopulate_block_cache is set to
BlockBasedTableOptions::PrepopulateBlockCache::kFlushOnly and
table_options.partition_filters is also set true, then there is
segmentation failure when top level filter is fetched because its
entered with wrong type in cache.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9263
Test Plan:
Updated unit tests;
Ran db_stress: make crash_test -j32
Reviewed By: pdillinger
Differential Revision: D32936566
Pulled By: akankshamahajan15
fbshipit-source-id: 8bd79e53830d3e3c1bb79787e1ffbc3cb46d4426
Summary:
We have three layers of block cache that often use the same key
but map to different physical data:
* BlockBasedTableOptions::block_cache
* BlockBasedTableOptions::block_cache_compressed
* BlockBasedTableOptions::persistent_cache
If any two of these happen to share an underlying implementation and key
space (insertion into one shows up in another), then memory safety is
broken. The simplest case is block_cache == block_cache_compressed.
(Credit mrambacher for asking about this case in a review.)
With this change, we explicitly check for overlap and preemptively and
safely fail with a Status code.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9172
Test Plan: test added. Crashes without new check
Reviewed By: anand1976
Differential Revision: D32465659
Pulled By: pdillinger
fbshipit-source-id: 3876b45b6dce6167e5a7a642725ddc86b96f8e40
Summary:
kFlushOnly currently means "always" except in the case of
remote compaction. This makes it flushes only.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8750
Test Plan: test updated
Reviewed By: akankshamahajan15
Differential Revision: D30968034
Pulled By: pdillinger
fbshipit-source-id: 5dbd24dde18852a0e937a540995fba9bfbe89037
Summary:
I very recently realized that with https://github.com/facebook/rocksdb/issues/8669 we cannot later add
file numbers to external SST files (so that more can share db session
ids for better uniqueness properties), because of forward compatibility.
We would have a version of RocksDB that assumes session IDs are unique
on external SST files and therefore can't really break that invariant in
future files.
This change adds a table property for "orig_file_number" which is
populated by normal SST files and also external SST files generated by
SstFileWriter. SstFileWriter now keeps a db_session_id for life of the
object and increments its own file numbers for embedding in table
properties. (They are arguably "fake" file numbers because these numbers
and not embedded in the file name.)
While updating block_based_table_builder, I removed several unnecessary
fields from Rep, because following the pattern would have created
another unnecessary field.
This change also updates block_based_table_reader to use this new
property when available, which means that for newer SST files, we can
determine the stable/original <db_session_id,file_number> unique
identifier using just the file contents, not the file name. (It's a bit
complicated; detailed comments in block_based_table_reader.)
Also added DB host id to properties listing by sst_dump, which could be
useful in debugging.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8686
Test Plan: majorly overhauled StableCacheKeys test for this change
Reviewed By: zhichao-cao
Differential Revision: D30457742
Pulled By: pdillinger
fbshipit-source-id: 2e5ae7dddeb94fb9d8eac8a928486aed8b8cd445
Summary:
Extends https://github.com/facebook/rocksdb/issues/8659 to work for ingested external SST files, even
the same file ingested into different DBs sharing a block cache.
Note: These new cache keys are currently only enabled when FileSystem
does not provide GetUniqueId. For now, they are typically larger,
so slightly less efficient.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8669
Test Plan: Extended unit test
Reviewed By: zhichao-cao
Differential Revision: D30398532
Pulled By: pdillinger
fbshipit-source-id: 1f13e2af4b8bfff5741953a69466e9589fbc23c7
Summary:
Use DB session ids in SST table properties to make cache keys
stable across DB re-open and copy / move / restore / etc.
These new cache keys are currently only enabled when FileSystem does not
provide GetUniqueId. For now, they are typically larger, so slightly
less efficient.
Relevant to https://github.com/facebook/rocksdb/issues/7405
This change has a minor regression in PersistentCache functionality:
metaindex blocks are no longer cached in PersistentCache. Table properties
blocks already were not but ideally should be. I didn't spent effort to
fix & test these issues because we don't believe PersistentCache is used much
if at all and expect SecondaryCache to replace it. (Though PRs are welcome.)
FIXME: there is more to be fixed for stable cache keys on external SST files
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8659
Test Plan:
new unit test added, which fails when disabling new
functionality
Reviewed By: zhichao-cao
Differential Revision: D30297705
Pulled By: pdillinger
fbshipit-source-id: e8539a5c8802a79340405629870f2e3fb3822d3a
Summary:
Insert warm blocks (data, uncompressed dict, index and filter blocks) during flush in Block cache which is enabled under option BlockBasedTableOptions.prepopulate_block_cache.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8561
Test Plan: Added unit test
Reviewed By: anand1976
Differential Revision: D29773411
Pulled By: akankshamahajan15
fbshipit-source-id: 6631123c10134340ef0bd7e90baafaa6deba0e66
Summary:
The PerThreadDBPath has already specified a slash. It does not need to be specified when initializing the test path.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8555
Reviewed By: ajkr
Differential Revision: D29758399
Pulled By: jay-zhuang
fbshipit-source-id: 6d2b878523e3e8580536e2829cb25489844d9011
Summary:
I previously didn't notice the DB mutex was being held during
block cache entry stat scans, probably because I primarily checked for
read performance regressions, because they require the block cache and
are traditionally latency-sensitive.
This change does some refactoring to avoid holding DB mutex and to
avoid triggering and waiting for a scan in GetProperty("rocksdb.cfstats").
Some tests have to be updated because now the stats collector is
populated in the Cache aggressively on DB startup rather than lazily.
(I hope to clean up some of this added complexity in the future.)
This change also ensures proper treatment of need_out_of_mutex for
non-int DB properties.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8538
Test Plan:
Added unit test logic that uses sync points to fail if the DB mutex
is held during a scan, covering the various ways that a scan might be
triggered.
Performance test - the known impact to holding the DB mutex is on
TransactionDB, and the easiest way to see the impact is to hack the
scan code to almost always miss and take an artificially long time
scanning. Here I've injected an unconditional 5s sleep at the call to
ApplyToAllEntries.
Before (hacked):
$ TEST_TMPDIR=/dev/shm ./db_bench.base_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 433.219 micros/op 2308 ops/sec; 0.1 MB/s ( transactions:78999 aborts:0)
rocksdb.db.write.micros P50 : 16.135883 P95 : 36.622503 P99 : 66.036115 P100 : 5000614.000000 COUNT : 149677 SUM : 8364856
$ TEST_TMPDIR=/dev/shm ./db_bench.base_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 448.802 micros/op 2228 ops/sec; 0.1 MB/s ( transactions:75999 aborts:0)
rocksdb.db.write.micros P50 : 16.629221 P95 : 37.320607 P99 : 72.144341 P100 : 5000871.000000 COUNT : 143995 SUM : 13472323
Notice the 5s P100 write time.
After (hacked):
$ TEST_TMPDIR=/dev/shm ./db_bench.new_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 303.645 micros/op 3293 ops/sec; 0.1 MB/s ( transactions:98999 aborts:0)
rocksdb.db.write.micros P50 : 16.061871 P95 : 33.978834 P99 : 60.018017 P100 : 616315.000000 COUNT : 187619 SUM : 4097407
$ TEST_TMPDIR=/dev/shm ./db_bench.new_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 310.383 micros/op 3221 ops/sec; 0.1 MB/s ( transactions:96999 aborts:0)
rocksdb.db.write.micros P50 : 16.270026 P95 : 35.786844 P99 : 64.302878 P100 : 603088.000000 COUNT : 183819 SUM : 4095918
P100 write is now ~0.6s. Not good, but it's the same even if I completely bypass all the scanning code:
$ TEST_TMPDIR=/dev/shm ./db_bench.new_skip -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 311.365 micros/op 3211 ops/sec; 0.1 MB/s ( transactions:96999 aborts:0)
rocksdb.db.write.micros P50 : 16.274362 P95 : 36.221184 P99 : 68.809783 P100 : 649808.000000 COUNT : 183819 SUM : 4156767
$ TEST_TMPDIR=/dev/shm ./db_bench.new_skip -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
randomtransaction : 308.395 micros/op 3242 ops/sec; 0.1 MB/s ( transactions:97999 aborts:0)
rocksdb.db.write.micros P50 : 16.106222 P95 : 37.202403 P99 : 67.081875 P100 : 598091.000000 COUNT : 185714 SUM : 4098832
No substantial difference.
Reviewed By: siying
Differential Revision: D29738847
Pulled By: pdillinger
fbshipit-source-id: 1c5c155f5a1b62e4fea0fd4eeb515a8b7474027b
Summary:
This PR is for https://github.com/facebook/rocksdb/issues/8453
We need to update `s = biter.status();` when `biter.status().IsIncomplete()` is true. By doing this, can fix the problem in issue.
Besides, we still need to update `db_statistics` in `get_context.ReportCounters()` before return back.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8485
Reviewed By: jay-zhuang
Differential Revision: D29604835
Pulled By: ajkr
fbshipit-source-id: c7f2f1cd058223ce1b507ec05d57cf264b9c9710
Summary:
This PR prepopulates warm/hot data blocks which are already in memory
into block cache at the time of flush. On a flush, the data block that is
in memory (in memtables) get flushed to the device. If using Direct IO,
additional IO is incurred to read this data back into memory again, which
is avoided by enabling newly added option.
Right now, this is enabled only for flush for data blocks. We plan to
expand this option to cover compactions in the future and for other types
of blocks.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8242
Test Plan: Add new unit test
Reviewed By: anand1976
Differential Revision: D28521703
Pulled By: akankshamahajan15
fbshipit-source-id: 7219d6958821cedce689a219c3963a6f1a9d5f05
Summary:
If the block Cache is full with strict_capacity_limit=false,
then our CacheEntryStatsCollector could be immediately evicted on
release, so iterating through column families with shared block cache
could trigger re-scan for each CF. This change fixes that problem by
pinning the CacheEntryStatsCollector from InternalStats so that it's not
evicted.
I had originally thought that this object could participate in LRU like
everything else, but even though a re-load+re-scan only touches memory,
it can be orders of magnitude more expensive than other cache misses.
One service in Facebook has scans that take ~20s over 100GB block cache
that is mostly 4KB entries. (The up-side of this bug and https://github.com/facebook/rocksdb/issues/8369 is that
we had a natural experiment on the effect on some service metrics even
with block cache scans running continuously in the background--a kind
of worst case scenario. Metrics like latency were not affected enough
to trigger warnings.)
Other smaller fixes:
20s is already a sizable portion of 600s stats dump period, or 180s
default max age to force re-scan, so added logic to ensure that (for
each block cache) we don't spend more than 0.2% of our background thread
time scanning it. Nevertheless, "foreground" requests for cache entry
stats (calls to `db->GetMapProperty(DB::Properties::kBlockCacheEntryStats)`)
are permitted to consume more CPU.
Renamed field to cache_entry_stats_ to match code style.
This change is intended for patching in 6.21 release.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8385
Test Plan:
unit test expanded to cover new logic (detect regression),
some manual testing with db_bench
Reviewed By: ajkr
Differential Revision: D29042759
Pulled By: pdillinger
fbshipit-source-id: 236faa902397f50038c618f50fbc8cf3f277308c
Summary:
In final polishing of https://github.com/facebook/rocksdb/issues/8297 (after most manual testing), I
broke my own caching layer by sanitizing an input parameter with
std::min(0, x) instead of std::max(0, x). I resisted unit testing the
timing part of the result caching because historically, these test
are either flaky or difficult to write, and this was not a correctness
issue. This bug is essentially unnoticeable with a small number
of column families but can explode background work with a
large number of column families.
This change fixes the logical error, removes some unnecessary related
optimization, and adds mock time/sleeps to the unit test to ensure we
can cache hit within the age limit.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8369
Test Plan: added time testing logic to existing unit test
Reviewed By: ajkr
Differential Revision: D28950892
Pulled By: pdillinger
fbshipit-source-id: e79cd4ff3eec68fd0119d994f1ed468c38026c3b
Summary:
This change gathers and publishes statistics about the
kinds of items in block cache. This is especially important for
profiling relative usage of cache by index vs. filter vs. data blocks.
It works by iterating over the cache during periodic stats dump
(InternalStats, stats_dump_period_sec) or on demand when
DB::Get(Map)Property(kBlockCacheEntryStats), except that for
efficiency and sharing among column families, saved data from
the last scan is used when the data is not considered too old.
The new information can be seen in info LOG, for example:
Block cache LRUCache@0x7fca62229330 capacity: 95.37 MB collections: 8 last_copies: 0 last_secs: 0.00178 secs_since: 0
Block cache entry stats(count,size,portion): DataBlock(7092,28.24 MB,29.6136%) FilterBlock(215,867.90 KB,0.888728%) FilterMetaBlock(2,5.31 KB,0.00544%) IndexBlock(217,180.11 KB,0.184432%) WriteBuffer(1,256.00 KB,0.262144%) Misc(1,0.00 KB,0%)
And also through DB::GetProperty and GetMapProperty (here using
ldb just for demonstration):
$ ./ldb --db=/dev/shm/dbbench/ get_property rocksdb.block-cache-entry-stats
rocksdb.block-cache-entry-stats.bytes.data-block: 0
rocksdb.block-cache-entry-stats.bytes.deprecated-filter-block: 0
rocksdb.block-cache-entry-stats.bytes.filter-block: 0
rocksdb.block-cache-entry-stats.bytes.filter-meta-block: 0
rocksdb.block-cache-entry-stats.bytes.index-block: 178992
rocksdb.block-cache-entry-stats.bytes.misc: 0
rocksdb.block-cache-entry-stats.bytes.other-block: 0
rocksdb.block-cache-entry-stats.bytes.write-buffer: 0
rocksdb.block-cache-entry-stats.capacity: 8388608
rocksdb.block-cache-entry-stats.count.data-block: 0
rocksdb.block-cache-entry-stats.count.deprecated-filter-block: 0
rocksdb.block-cache-entry-stats.count.filter-block: 0
rocksdb.block-cache-entry-stats.count.filter-meta-block: 0
rocksdb.block-cache-entry-stats.count.index-block: 215
rocksdb.block-cache-entry-stats.count.misc: 1
rocksdb.block-cache-entry-stats.count.other-block: 0
rocksdb.block-cache-entry-stats.count.write-buffer: 0
rocksdb.block-cache-entry-stats.id: LRUCache@0x7f3636661290
rocksdb.block-cache-entry-stats.percent.data-block: 0.000000
rocksdb.block-cache-entry-stats.percent.deprecated-filter-block: 0.000000
rocksdb.block-cache-entry-stats.percent.filter-block: 0.000000
rocksdb.block-cache-entry-stats.percent.filter-meta-block: 0.000000
rocksdb.block-cache-entry-stats.percent.index-block: 2.133751
rocksdb.block-cache-entry-stats.percent.misc: 0.000000
rocksdb.block-cache-entry-stats.percent.other-block: 0.000000
rocksdb.block-cache-entry-stats.percent.write-buffer: 0.000000
rocksdb.block-cache-entry-stats.secs_for_last_collection: 0.000052
rocksdb.block-cache-entry-stats.secs_since_last_collection: 0
Solution detail - We need some way to flag what kind of blocks each
entry belongs to, preferably without changing the Cache API.
One of the complications is that Cache is a general interface that could
have other users that don't adhere to whichever convention we decide
on for keys and values. Or we would pay for an extra field in the Handle
that would only be used for this purpose.
This change uses a back-door approach, the deleter, to indicate the
"role" of a Cache entry (in addition to the value type, implicitly).
This has the added benefit of ensuring proper code origin whenever we
recognize a particular role for a cache entry; if the entry came from
some other part of the code, it will use an unrecognized deleter, which
we simply attribute to the "Misc" role.
An internal API makes for simple instantiation and automatic
registration of Cache deleters for a given value type and "role".
Another internal API, CacheEntryStatsCollector, solves the problem of
caching the results of a scan and sharing them, to ensure scans are
neither excessive nor redundant so as not to harm Cache performance.
Because code is added to BlocklikeTraits, it is pulled out of
block_based_table_reader.cc into its own file.
This is a reformulation of https://github.com/facebook/rocksdb/issues/8276, without the type checking option
(could still be added), and with actual stat gathering.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8297
Test Plan: manual testing with db_bench, and a couple of basic unit tests
Reviewed By: ltamasi
Differential Revision: D28488721
Pulled By: pdillinger
fbshipit-source-id: 472f524a9691b5afb107934be2d41d84f2b129fb
Summary:
Defined the abstract interface for a secondary cache in include/rocksdb/secondary_cache.h, and updated LRUCacheOptions to take a std::shared_ptr<SecondaryCache>. An item is initially inserted into the LRU (primary) cache. When it ages out and evicted from memory, its inserted into the secondary cache. On a LRU cache miss and successful lookup in the secondary cache, the item is promoted to the LRU cache. Only support synchronous lookup currently. The secondary cache would be used to implement a persistent (flash cache) or compressed cache.
Tests:
Results from cache_bench and db_bench don't show any regression due to these changes.
cache_bench results before and after this change -
Command
```./cache_bench -ops_per_thread=10000000 -threads=1```
Before
```Complete in 40.688 s; QPS = 245774```
```Complete in 40.486 s; QPS = 246996```
```Complete in 42.019 s; QPS = 237989```
After
```Complete in 40.672 s; QPS = 245869```
```Complete in 44.622 s; QPS = 224107```
```Complete in 42.445 s; QPS = 235599```
db_bench results before this change, and with this change + https://github.com/facebook/rocksdb/issues/8213 and https://github.com/facebook/rocksdb/issues/8191 -
Commands
```./db_bench --benchmarks="fillseq,compact" -num=30000000 -key_size=32 -value_size=256 -use_direct_io_for_flush_and_compaction=true -db=/home/anand76/nvm_cache/db -partition_index_and_filters=true```
```./db_bench -db=/home/anand76/nvm_cache/db -use_existing_db=true -benchmarks=readrandom -num=30000000 -key_size=32 -value_size=256 -use_direct_reads=true -cache_size=1073741824 -cache_numshardbits=6 -cache_index_and_filter_blocks=true -read_random_exp_range=17 -statistics -partition_index_and_filters=true -threads=16 -duration=300```
Before
```
DB path: [/home/anand76/nvm_cache/db]
readrandom : 80.702 micros/op 198104 ops/sec; 54.4 MB/s (3708999 of 3708999 found)
```
```
DB path: [/home/anand76/nvm_cache/db]
readrandom : 87.124 micros/op 183625 ops/sec; 50.4 MB/s (3439999 of 3439999 found)
```
After
```
DB path: [/home/anand76/nvm_cache/db]
readrandom : 77.653 micros/op 206025 ops/sec; 56.6 MB/s (3866999 of 3866999 found)
```
```
DB path: [/home/anand76/nvm_cache/db]
readrandom : 84.962 micros/op 188299 ops/sec; 51.7 MB/s (3535999 of 3535999 found)
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8271
Reviewed By: zhichao-cao
Differential Revision: D28357511
Pulled By: anand1976
fbshipit-source-id: d1cfa236f00e649a18c53328be10a8062a4b6da2
Summary:
In dictionary compression's initial implementation, in order to save CPU overhead, we only enabled it
for bottom level under the assumption that the vast majority of data is
stored there. At that time, there was no
such thing as `ColumnFamilyOptions::bottommost_compression_opts`, so we just
hardcoded disabling dictionary compression in flush and compactions to
non-bottommost level. Now, we have users who generate all their files
through flush and are considering using dictionary compression.
To support such a use case, this PR expands the scope of `ColumnFamilyOptions::compression_opts` to
additionally include flushed files and files generated by compaction to
a non-bottommost level. Users can still get the old behavior by moving
their dictionary settings to `ColumnFamilyOptions::bottommost_compression_opts`
and explicitly enabling both that and `ColumnFamilyOptions::bottommost_compression`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7619
Reviewed By: ltamasi
Differential Revision: D24665610
Pulled By: ajkr
fbshipit-source-id: 656b90bce1033fe21c71e09af931ef5bde3e464c
Summary:
The old flag-based APIs (`BlockBasedTableOptions::pin_l0_filter_and_index_blocks_in_cache` and `BlockBasedTableOptions::pin_top_level_index_and_filter`) were insufficient for our needs. For example, it was impossible to pin only unpartitioned meta-blocks, which could prevent block cache contention when turning on dictionary compression or during a migration to partitioned indexes/filters. It was also impossible to pin all meta-blocks in memory while having predictable memory usage via block cache. If we had continued adding flags to address these scenarios, they would have had significant overlap causing confusion. Instead, this PR deprecates the flags and starts a new API with non-overlapping options.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7520
Test Plan:
- new unit test
- added new options to stress/crash test and ran for a while: `$ python tools/db_crashtest.py blackbox --simple --max_key=1000000 -write_buffer_size=1048576 -target_file_size_base=1048576 -max_bytes_for_level_base=4194304 --interval=10 -value_size_mult=33 -column_families=1 -reopen=0`
Reviewed By: pdillinger
Differential Revision: D24200034
Pulled By: ajkr
fbshipit-source-id: 3fa7cfc71e7960f7a867511dd6ae5834dd73b13e
Summary:
This PR merges the functionality of making the ColumnFamilyOptions, TableFactory, and DBOptions into Configurable into a single PR, resolving any merge conflicts
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5753
Reviewed By: ajkr
Differential Revision: D23385030
Pulled By: zhichao-cao
fbshipit-source-id: 8b977a7731556230b9b8c5a081b98e49ee4f160a
Summary:
After https://github.com/facebook/rocksdb/pull/7036, we still see extra DBTest that can timeout when running 10 or 20 in parallel. Expand skip-fsync mode in whole DBTest. Still preserve other tests from doing this mode to be conservative.
This commit reinstates https://github.com/facebook/rocksdb/issues/7049, whose un-revert was lost in an automatic
infrastructure mis-merge.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7274
Test Plan: Run all existing files.
Reviewed By: pdillinger
Differential Revision: D23177444
fbshipit-source-id: 1f61690b2ac6333c3b2c87176fef6b2cba086b33
