Summary:
This PR updates `VersionEditHandlerPointInTime` to recover all or none of the updates in an AtomicGroup. This makes best-effort recovery properly handle atomic flushes during recovery, so the features are now allowed to both be enabled at once.
The new logic requires that AtomicGroups do not contain column family additions or removals. AtomicGroups are currently written for atomic flush, which does not include such edits.
Column family additions or removals are recovered independently of AtomicGroups. The new logic needs to be aware of removal, though, so that a dropped CF does not prevent completion of an AtomicGroup recovery.
The new logic treats each AtomicGroup as if it contains updates for all existing column families, even though it is possible to create AtomicGroups that only affect a subset of column families. This simplifies the logic at the expense of recovering less data in certain edge case scenarios.
The usage of `MaybeCreateVersion()` is pretty tricky. The goal is to create a barrier at the start of an AtomicGroup such that all valid states up to that point will be applied to `versions_`. Here is a summary.
- `MaybeCreateVersion(..., false)` creates a `Version` on a negative edge trigger (transition from valid to invalid). It was previously called when applying each update. Now, it is only called when applying non-AtomicGroup updates.
- `MaybeCreateVersion(..., true)` creates a `Version` on a positive level trigger (valid state). It was previously called only at the end of iteration. Now, it is additionally called before processing an AtomicGroup.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12406
Reviewed By: jaykorean, cbi42
Differential Revision: D54494904
Pulled By: ajkr
fbshipit-source-id: 0114a9fe1d04b471d086dcab5978ea8a3a56ad52
Summary:
Some of the errors like data race and heap-after-use are error out based on crash test reporting them as error by relying on stderr. So reverting back to original form unless we come up with a more reliable solution to error out.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12335
Reviewed By: cbi42
Differential Revision: D53534781
Pulled By: akankshamahajan15
fbshipit-source-id: b19aa560d1560ac2281f7bc04e13961ed751f178
Summary:
For the user defined timestamps in memtable only feature, some special handling for range deletion blocks are needed since both the key (start_key) and the value (end_key) of a range tombstone can contain user-defined timestamps. Handling for the key is taken care of in the same way as the other data blocks in the block based table. This PR adds the special handling needed for the value (end_key) part. This includes:
1) On the write path, when L0 SST files are first created from flush, user-defined timestamps are removed from an end key of a range tombstone. There are places where it's logically removed (replaced with a min timestamp) because there is still logic with the running comparator that expects a user key that contains timestamp. And in the block based builder, it is eventually physically removed before persisted in a block.
2) On the read path, when range deletion block is being read, we artificially pad a min timestamp to the end key of a range tombstone in `BlockBasedTableReader`.
3) For file boundary `FileMetaData.largest`, we artificially pad a max timestamp to it if it contains a range deletion sentinel. Anytime when range deletion end_key is used to update file boundaries, it's using max timestamp instead of the range tombstone's actual timestamp to mark it as an exclusive end. d69628e6ce/db/dbformat.h (L923-L935)
This max timestamp is removed when in memory `FileMetaData.largest` is persisted into Manifest, we pad it back when it's read from Manifest while handling related `VersionEdit` in `VersionEditHandler`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12254
Test Plan: Added unit test and enabled this feature combination's stress test.
Reviewed By: cbi42
Differential Revision: D52965527
Pulled By: jowlyzhang
fbshipit-source-id: e8315f8a2c5268e2ae0f7aec8012c266b86df985
Summary:
While working on Meta's internal test triaging process, I found that `db_crashtest.py` was printing out `stdout` and `stderr` altogether. Adding an option to print `stderr` separately so that it's easy to extract only `stderr` from the test run.
`print_stderr_separately` is introduced as an optional parameter with default value `False` to keep the existing behavior as is (except a few minor changes).
Minor changes to the existing behavior
- We no longer print `stderr has error message:` and `***` prefix to each line. We simply print `stderr:` before printing `stderr` if stderr is printed in stdout and print `stderr` as is.
- We no longer print `times error occurred in output is ...` which doesn't appear to have any values
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12301
Test Plan:
**Default Behavior (blackbox)**
Run printed everything as is
```
$> python3 tools/db_crashtest.py blackbox --simple --max_key=25000000 --write_buffer_size=4194304 2> /tmp/error.log
Running blackbox-crash-test with
interval_between_crash=120
total-duration=6000
...
Integrated BlobDB: blob files enabled 0, min blob size 0, blob file size 268435456, blob compression type NoCompression, blob GC enabled 0, cutoff 0.250000, force threshold 1.000000, blob compaction readahead size 0, blob file starting level 0
Integrated BlobDB: blob cache disabled
DB path: [/tmp/jewoongh/rocksdb_crashtest_blackboxwh7yxpec]
(Re-)verified 0 unique IDs
2024/01/29-09:16:30 Initializing worker threads
Crash-recovery verification passed :)
2024/01/29-09:16:35 Starting database operations
2024/01/29-09:16:35 Starting verification
Stress Test : 543.600 micros/op 8802 ops/sec
: Wrote 0.00 MB (0.27 MB/sec) (50% of 10 ops)
: Wrote 5 times
: Deleted 1 times
: Single deleted 0 times
: 4 read and 0 found the key
: Prefix scanned 0 times
: Iterator size sum is 0
: Iterated 0 times
: Deleted 0 key-ranges
: Range deletions covered 0 keys
: Got errors 0 times
: 0 CompactFiles() succeed
: 0 CompactFiles() did not succeed
stderr:
WARNING: prefix_size is non-zero but memtablerep != prefix_hash
Error : jewoongh injected test error This is not a real failure.
Verification failed :(
```
Nothing in stderr
```
$> cat /tmp/error.log
```
**Default Behavior (whitebox)**
Run printed everything as is
```
$> python3 tools/db_crashtest.py whitebox --simple --max_key=25000000 --write_buffer_size=4194304 2> /tmp/error.log
Running whitebox-crash-test with
total-duration=10000
...
(Re-)verified 571 unique IDs
2024/01/29-09:33:53 Initializing worker threads
Crash-recovery verification passed :)
2024/01/29-09:35:16 Starting database operations
2024/01/29-09:35:16 Starting verification
Stress Test : 97248.125 micros/op 10 ops/sec
: Wrote 0.00 MB (0.00 MB/sec) (12% of 8 ops)
: Wrote 1 times
: Deleted 0 times
: Single deleted 0 times
: 4 read and 1 found the key
: Prefix scanned 1 times
: Iterator size sum is 120868
: Iterated 4 times
: Deleted 0 key-ranges
: Range deletions covered 0 keys
: Got errors 0 times
: 0 CompactFiles() succeed
: 0 CompactFiles() did not succeed
stderr:
WARNING: prefix_size is non-zero but memtablerep != prefix_hash
Error : jewoongh injected test error This is not a real failure.
New cache capacity = 4865393
Verification failed :(
TEST FAILED. See kill option and exit code above!!!
```
Nothing in stderr
```
$> cat /tmp/error.log
```
**New option added (blackbox)**
```
$> python3 tools/db_crashtest.py blackbox --simple --max_key=25000000 --write_buffer_size=4194304 --print_stderr_separately 2> /tmp/error.log
Running blackbox-crash-test with
interval_between_crash=120
total-duration=6000
...
Integrated BlobDB: blob files enabled 0, min blob size 0, blob file size 268435456, blob compression type NoCompression, blob GC enabled 0, cutoff 0.250000, force threshold 1.000000, blob compaction readahead size 0, blob file starting level 0
Integrated BlobDB: blob cache disabled
DB path: [/tmp/jewoongh/rocksdb_crashtest_blackbox7ybna32z]
(Re-)verified 0 unique IDs
Compaction filter factory: DbStressCompactionFilterFactory
2024/01/29-09:05:39 Initializing worker threads
Crash-recovery verification passed :)
2024/01/29-09:05:46 Starting database operations
2024/01/29-09:05:46 Starting verification
Stress Test : 235.917 micros/op 16000 ops/sec
: Wrote 0.00 MB (0.16 MB/sec) (16% of 12 ops)
: Wrote 2 times
: Deleted 1 times
: Single deleted 0 times
: 9 read and 0 found the key
: Prefix scanned 0 times
: Iterator size sum is 0
: Iterated 0 times
: Deleted 0 key-ranges
: Range deletions covered 0 keys
: Got errors 0 times
: 0 CompactFiles() succeed
: 0 CompactFiles() did not succeed
```
stderr printed separately
```
$> cat /tmp/error.log
WARNING: prefix_size is non-zero but memtablerep != prefix_hash
Error : jewoongh injected test error This is not a real failure.
New cache capacity = 19461571
Verification failed :(
```
**New option added (whitebox)**
```
$> python3 tools/db_crashtest.py whitebox --simple --max_key=25000000 --write_buffer_size=4194304 --print_stderr_separately 2> /tmp/error.log
Running whitebox-crash-test with
total-duration=10000
...
Integrated BlobDB: blob files enabled 0, min blob size 0, blob file size 268435456, blob compression type NoCompression, blob GC enabled 0, cutoff 0.250000, force threshold 1.000000, blob compaction readahead size 0, blob file starting level 0
Integrated BlobDB: blob cache disabled
DB path: [/tmp/jewoongh/rocksdb_crashtest_whiteboxtwj0ihn6]
(Re-)verified 157 unique IDs
2024/01/29-09:39:59 Initializing worker threads
Crash-recovery verification passed :)
2024/01/29-09:40:16 Starting database operations
2024/01/29-09:40:16 Starting verification
Stress Test : 742.474 micros/op 11801 ops/sec
: Wrote 0.00 MB (0.27 MB/sec) (36% of 19 ops)
: Wrote 7 times
: Deleted 1 times
: Single deleted 0 times
: 8 read and 0 found the key
: Prefix scanned 0 times
: Iterator size sum is 0
: Iterated 4 times
: Deleted 0 key-ranges
: Range deletions covered 0 keys
: Got errors 0 times
: 0 CompactFiles() succeed
: 0 CompactFiles() did not succeed
TEST FAILED. See kill option and exit code above!!!
```
stderr printed separately
```
$> cat /tmp/error.log
WARNING: prefix_size is non-zero but memtablerep != prefix_hash
Error : jewoongh injected test error This is not a real failure.
Error : jewoongh injected test error This is not a real failure.
Error : jewoongh injected test error This is not a real failure.
New cache capacity = 4865393
Verification failed :(
```
Reviewed By: akankshamahajan15
Differential Revision: D53187491
Pulled By: jaykorean
fbshipit-source-id: 76f9100d08b96d014e41b7b88b206d69f0ae932b
Summary:
Right now crash_test relies on std::errors too to check for only errors/failures along with verification. However, that's not a reliable solution and many internal services logs benign errors/warnings in which case our test script fails.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12265
Test Plan: Keep std::errors but printout instead of failing and will monitor crash tests internally to see if there is any scenario which solely relies on std::error, in which case stress tests can be improve.
Reviewed By: ajkr, cbi42
Differential Revision: D52967000
Pulled By: akankshamahajan15
fbshipit-source-id: 5328c8b69480c7946fe6a9c72f9ffeede70ac2ad
Summary:
Currently, we treat the long-running whitebox_crash_test as passing. However, we were not cleaning up after ourselves when we killed the running test for running too long, which often caused out-of-space errors in subsequent tests (e.g., blackbox_crash_test after whitebox_crash_test).
Unless we want to start treating these timeouts as failures and need the DB output for investigation now, we should properly clean up the tmp dir.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12248
Test Plan:
```
$> make crash_test -j
```
Reviewed By: ajkr
Differential Revision: D52885342
Pulled By: jaykorean
fbshipit-source-id: 7c1f2ca7cf03d0705bb14155ee44d5d7a411c132
Summary:
We test LockWAL() and UnlockWAL() by checking that latest sequence number is not changed: 1a1f9f1660/db_stress_tool/db_stress_test_base.cc (L920-L937). With writeprepared transaction, sequence number can be advanced in SwitchMemtable::WriteRecoverableState() when writing recoverable state: 1a1f9f1660/db/db_impl/db_impl_write.cc (L1560)
This PR disables LockWAL() tests for writeprepared transaction for now. We probably need to change how we test LockWAL() for writeprepared before re-enabling this test.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12221
Reviewed By: ajkr
Differential Revision: D52677076
Pulled By: cbi42
fbshipit-source-id: 27ee694878edf63e8f4ad52f769d4db401f511bc
Summary:
This feature combination is not fully working yet. Disable them so the stress tests have less noise.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12218
Reviewed By: cbi42
Differential Revision: D52643957
Pulled By: jowlyzhang
fbshipit-source-id: 8815a18a3b5814cad4f7ec41f3fb94869302081e
Summary:
**Context/Summary:**
Continued from https://github.com/facebook/rocksdb/pull/12127, we can randomly reduce the # max key to coerce more operations on the same key. My experimental run shows it surfaced more issue than just https://github.com/facebook/rocksdb/pull/12127.
I also randomly reduce the related parameters, write buffer size and target file base, to adapt to randomly lower number of # max key. This creates 4 situations of testing, 3 of which are new:
1. **high** # max key with **high** write buffer size and target file base (existing)
2. **high** # max key with **low** write buffer size and target file base (new, will go through some rehearsal testing to ensure we don't run out of space with many files)
3. **low** # max key with **high** write buffer size and target file base (new, keys will stay in memory longer)
4. **low** # max key with **low** write buffer size and target file base (new, experimental runs show it surfaced even more issues)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12148
Test Plan:
- [Ongoing] Rehearsal stress test
- Monitor production stress test
Reviewed By: jaykorean
Differential Revision: D52174980
Pulled By: hx235
fbshipit-source-id: bd5e11280826819ca9314c69bbbf05d481c6d105
Summary:
This PR adds initial stress testing for the user-defined timestamps in memtable only feature. Each flavor of the `*_ts` crash test get a 1 in 3 chance to run with timestamps not persisted, this setting is initialized once and kept consistent across the following re-runs.
This initial stress test included these things besides disabling incompatible feature combinations to make the test run more stably:
1) It currently only run test methods that validates db state with expected state. Not the ones that validate db state by comparing result from one API to another API. Such as `TestMultiGet` (compared with `Get`), similarly `TestMultiGetEntity`, `TestIterate` (compare src iterator to a control iterator). Due to timestamps being removed, results from one API to another API is not directly comparable as it is now. More test logic to handle that need to be added, will do that in a follow up.
2) Even when comparing db state to expected state, sometimes the db can receive `InvalidArgument` too due to timestamps getting flushed and removed. Added some logic to handle that.
3) When timestamps are not persisted, we don't try to read with older timestamp. Since that's making it easier to get `InvalidArgument`. And this capability is not yet needed by our customer so it's disabled for now.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12124
Test Plan: running multiple flavor of this test on continuous run for sometime before checkin
Reviewed By: ltamasi
Differential Revision: D51916267
Pulled By: jowlyzhang
fbshipit-source-id: 3f3eb5f9618d05d296062820e0ef5cb8edc7c2b2
Summary:
**Context/Summary:**
My experimental stress runs with more frequent "xxx_one_in" surfaced a couple interesting bugs/issues with RocksDB or crash test framework in the past. We now consider changing the default value so they are run more frequently in production testing environment.
Increase frequency by 2 orders of magnitude for most parameters, except for error-prone features e.g, manual compaction and file ingestion (increased by 3 orders) and expensive features e.g, checksum verification (increased by 1 order)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12127
Test Plan: Monitor CI to see if it did surface more interesting bugs/issues. If not, we may consider intensify even more.
Reviewed By: pdillinger
Differential Revision: D51954235
Pulled By: hx235
fbshipit-source-id: 92046cb7c52a37212f19ab7965b40f77b90b08b1
Summary:
This is a simple refactor for the crash test script to put shared logic for parsing stderr into a function. There is no functional change.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12109
Test Plan: manually tested the script
Reviewed By: ajkr
Differential Revision: D51692172
Pulled By: jowlyzhang
fbshipit-source-id: d346d64e981d9c489c380ff6ce33296a224b5877
Summary:
I have finally tracked down and fixed a bug affecting AutoHCC that was causing CI crash test assertion failures in AutoHCC when using secondary cache, but I was only able to reproduce locally a couple of times, after very long runs/repetitions.
It turns out that the essential feature used by secondary cache to trigger the bug is Insert without keeping a handle, which is otherwise rarely used in RocksDB and not incorporated into cache_bench (also used for targeted correctness stress testing) until this change (new option `-blind_insert_percent`).
The problem was in copying some logic from FixedHCC that makes the entry "sharable" but unreferenced once populated, if no reference is to be saved. The problem in AutoHCC is that we can only add the entry to a chain after it is in the sharable state, and must be removed from the chain while in the "under (de)construction" state and before it is back in the "empty" state. Also, it is possible for Lookup to find entries that are not connected to any chain, by design for efficiency, and for Release to erase_if_last_ref. Therefore, we could have
* Thread 1 starts to Insert a cache entry without keeping ref, and pauses before adding to the chain.
* Thread 2 finds it with Lookup optimizations, and then does Release with `erase_if_last_ref=true` causing it to trigger erasure on the entry. It successfully locks the home chain for the entry and purges any entries pending erasure. It is OK that this entry is not found on the chain, as another thread is allowed to remove it from the chain before we are able to (but after is it marked for (de)construction). And after the purge of the chain, the entry is marked empty.
* Thread 1 resumes in adding the slot (presumed entry) to the home chain for what was being inserted, but that now violates invariants and sets up a race or double-chain-reference as another thread could insert a new entry in the slot and try to insert into a different chain.
This is easily fixed by holding on to a reference until inserted onto the chain.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12046
Test Plan:
As I don't have a reliable local reproducer, I triggered 20 runs of internal CI on fbcode_blackbox_crash_test that were previously failing in AutoHCC with about 1/3 probability, and they all passed.
Also re-enabling AutoHCC in the crash test with this change. (Revert https://github.com/facebook/rocksdb/issues/12000)
Reviewed By: jowlyzhang
Differential Revision: D51016979
Pulled By: pdillinger
fbshipit-source-id: 3840fb829d65b97c779d8aed62a4a4a433aeff2b
Summary:
db_stress flag `verify_iterator_with_expected_state_one_in` is only enabled for in crash test if --simple flag is set. This PR enables it for all supported crash tests by enabling it by default. This adds coverage for --txn and --enable_ts crash tests.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12040
Test Plan:
ran crash tests that disabled this flag before for a few hours
```
python3 ./tools/db_crashtest.py blackbox --verify_iterator_with_expected_state_one_in=1 --txn --txn_write_policy=[0,1,2]
python3 ./tools/db_crashtest.py blackbox --verify_iterator_with_expected_state_one_in=1 --enable_ts
```
Reviewed By: ajkr, hx235
Differential Revision: D50980001
Pulled By: cbi42
fbshipit-source-id: 3daf6b4c32bdddc5df057240068162aa1a907587
Summary:
- Right now in blackbox test we don't exit if there are std::error as we do in whitebox crash tests. As result those errors are swallowed.
It only errors out if state is unexpected.
One example that was noticed in blackbox crash test -
```
stderr has error message:
***Error restoring historical expected values: Corruption: DB is older than any restorable expected state***
Running db_stress with pid=30454: /packages/rocksdb_db_stress_internal_repo/rocks_db_stress ....
```
- This diff also provided support to export files - db_crashtest.py file to be used by different repo.
Reviewed By: ajkr
Differential Revision: D50564889
fbshipit-source-id: 7bafbbc6179dc79467ca2b680fe83afc7850616a
Summary:
... until I can reproduce and resolve assertion failures (mostly in PurgeImplLocked) seen in crash test.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12000
Test Plan: make blackbox_crash_test
Reviewed By: hx235
Differential Revision: D50565984
Pulled By: pdillinger
fbshipit-source-id: 5eea1638ff2683c41b4f65ee1ffc2398071911e7
Summary:
When `secondary_cache_uri` is non-empty and the `cache_type` is not a tiered cache, then sanitize `compressed_secondary_cache_size` to 0.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11967
Test Plan: Run crash test
Reviewed By: akankshamahajan15
Differential Revision: D50346157
Pulled By: anand1976
fbshipit-source-id: 57bcbad2ec81fa736f1539a0a41ed6854ded2077
Summary:
Thanks ltamasi and ajkr for initial investigations on the test failure. Per the investigations, the following scenario is likely causing the test to fail.
1. Recovery is needed (could be any reason during crash test)
2. Trying to recover from the latest manifest fails (likely due to read error injection)
3. DB opens with recovery from the next manifest which is different from step 2.
4. Expected state is based on the manifest we tried and failed in step 2.
5. Two manifests used in step 2 and 3 are confirmed to have difference in LSM trees (Thanks ltamasi again for the finding).
```
2023/10/05-11:24:18.942189 56341 [db/version_set.cc:6079] Trying to recover from manifest: /dev/shm/rocksdb_test/rocksdb_crashtest_blackbox/MANIFEST-007184
...
2023/10/05-11:24:18.978007 56341 [db/version_set.cc:6079] Trying to recover from manifest: /dev/shm/rocksdb_test/rocksdb_crashtest_blackbox/MANIFEST-007180
```
```
[ltamasi@devbig1024.prn1 /tmp/x]$ ldb manifest_dump --hex --path=MANIFEST-007184_renamed_ > 2
[ltamasi@devbig1024.prn1 /tmp/x]$ ldb manifest_dump --hex --path=MANIFEST-007180_renamed_ > 1
[ltamasi@devbig1024.prn1 /tmp/x]$ diff 1 2
--- 1 2023-10-09 10:29:16.966215207 -0700
+++ 2 2023-10-09 10:29:11.984241645 -0700
@@ -13,7 +13,7 @@
7174:3950254[1875617 .. 2203952]['000000000003415B000000000000012B000000000000007D' seq:1906214, type:1 .. '000000000003CA59000000000000012B000000000000005C' seq:2039838, type:1]
7175:88060[2074748 .. 2203892]['000000000003CA6300000000000000CF78787878787878' seq:2167539, type:2 .. '000000000003D08F000000000000012B0000000000000130' seq:2112478, type:0]
--- level 6 --- version# 1 ---
- 7057:3132633[0 .. 2046144]['0000000000000009000000000000000978' seq:0, type:1 .. '0000000000005F8B000000000000012B00000000000002AC' seq:0, type:1]
+ 7219:2135565[0 .. 2046144]['0000000000000009000000000000000978' seq:0, type:1 .. '0000000000005F8B000000000000012B00000000000002AC' seq:0, type:1]
7061:827724[0 .. 2046131]['0000000000005F95000000000000000778787878787878' seq:0, type:1 .. '000000000000784F000000000000012B0000000000000113' seq:0, type:1]
6763:1352[0 .. 0]['000000000000784F000000000000012B0000000000000129' seq:0, type:1 .. '000000000000784F000000000000012B0000000000000129' seq:0, type:1]
7173:4812291[0 .. 2203957]['000000000000784F000000000000012B0000000000000138' seq:0, type:1 .. '0000000000020FAE787878787878' seq:0, type:1]
@@ -77,4 +77,4 @@
--- level 61 --- version# 1 ---
--- level 62 --- version# 1 ---
--- level 63 --- version# 1 ---
-next_file_number 7182 last_sequence 2203963 prev_log_number 0 max_column_family 0 min_log_number_to_keep 7015
+next_file_number 7221 last_sequence 2203963 prev_log_number 0 max_column_family 0 min_log_number_to_keep 7015
```
We have two options to fix this. Either skip verification against expected state or disable read injection when BE recovery is enabled. I chose to skip verification against expected state per discussion. (See comments in this PR)
Please note that some linter changes were included in this PR.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11938
Test Plan:
```
TEST_TMPDIR=/dev/shm/rocksdb make crash_test_with_best_efforts_recovery
```
Reviewed By: ltamasi
Differential Revision: D50136341
Pulled By: jaykorean
fbshipit-source-id: ac7434d592aebc148bfc3a4fcaa34936f136b95c
Summary:
This PR depends on https://github.com/facebook/rocksdb/issues/11879 . Enable write fault injection for the basic whitebox, blackbox, and cf_consistency modes. For other test modes like multiops_txn, best_efforts_recovery etc., leave it disabled for now until we can do more testing.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11924
Reviewed By: ajkr
Differential Revision: D50178252
Pulled By: anand1976
fbshipit-source-id: 5794f81c14cded1eb28762b2de818dfff1c1a34c
Summary:
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11906
The patch adds stress test coverage for the wide-column aware `FullMergeV3` API by implementing a new `DBStressWideMergeOperator`. This operator is similar to `PutOperator` / `PutOperatorV2` in the sense that its result is based on the last merge operand; however, the merge result can be either a plain value or a wide-column entity, depending on the value base encoded into the operand and the value of the `use_put_entity_one_in` stress test parameter. Following the same rule for merge results that we do for writes ensures that the queries issued by the validation logic receive the expected results. The new operator is used instead of `PutOperatorV2` whenever `use_put_entity_one_in` is positive. Note that the patch also makes it possible to set `use_put_entity_one_in` and `use_merge` (but not `use_full_merge_v1`) at the same time, giving `use_put_entity_one_in` precedence, so the stress test will use `PutEntity` for writes passing the `use_put_entity_one_in` check described above and `Merge` for any other writes.
Reviewed By: jaykorean
Differential Revision: D49760024
fbshipit-source-id: 3893602c3e7935381b484f4f5026f1983e3a04a9
Summary:
Crash tests are failing with recent change of auto_readahead_size. Disable it in stress tests and enable it with fix to clear the crash tests failures.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11883
Reviewed By: pdillinger
Differential Revision: D49597854
Pulled By: akankshamahajan15
fbshipit-source-id: 0af8ca7414ee9b92f244ee0fb811579c3c052b41
Summary:
This PR contains two fixes:
1. disable write fault injection since it caused several other kinds of internal stress test failures. I'll try to fix those separately before enabling it again.
2. Fix segfault like
```
https://github.com/facebook/rocksdb/issues/5 0x000000000083dc43 in rocksdb::port::Mutex::Lock (this=0x30) at internal_repo_rocksdb/repo/port/port_posix.cc:80
80 internal_repo_rocksdb/repo/port/port_posix.cc: No such file or directory.
https://github.com/facebook/rocksdb/issues/6 0x0000000000465142 in rocksdb::MutexLock::MutexLock (mu=0x30, this=<optimized out>) at internal_repo_rocksdb/repo/util/mutexlock.h:37
37 internal_repo_rocksdb/repo/util/mutexlock.h: No such file or directory.
https://github.com/facebook/rocksdb/issues/7 rocksdb::FaultInjectionTestFS::DisableWriteErrorInjection (this=0x0) at internal_repo_rocksdb/repo/utilities/fault_injection_fs.h:505
505 internal_repo_rocksdb/repo/utilities/fault_injection_fs.h: No such file or directory.
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11859
Test Plan: db_stress with no fault injection: `./db_stress --write_fault_one_in=0 --read_fault_one_in=0 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=0 --sync_fault_injection=0`
Reviewed By: jaykorean
Differential Revision: D49408247
Pulled By: cbi42
fbshipit-source-id: 0ca01f20e6e81bf52af77818b50d562ef7462165
Summary:
* db_crashtest.py now may set `write_fault_one_in` to 500 for blackbox and whitebox simple test.
* Error injection only applies to writing to SST files. Flush error will cause DB to pause background operations and auto-resume. Compaction error will just re-schedule later.
* File ingestion and back up tests are updated to check if the result status is due to an injected error.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11829
Test Plan:
a full round of whitebox simple and blackbox simple crash test
* `python3 ./tools/db_crashtest.py whitebox/blackbox --simple --write_fault_one_in=500`
Reviewed By: ajkr
Differential Revision: D49256962
Pulled By: cbi42
fbshipit-source-id: 68e0c9648d8e03bad39c7672b25d5500fc286d97
Summary:
An internal user wants to be able to dynamically switch between Bloom and Ribbon filters, without a custom FilterPolicy. Making `filter_policy` mutable would actually make issue https://github.com/facebook/rocksdb/issues/10079 worse, because it would be a race on a pointer field, not just on scalars.
As a reasonable compromise until that is fixed, I am enabling dynamic control over Bloom vs. Ribbon choice by making
RibbonFilterPolicy::bloom_before_level mutable, and doing that safely by using an atomic.
I've also slightly tweaked the interpretation of that field so that setting it to INT_MAX really means "always Bloom."
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11838
Test Plan: unit tests added/extended. crash test updated for SetOptions call and tested under TSAN with amplified probability (lower set_options_one_in).
Reviewed By: ajkr
Differential Revision: D49296284
Pulled By: pdillinger
fbshipit-source-id: e4251c077510df9a9c719876f482448c0d15402a
Summary:
- Fixed misspellings of "inject"
- Made user read errors retryable when `FLAGS_inject_error_severity == 1`
- Added compaction read errors when `FLAGS_read_fault_one_in > 0`. These are always retryable so that the DB will keep accepting writes
- Reenabled setting `compaction_readahead_size` in crash test. The reason for disabling it was to "keep the test clean", which is not a good enough reason to skip testing it
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11789
Test Plan:
With https://github.com/facebook/rocksdb/issues/11782 reverted, reproduced the bug:
- Build: `make -j56 db_stress`
- Command: `TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=524288 --target_file_size_base=524288 --max_bytes_for_level_base=2097152 --interval=10 --max_key=1000000`
- Output:
```
stderr has error message:
***put or merge error: Corruption: Compaction number of input keys does not match number of keys processed.***
```
Reviewed By: cbi42
Differential Revision: D48939994
Pulled By: ajkr
fbshipit-source-id: a1efb799efecdfd5d9cfd185e4a6321db8fccfbb
Summary:
Same as title
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11729
Test Plan: make crash_test -j32
Reviewed By: anand1976
Differential Revision: D48534820
Pulled By: akankshamahajan15
fbshipit-source-id: 3a2a28af98dfad164b82ddaaf9fddb94c53a652e
Summary:
In blackbox tests, db_stress command always run with timeout. Timeout can happen during validation, leaving some of the keys not checked. Since key validation is done in order, it is quite likely that keys those are towards to the end of the set are never validated. This PR adds a final execution, without timeout, to ensure validation is executed for all keys, at least once.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11592
Reviewed By: cbi42
Differential Revision: D48003998
Pulled By: hx235
fbshipit-source-id: 72543475a932f12cf0f57534b7e3b6e07e87080f
Summary:
For leveled compaction, RocksDB has a special kind of compaction with reason "kBottommmostFiles" that compacts bottommost level files to clear data held by snapshots (more detail in https://github.com/facebook/rocksdb/issues/3009). Such compactions can happen soon after a relevant snapshot is released. For some use cases, a bottommost file may contain only a small amount of keys that can be cleared, so compacting such a file has a high write amp. In addition, these bottommost files may be compacted in compactions with reason other than "kBottommmostFiles" if we wait for some time (so that enough data is ingested to trigger such a compaction). This PR introduces an option `bottommost_file_compaction_delay` to specify the delay of these bottommost level single file compactions.
* The main change is in `VersionStorageInfo::ComputeBottommostFilesMarkedForCompaction()` where we only add a file to `bottommost_files_marked_for_compaction_` if it oldest_snapshot is larger than its non-zero largest_seqno **and** the file is old enough. Note that if a file is not old enough but its largest_seqno is less than oldest_snapshot, we exclude it from the calculation of `bottommost_files_mark_threshold_`. This makes the change simpler, but such a file's eligibility for compaction will only be checked the next time `ComputeBottommostFilesMarkedForCompaction()` is called. This happens when a new Version is created (compaction, flush, SetOptions()...), a new enough snapshot is released (`VersionStorageInfo::UpdateOldestSnapshot()`) or when a compaction is picked and compaction score has to be re-calculated.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11701
Test Plan:
* Add two unit tests to test when bottommost_file_compaction_delay > 0.
* Ran crash test with the new option.
Reviewed By: jaykorean, ajkr
Differential Revision: D48331564
Pulled By: cbi42
fbshipit-source-id: c584f3dc5f6354fce3ed65f4c6366dc450b15ba8
Summary:
* The plan is for AutoHyperClockCache to be selected when HyperClockCacheOptions::estimated_entry_charge == 0, and in that case to use a new configuration option min_avg_entry_charge for determining an extreme case maximum size for the hash table. For the placeholder, a hack is in place in HyperClockCacheOptions::MakeSharedCache() to make the unit tests happy despite the new options not really making sense with the current implementation.
* Mostly updating and refactoring tests to test both the current HCC (internal name FixedHyperClockCache) and a placeholder for the new version (internal name AutoHyperClockCache).
* Simplify some existing tests not to depend directly on cache type.
* Type-parameterize the shard-level unit tests, which unfortunately requires more syntax like `this->` in places for disambiguation.
* Added means of choosing auto_hyper_clock_cache to cache_bench, db_bench, and db_stress, including add to crash test.
* Add another templated class BaseHyperClockCache to reduce future copy-paste
* Added ReportProblems support to cache_bench
* Added a DEBUG-level diagnostic to ReportProblems for the variance in load factor throughout the table, which will become more of a concern with linear hashing to be used in the Auto implementation. Example with current Fixed HCC:
```
2023/08/10-13:41:41.602450 6ac36 [DEBUG] [che/clock_cache.cc:1507] Slot occupancy stats: Overall 49% (129008/262144), Min/Max/Window = 39%/60%/500, MaxRun{Pos/Neg} = 18/17
```
In other words, with overall occupancy of 49%, the lowest across any 500 contiguous cells is 39% and highest 60%. Longest run of occupied is 18 and longest run of unoccupied is 17. This seems consistent with random samples from a uniform distribution.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11692
Test Plan: Shouldn't be any meaningful changes yet to production code or to what is tested, but there is temporary redundancy in testing until the new implementation is plugged in.
Reviewed By: jowlyzhang
Differential Revision: D48247413
Pulled By: pdillinger
fbshipit-source-id: 11541f996d97af403c2e43c92fb67ff22dd0b5da
Summary:
**Context/Summary**
After https://github.com/facebook/rocksdb/pull/11058, we no longer lock the key range to iterate in TestIterateAgainstExpected, except for working with timestamp feature.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11695
Test Plan: no code change
Reviewed By: ajkr
Differential Revision: D48276668
Pulled By: hx235
fbshipit-source-id: dc92a3708b2281dc737c0877fb755548bf03a9fc
Summary:
**Context/Summary:**
- Similar to https://github.com/facebook/rocksdb/pull/11288 but for user read such as `Get(), MultiGet(), DBIterator::XXX(), Verify(File)Checksum()`.
- For this, I refactored some user-facing `MultiGet` calls in `TransactionBase` and various types of `DB` so that it does not call a user-facing `Get()` but `GetImpl()` for passing the `ReadOptions::io_activity` check (see PR conversation)
- New user read stats breakdown are guarded by `kExceptDetailedTimers` since measurement shows they have 4-5% regression to the upstream/main.
- Misc
- More refactoring: with https://github.com/facebook/rocksdb/pull/11288, we complete passing `ReadOptions/IOOptions` to FS level. So we can now replace the previously [added](https://github.com/facebook/rocksdb/pull/9424) `rate_limiter_priority` parameter in `RandomAccessFileReader`'s `Read/MultiRead/Prefetch()` with `IOOptions::rate_limiter_priority`
- Also, `ReadAsync()` call time is measured in `SST_READ_MICRO` now
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11444
Test Plan:
- CI fake db crash/stress test
- Microbenchmarking
**Build** `make clean && ROCKSDB_NO_FBCODE=1 DEBUG_LEVEL=0 make -jN db_basic_bench`
- google benchmark version: 604f6fd3f4
- db_basic_bench_base: upstream
- db_basic_bench_pr: db_basic_bench_base + this PR
- asyncread_db_basic_bench_base: upstream + [db basic bench patch for IteratorNext](https://github.com/facebook/rocksdb/compare/main...hx235:rocksdb:micro_bench_async_read)
- asyncread_db_basic_bench_pr: asyncread_db_basic_bench_base + this PR
**Test**
Get
```
TEST_TMPDIR=/dev/shm ./db_basic_bench_{null_stat|base|pr} --benchmark_filter=DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/negative_query:0/enable_filter:0/mmap:1/threads:1 --benchmark_repetitions=1000
```
Result
```
Coming soon
```
AsyncRead
```
TEST_TMPDIR=/dev/shm ./asyncread_db_basic_bench_{base|pr} --benchmark_filter=IteratorNext/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/async_io:1/include_detailed_timers:0 --benchmark_repetitions=1000 > syncread_db_basic_bench_{base|pr}.out
```
Result
```
Base:
1956,1956,1968,1977,1979,1986,1988,1988,1988,1990,1991,1991,1993,1993,1993,1993,1994,1996,1997,1997,1997,1998,1999,2001,2001,2002,2004,2007,2007,2008,
PR (2.3% regression, due to measuring `SST_READ_MICRO` that wasn't measured before):
1993,2014,2016,2022,2024,2027,2027,2028,2028,2030,2031,2031,2032,2032,2038,2039,2042,2044,2044,2047,2047,2047,2048,2049,2050,2052,2052,2052,2053,2053,
```
Reviewed By: ajkr
Differential Revision: D45918925
Pulled By: hx235
fbshipit-source-id: 58a54560d9ebeb3a59b6d807639692614dad058a
Summary:
Add a mutable column family option `memtable_max_range_deletions`. When non-zero, RocksDB will try to flush the current memtable after it has at least `memtable_max_range_deletions` range deletions. Java API is added and crash test is updated accordingly to randomly enable this option.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11358
Test Plan:
* New unit test: `DBRangeDelTest.MemtableMaxRangeDeletions`
* Ran crash test `python3 ./tools/db_crashtest.py whitebox --simple --memtable_max_range_deletions=20` and saw logs showing flushed memtables usually with 20 range deletions.
Reviewed By: ajkr
Differential Revision: D46582680
Pulled By: cbi42
fbshipit-source-id: f23d6fa8d8264ecf0a18d55c113ba03f5e2504da
Summary:
## Context checksum
All RocksDB checksums currently use 32 bits of checking
power, which should be 1 in 4 billion false negative (FN) probability (failing to
detect corruption). This is true for random corruptions, and in some cases
small corruptions are guaranteed to be detected. But some possible
corruptions, such as in storage metadata rather than storage payload data,
would have a much higher FN rate. For example:
* Data larger than one SST block is replaced by data from elsewhere in
the same or another SST file. Especially with block_align=true, the
probability of exact block size match is probably around 1 in 100, making
the FN probability around that same. Without `block_align=true` the
probability of same block start location is probably around 1 in 10,000,
for FN probability around 1 in a million.
To solve this problem in new format_version=6, we add "context awareness"
to block checksum checks. The stored and expected checksum value is
modified based on the block's position in the file and which file it is in. The
modifications are cleverly chosen so that, for example
* blocks within about 4GB of each other are guaranteed to use different context
* blocks that are offset by exactly some multiple of 4GiB are guaranteed to use
different context
* files generated by the same process are guaranteed to use different context
for the same offsets, until wrap-around after 2^32 - 1 files
Thus, with format_version=6, if a valid SST block and checksum is misplaced,
its checksum FN probability should be essentially ideal, 1 in 4B.
## Footer checksum
This change also adds checksum protection to the SST footer (with
format_version=6), for the first time without relying on whole file checksum.
To prevent a corruption of the format_version in the footer (e.g. 6 -> 5) to
defeat the footer checksum, we change much of the footer data format
including an "extended magic number" in format_version 6 that would be
interpreted as empty index and metaindex block handles in older footer
versions. We also change the encoding of handles to free up space for
other new data in footer.
## More detail: making space in footer
In order to keep footer the same size in format_version=6 (avoid change to IO
patterns), we have to free up some space for new data. We do this two ways:
* Metaindex block handle is encoded down to 4 bytes (from 10) by assuming
it immediately precedes the footer, and by assuming it is < 4GB.
* Index block handle is moved into metaindex. (I don't know why it was
in footer to begin with.)
## Performance
In case of small performance penalty, I've made a "pay as you go" optimization
to compensate: replace `MutableCFOptions` in BlockBasedTableBuilder::Rep
with the only field used in that structure after construction: `prefix_extractor`.
This makes the PR an overall performance improvement (results below).
Nevertheless I'm seeing essentially no difference going from fv=5 to fv=6,
even including that improvement for both. That's based on extreme case table
write performance testing, many files with many blocks. This is relatively
checksum intensive (small blocks) and salt generation intensive (small files).
```
(for I in `seq 1 100`; do TEST_TMPDIR=/dev/shm/dbbench2 ./db_bench -benchmarks=fillseq -memtablerep=vector -disable_wal=1 -allow_concurrent_memtable_write=false -num=3000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -write_buffer_size=100000 -compression_type=none -block_size=1000; done) 2>&1 | grep micros/op | tee out
awk '{ tot += $5; n += 1; } END { print int(1.0 * tot / n) }' < out
```
Each value below is ops/s averaged over 100 runs, run simultaneously with competing
configuration for load fairness
Before -> after (both fv=5): 483530 -> 483673 (negligible)
Re-run 1: 480733 -> 485427 (1.0% faster)
Re-run 2: 483821 -> 484541 (0.1% faster)
Before (fv=5) -> after (fv=6): 482006 -> 485100 (0.6% faster)
Re-run 1: 482212 -> 485075 (0.6% faster)
Re-run 2: 483590 -> 484073 (0.1% faster)
After fv=5 -> after fv=6: 483878 -> 485542 (0.3% faster)
Re-run 1: 485331 -> 483385 (0.4% slower)
Re-run 2: 485283 -> 483435 (0.4% slower)
Re-run 3: 483647 -> 486109 (0.5% faster)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9058
Test Plan:
unit tests included (table_test, db_properties_test, salt in env_test). General DB tests
and crash test updated to test new format_version.
Also temporarily updated the default format version to 6 and saw some test failures. Almost all
were due to an inadvertent additional read in VerifyChecksum to verify the index block checksum,
though it's arguably a bug that VerifyChecksum does not appear to (re-)verify the index block
checksum, just assuming it was verified in opening the index reader (probably *usually* true but
probably not always true). Some other concerns about VerifyChecksum are left in FIXME
comments. The only remaining test failure on change of default (in block_fetcher_test) now
has a comment about how to upgrade the test.
The format compatibility test does not need updating because we have not updated the default
format_version.
Reviewed By: ajkr, mrambacher
Differential Revision: D33100915
Pulled By: pdillinger
fbshipit-source-id: 8679e3e572fa580181a737fd6d113ed53c5422ee
Summary:
Add `skip_tmpdir_check` argument in crash script. If `tmp_dir` is on remote storage and exist, `isdir` will be false (checking on local storage) leading to exit. By passing `skip_tmpdir_check` with `crashtest.py`, the dir check can be skipped.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11539
Test Plan: Ran locally
Reviewed By: anand1976
Differential Revision: D46740456
Pulled By: akankshamahajan15
fbshipit-source-id: 8726882ef53d2c84b604c7515e84eda6d1bf797c
Summary:
Context:
OptimisticTransactionDB has not been covered by db_stress (including crash test) like TransactionDB.
1. Adding the following gflag options to to test OptimisticTransactionDB
- `use_optimistic_txn`: When true, open OptimisticTransactionDB to test
- `occ_validation_policy`: `OccValidationPolicy::kValidateParallel = 1` by default.
- `share_occ_lock_buckets`: Use shared occ locks
- `occ_lock_bucket_count`: 500 by default. Number of buckets to use for shared occ lock.
2. Opening OptimisticTransactionDB and NewTxn/Commit added per `use_optimistic_txn` flag in `db_stress_test_base.cc`
3. OptimisticTransactionDB blackbox/whitebox test added in crash_test.mk
Please note that the existing flag `use_txn` is being used here. When `use_txn == true` and `use_optimistic_txn == false`, we use `TransactionDB` (a.k.a. pessimistic transaction db). When both `use_txn` and `use_optimistic_txn` are true, we use `OptimisticTransactionDB`. If `use_txn == false` but `use_optimistic_txn == true` throw error with message _"You cannot set use_optimistic_txn true while use_txn is false. Please set use_txn true if you want to use OptimisticTransactionDB"_.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11513
Test Plan:
**Crash Test**
Serial Validation
```
export CRASH_TEST_EXT_ARGS="--use_optimistic_txn=1 --use_txn=1 --use_put_entity_one_in=0 --occ_validation_policy=0"
make crash_test -j
```
Parallel Validation (no share bucket)
```
export CRASH_TEST_EXT_ARGS="--use_optimistic_txn=1 --use_txn=1 --use_put_entity_one_in=0 --occ_validation_policy=1 --share_occ_lock_buckets=0"
make crash_test -j
```
Parallel Validation (share bucket)
```
export CRASH_TEST_EXT_ARGS="--use_optimistic_txn=1 --use_txn=1 --use_put_entity_one_in=0 --occ_validation_policy=1 --share_occ_lock_buckets=1 --occ_lock_bucket_count=500"
make crash_test -j
```
**Stress Test**
```
./db_stress -use_optimistic_txn -threads=32
```
Reviewed By: pdillinger
Differential Revision: D46547387
Pulled By: jaykorean
fbshipit-source-id: ca19819ca6e0281694966998014b40d95d4e5960
Summary:
- add TEST_TMPDIR_EXPECTED env in crash test if expected dir is on different filesystem. If TEST_TMPDIR_EXPECTED is not specified, it'll fallback to default value of TEST_TMPDIR (Same as before)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11448
Test Plan: Ran locally
Reviewed By: anand1976
Differential Revision: D45870268
Pulled By: akankshamahajan15
fbshipit-source-id: 52a7b961d3647dde023dcf7f20341558e8a5b528
Summary:
add option `block_protection_bytes_per_key` and implementation for block per key-value checksum. The main changes are
1. checksum construction and verification in block.cc/h
2. pass the option `block_protection_bytes_per_key` around (mainly for methods defined in table_cache.h)
3. unit tests/crash test updates
Tests:
* Added unit tests
* Crash test: `python3 tools/db_crashtest.py blackbox --simple --block_protection_bytes_per_key=1 --write_buffer_size=1048576`
Follow up (maybe as a separate PR): make sure corruption status returned from BlockIters are correctly handled.
Performance:
Turning on block per KV protection has a non-trivial negative impact on read performance and costs additional memory.
For memory, each block includes additional 24 bytes for checksum-related states beside checksum itself. For CPU, I set up a DB of size ~1.2GB with 5M keys (32 bytes key and 200 bytes value) which compacts to ~5 SST files (target file size 256 MB) in L6 without compression. I tested readrandom performance with various block cache size (to mimic various cache hit rates):
```
SETUP
make OPTIMIZE_LEVEL="-O3" USE_LTO=1 DEBUG_LEVEL=0 -j32 db_bench
./db_bench -benchmarks=fillseq,compact0,waitforcompaction,compact,waitforcompaction -write_buffer_size=33554432 -level_compaction_dynamic_level_bytes=true -max_background_jobs=8 -target_file_size_base=268435456 --num=5000000 --key_size=32 --value_size=200 --compression_type=none
BENCHMARK
./db_bench --use_existing_db -benchmarks=readtocache,readrandom[-X10] --num=5000000 --key_size=32 --disable_auto_compactions --reads=1000000 --block_protection_bytes_per_key=[0|1] --cache_size=$CACHESIZE
The readrandom ops/sec looks like the following:
Block cache size: 2GB 1.2GB * 0.9 1.2GB * 0.8 1.2GB * 0.5 8MB
Main 240805 223604 198176 161653 139040
PR prot_bytes=0 238691 226693 200127 161082 141153
PR prot_bytes=1 214983 193199 178532 137013 108211
prot_bytes=1 vs -10% -15% -10.8% -15% -23%
prot_bytes=0
```
The benchmark has a lot of variance, but there was a 5% to 25% regression in this benchmark with different cache hit rates.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11287
Reviewed By: ajkr
Differential Revision: D43970708
Pulled By: cbi42
fbshipit-source-id: ef98d898b71779846fa74212b9ec9e08b7183940
Summary:
**Context:**
The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them.
**Summary**
- Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros`
- Fixed the default histogram in `RandomAccessFileReader`
- New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader`
- Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288
Test Plan:
- **Stress test**
- **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob)
- May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads.
```
./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10)
```
```
// BlockBasedTable
rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805
rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116
rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689
// PlainTable
Does not apply
```
- **Db bench 2: performance**
**Read**
SETUP: db with 900 files
```
./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none
```run till convergence
```
./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3
```
Pre-change
`readrandom [AVG 60 runs] : 21568 (± 248) ops/sec`
Post-change (no regression, -0.3%)
`readrandom [AVG 60 runs] : 21486 (± 236) ops/sec`
**Compaction/Flush**run till convergence
```
./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none
rocksdb.sst.read.micros COUNT : 33820
rocksdb.sst.read.flush.micros COUNT : 1800
rocksdb.sst.read.compaction.micros COUNT : 32020
```
Pre-change
`fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec`
Post-change (no regression, ~-0.4%)
`fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec`
Reviewed By: ajkr
Differential Revision: D44007011
Pulled By: hx235
fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
Summary:
Similarly to `GetEntity` prior to https://github.com/facebook/rocksdb/issues/11303, the `MultiGetEntity` API is currently
only used in the DB verification logic of the stress tests. The patch introduces
a new mode where all point lookups are performed using `MultiGetEntity`,
and implements the corresponding logic in the non-batched, batched, and
CF consistency tests.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11336
Test Plan: Ran simple blackbox tests for the various stress test flavors.
Reviewed By: akankshamahajan15
Differential Revision: D44513285
Pulled By: ltamasi
fbshipit-source-id: c3db098501bf875b6a356b09fc676a0268d92c35
Summary:
…evel_bytes
During DB open, if a column family uses level compaction with level_compaction_dynamic_level_bytes=true, trivially move its files down in the LSM such that the bottommost files are in Lmax, the second from bottommost level files are in Lmax-1 and so on. This is aimed to make it easier to migrate level_compaction_dynamic_level_bytes from false to true. Before this change, a full manual compaction is suggested for such migration. After this change, user can just restart DB to turn on this option. db_crashtest.py is updated to randomly choose value for level_compaction_dynamic_level_bytes.
Note that there may still be too many unnecessary levels if a user is migrating from universal compaction or level compaction with a smaller level multiplier. A full manual compaction may still be needed in that case before some PR that automatically drain unnecessary levels like https://github.com/facebook/rocksdb/issues/3921 lands. Eventually we may want to change the default value of option level_compaction_dynamic_level_bytes to true.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11321
Test Plan:
1. Added unit tests.
2. Crash test: ran a variation of db_crashtest.py (like 32516507e77521ae887e45091b69139e32e8efb7) that turns level_compaction_dynamic_level_bytes on and off and switches between LC and UC for the same DB.
TODO: Update `OptionChangeMigration`, either after this PR or https://github.com/facebook/rocksdb/issues/3921.
Reviewed By: ajkr
Differential Revision: D44341930
Pulled By: cbi42
fbshipit-source-id: 013de19a915c6a0502be569f07c4cc8f1c3c6be2
Summary:
Internally refactors SecondaryCache integration out of LRUCache specifically and into a wrapper/adapter class that works with various Cache implementations. Notably, this relies on separating the notion of async lookup handles from other cache handles, so that HyperClockCache doesn't have to deal with the problem of allocating handles from the hash table for lookups that might fail anyway, and might be on the same key without support for coalescing. (LRUCache's hash table can incorporate previously allocated handles thanks to its pointer indirection.) Specifically, I'm worried about the case in which hundreds of threads try to access the same block and probing in the hash table degrades to linear search on the pile of entries with the same key.
This change is a big step in the direction of supporting stacked SecondaryCaches, but there are obstacles to completing that. Especially, there is no SecondaryCache hook for evictions to pass from one to the next. It has been proposed that evictions be transmitted simply as the persisted data (as in SaveToCallback), but given the current structure provided by the CacheItemHelpers, that would require an extra copy of the block data, because there's intentionally no way to ask for a contiguous Slice of the data (to allow for flexibility in storage). `AsyncLookupHandle` and the re-worked `WaitAll()` should be essentially prepared for stacked SecondaryCaches, but several "TODO with stacked secondaries" issues remain in various places.
It could be argued that the stacking instead be done as a SecondaryCache adapter that wraps two (or more) SecondaryCaches, but at least with the current API that would require an extra heap allocation on SecondaryCache Lookup for a wrapper SecondaryCacheResultHandle that can transfer a Lookup between secondaries. We could also consider trying to unify the Cache and SecondaryCache APIs, though that might be difficult if `AsyncLookupHandle` is kept a fixed struct.
## cache.h (public API)
Moves `secondary_cache` option from LRUCacheOptions to ShardedCacheOptions so that it is applicable to HyperClockCache.
## advanced_cache.h (advanced public API)
* Add `Cache::CreateStandalone()` so that the SecondaryCache support wrapper can use it.
* Add `SetEvictionCallback()` / `eviction_callback_` so that the SecondaryCache support wrapper can use it. Only a single callback is supported for efficiency. If there is ever a need for more than one, hopefully that can be handled with a broadcast callback wrapper.
These are essentially the two "extra" pieces of `Cache` for pulling out specific SecondaryCache support from the `Cache` implementation. I think it's a good trade-off as these are reasonable, limited, and reusable "cut points" into the `Cache` implementations.
* Remove async capability from standard `Lookup()` (getting rid of awkward restrictions on pending Handles) and add `AsyncLookupHandle` and `StartAsyncLookup()`. As noted in the comments, the full struct of `AsyncLookupHandle` is exposed so that it can be stack allocated, for efficiency, though more data is being copied around than before, which could impact performance. (Lookup info -> AsyncLookupHandle -> Handle vs. Lookup info -> Handle)
I could foresee a future in which a Cache internally saves a pointer to the AsyncLookupHandle, which means it's dangerous to allow it to be copyable or even movable. It also means it's not compatible with std::vector (which I don't like requiring as an API parameter anyway), so `WaitAll()` expects any contiguous array of AsyncLookupHandles. I believe this is best for common case efficiency, while behaving well in other cases also. For example, `WaitAll()` has no effect on default-constructed AsyncLookupHandles, which look like a completed cache miss.
## cacheable_entry.h
A couple of functions are obsolete because Cache::Handle can no longer be pending.
## cache.cc
Provides default implementations for new or revamped Cache functions, especially appropriate for non-blocking caches.
## secondary_cache_adapter.{h,cc}
The full details of the Cache wrapper adding SecondaryCache support. Essentially replicates the SecondaryCache handling that was in LRUCache, but obviously refactored. There is a bit of logic duplication, where Lookup() is essentially a manually optimized version of StartAsyncLookup() and Wait(), but it's roughly a dozen lines of code.
## sharded_cache.h, typed_cache.h, charged_cache.{h,cc}, sim_cache.cc
Simply updated for Cache API changes.
## lru_cache.{h,cc}
Carefully remove SecondaryCache logic, implement `CreateStandalone` and eviction handler functionality.
## clock_cache.{h,cc}
Expose existing `CreateStandalone` functionality, add eviction handler functionality. Light refactoring.
## block_based_table_reader*
Mostly re-worked the only usage of async Lookup, which is in BlockBasedTable::MultiGet. Used arrays in place of autovector in some places for efficiency. Simplified some logic by not trying to process some cache results before they're all ready.
Created new function `BlockBasedTable::GetCachePriority()` to reduce some pre-existing code duplication (and avoid making it worse).
Fixed at least one small bug from the prior confusing mixture of async and sync Lookups. In MaybeReadBlockAndLoadToCache(), called by RetrieveBlock(), called by MultiGet() with wait=false, is_cache_hit for the block_cache_tracer entry would not be set to true if the handle was pending after Lookup and before Wait.
## Intended follow-up work
* Figure out if there are any missing stats or block_cache_tracer work in refactored BlockBasedTable::MultiGet
* Stacked secondary caches (see above discussion)
* See if we can make up for the small MultiGet performance regression.
* Study more performance with SecondaryCache
* Items evicted from over-full LRUCache in Release were not being demoted to SecondaryCache, and still aren't to minimize unit test churn. Ideally they would be demoted, but it's an exceptional case so not a big deal.
* Use CreateStandalone for cache reservations (save unnecessary hash table operations). Not a big deal, but worthy cleanup.
* Somehow I got the contract for SecondaryCache::Insert wrong in #10945. (Doesn't take ownership!) That API comment needs to be fixed, but didn't want to mingle that in here.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11301
Test Plan:
## Unit tests
Generally updated to include HCC in SecondaryCache tests, though HyperClockCache has some different, less strict behaviors that leads to some tests not really being set up to work with it. Some of the tests remain disabled with it, but I think we have good coverage without them.
## Crash/stress test
Updated to use the new combination.
## Performance
First, let's check for regression on caches without secondary cache configured. Adding support for the eviction callback is likely to have a tiny effect, but it shouldn't be worrisome. LRUCache could benefit slightly from less logic around SecondaryCache handling. We can test with cache_bench default settings, built with DEBUG_LEVEL=0 and PORTABLE=0.
```
(while :; do base/cache_bench --cache_type=hyper_clock_cache | grep Rough; done) | awk '{ sum += $9; count++; print $0; print "Average: " int(sum / count) }'
```
**Before** this and #11299 (which could also have a small effect), running for about an hour, before & after running concurrently for each cache type:
HyperClockCache: 3168662 (average parallel ops/sec)
LRUCache: 2940127
**After** this and #11299, running for about an hour:
HyperClockCache: 3164862 (average parallel ops/sec) (0.12% slower)
LRUCache: 2940928 (0.03% faster)
This is an acceptable difference IMHO.
Next, let's consider essentially the worst case of new CPU overhead affecting overall performance. MultiGet uses the async lookup interface regardless of whether SecondaryCache or folly are used. We can configure a benchmark where all block cache queries are for data blocks, and all are hits.
Create DB and test (before and after tests running simultaneously):
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16
TEST_TMPDIR=/dev/shm base/db_bench -benchmarks=multireadrandom[-X30] -readonly -multiread_batched -batch_size=32 -num=30000000 -bloom_bits=16 -cache_size=6789000000 -duration 20 -threads=16
```
**Before**:
multireadrandom [AVG 30 runs] : 3444202 (± 57049) ops/sec; 240.9 (± 4.0) MB/sec
multireadrandom [MEDIAN 30 runs] : 3514443 ops/sec; 245.8 MB/sec
**After**:
multireadrandom [AVG 30 runs] : 3291022 (± 58851) ops/sec; 230.2 (± 4.1) MB/sec
multireadrandom [MEDIAN 30 runs] : 3366179 ops/sec; 235.4 MB/sec
So that's roughly a 3% regression, on kind of a *worst case* test of MultiGet CPU. Similar story with HyperClockCache:
**Before**:
multireadrandom [AVG 30 runs] : 3933777 (± 41840) ops/sec; 275.1 (± 2.9) MB/sec
multireadrandom [MEDIAN 30 runs] : 3970667 ops/sec; 277.7 MB/sec
**After**:
multireadrandom [AVG 30 runs] : 3755338 (± 30391) ops/sec; 262.6 (± 2.1) MB/sec
multireadrandom [MEDIAN 30 runs] : 3785696 ops/sec; 264.8 MB/sec
Roughly a 4-5% regression. Not ideal, but not the whole story, fortunately.
Let's also look at Get() in db_bench:
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom[-X30] -readonly -num=30000000 -bloom_bits=16 -cache_size=6789000000 -duration 20 -threads=16
```
**Before**:
readrandom [AVG 30 runs] : 2198685 (± 13412) ops/sec; 153.8 (± 0.9) MB/sec
readrandom [MEDIAN 30 runs] : 2209498 ops/sec; 154.5 MB/sec
**After**:
readrandom [AVG 30 runs] : 2292814 (± 43508) ops/sec; 160.3 (± 3.0) MB/sec
readrandom [MEDIAN 30 runs] : 2365181 ops/sec; 165.4 MB/sec
That's showing roughly a 4% improvement, perhaps because of the secondary cache code that is no longer part of LRUCache. But weirdly, HyperClockCache is also showing 2-3% improvement:
**Before**:
readrandom [AVG 30 runs] : 2272333 (± 9992) ops/sec; 158.9 (± 0.7) MB/sec
readrandom [MEDIAN 30 runs] : 2273239 ops/sec; 159.0 MB/sec
**After**:
readrandom [AVG 30 runs] : 2332407 (± 11252) ops/sec; 163.1 (± 0.8) MB/sec
readrandom [MEDIAN 30 runs] : 2335329 ops/sec; 163.3 MB/sec
Reviewed By: ltamasi
Differential Revision: D44177044
Pulled By: pdillinger
fbshipit-source-id: e808e48ff3fe2f792a79841ba617be98e48689f5
Summary:
The `GetEntity` API is currently used in the stress tests for verification purposes;
this patch extends the coverage by adding a mode where all point lookups in
the non-batched, batched, and CF consistency stress tests are done using this API.
The PR also includes a bit of refactoring to eliminate some boilerplate code around
the wide-column consistency checks.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11303
Test Plan: Ran stress tests of the batched, non-batched, and CF consistency varieties.
Reviewed By: akankshamahajan15
Differential Revision: D44148503
Pulled By: ltamasi
fbshipit-source-id: fecdbfd3e65a459bbf16ab7aa7b9173e19240077