Commit Graph

10 Commits

Author SHA1 Message Date
Peter Dillinger
37fd2b9694 Revert "Generate variable length keys in db_stress (#6165)" and follow-ups (#6243)
Summary:
This commit is suspected in some crash test failures such as

Verification failed for column family 0 key 78438077: Value not found: NotFound:
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6243

Test Plan: 'make check' and start 'make crash_test'

Differential Revision: D19220495

Pulled By: pdillinger

fbshipit-source-id: 6c4709cee80ab4344e06ce360f51e947d79fb3fa
2019-12-23 16:32:57 -08:00
anand76
3160edfdc7 Generate variable length keys in db_stress (#6165)
Summary:
Currently, db_stress generates fixed length keys of 8 bytes. This patch adds the ability to generate variable length keys. Most of the db_stress code continues to work with a numeric key randomly generated, and the numeric key also acts as an index into the values_ array. The numeric key is mapped to a variable length string key in a deterministic way. Furthermore, the ordering is preserved.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6165

Test Plan: run make crash_test

Differential Revision: D19204646

Pulled By: anand1976

fbshipit-source-id: d2d46a96615b4832a8be2a981f5913905f0e1ca7
2019-12-20 21:10:33 -08:00
sdong
338c149b92 crash_test to cover bottommost compression and some other changes (#6215)
Summary:
Several improvements to crash_test/stress_test:
(1) Stress_test to support an parameter of bottommost compression
(2) Rename those FLAGS_* variables that are not gflags to avoid confusion
(3) Crash_test to randomly generate compression type for bottommost compression with half the chance.
(4) Stress_test to sanitize unsupported compression type to snappy, so that crash_test to cover all possible compression types and people don't need to worry about they don't support all comrpession types in their environment.
(5) In crash_test, when generating db_stress command, sort arguments in alphabeta order, so that it is easier to find value for a specific argument.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6215

Test Plan: Run "make crash_test" for a while and see the botommost option shown in LOG files.

Differential Revision: D19171255

fbshipit-source-id: d7001e246c4ff9ee5760776eea0be97738650735
2019-12-20 16:14:52 -08:00
Yanqin Jin
c4fd9cf461 Remove an unnecessary check before running db_stress (#6231)
Summary:
As title. We can run non-cf-consistency stress tests with verify_db_one_in>0,
thus remove the check added previously.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6231

Test Plan:
```
make crash_test
```

Differential Revision: D19198295

Pulled By: riversand963

fbshipit-source-id: e874c701bb03ab76eaab00f059dd4032bb2f537f
2019-12-20 11:29:06 -08:00
Yanqin Jin
670a916d01 Add more verification to db_stress (#6173)
Summary:
Currently, db_stress performs verification by calling `VerifyDb()` at the end of test and optionally before tests start. In case of corruption or incorrect result, it will be too late. This PR adds more verification in two ways.
1. For cf consistency test, each test thread takes a snapshot and verifies every N ops. N is configurable via `-verify_db_one_in`. This option is not supported in other stress tests.
2. For cf consistency test, we use another background thread in which a secondary instance periodically tails the primary (interval is configurable). We verify the secondary. Once an error is detected, we terminate the test and report. This does not affect other stress tests.

Test plan (devserver)
```
$./db_stress -test_cf_consistency -verify_db_one_in=0 -ops_per_thread=100000 -continuous_verification_interval=100
$./db_stress -test_cf_consistency -verify_db_one_in=1000 -ops_per_thread=10000 -continuous_verification_interval=0
$make crash_test
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6173

Differential Revision: D19047367

Pulled By: riversand963

fbshipit-source-id: aeed584ad71f9310c111445f34975e5ab47a0615
2019-12-20 08:49:29 -08:00
sdong
35126dd874 db_stress: preserve all historic manifest files (#6142)
Summary:
compaction history is stored in manifest files. Preserve all of them in db_stress would help debugging.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6142

Test Plan: Run db_stress and observe that manifest files are preserved. Run whole crash_test and see how DB directory looks like.

Differential Revision: D19047026

fbshipit-source-id: f83c3e0bb5332b1b4768be5dcee56a24f9b760a9
2019-12-16 14:32:34 -08:00
Zhichao Cao
fbda25f57a db_stress: generate the key based on Zipfian distribution (hot key) (#6163)
Summary:
In the current db_stress, all the keys are generated randomly and follows the uniform distribution. In order to test some corner cases that some key are always updated or read, we need to generate the key based on other distributions. In this PR, the key is generated based on Zipfian distribution and the skewness can be controlled by setting hot_key_alpha (0.8 to 1.5 is suggested). The larger hot_key_alpha is, the more skewed will be. Not that, usually, if hot_key_alpha is larger than 2, there might be only 1 or 2 keys that are generated. If hot_key_alpha is 0, it generate the key follows uniform distribution (random key)

Testing plan: pass the db_stress and printed the keys to make sure it follows the distribution.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6163

Differential Revision: D18978480

Pulled By: zhichao-cao

fbshipit-source-id: e123b4865477f7478e83fb581f9576bada334680
2019-12-16 14:01:58 -08:00
Peter Dillinger
6380df5e10 Vary bloom_bits in db_crashtest (#6103)
Summary:
Especially with non-integral bits/key now supported,
db_crashtest should vary the bloom_bits configuration. The probabilities
look like this:

1/2 chance of a uniform int from 0 to 19. This includes overall 1/40
chance of 0 which disables the bloom filter.

1/2 chance of a float from a lognormal distribution with a median of 10.
This always produces positive values but with a decent chance of < 1
(overall ~1/40) or > 100 (overall ~1/40), the enforced/coerced
implementation limits.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6103

Test Plan:
start 'make blackbox_crash_test' several times and look at
configuration output

Differential Revision: D18734877

Pulled By: pdillinger

fbshipit-source-id: 4a38cb057d3b3fc1327f93199f65b9a9ffbd7316
2019-12-10 08:39:50 -08:00
sdong
a960287dee db_stress: Some code style improvements (#6137)
Summary:
Two changes:
1. Prevent static variables in a header file
2. Add "override" keyword when virtual functions are overridden.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6137

Test Plan: Build db_stress with or without LITE.

Differential Revision: D18892007

fbshipit-source-id: 295356427a34473b23ed36d6ed4ef3ae35a32db0
2019-12-09 14:38:42 -08:00
sdong
7d79b32618 Break db_stress_tool.cc to a list of source files (#6134)
Summary:
db_stress_tool.cc now is a giant file. In order to main it easier to improve and maintain, break it down to multiple source files.
Most classes are turned into their own files. Separate .h and .cc files are created for gflag definiations. Another .h and .cc files are created for some common functions. Some test execution logic that is only loosely related to class StressTest is moved to db_stress_driver.h and db_stress_driver.cc. All the files are located under db_stress_tool/. The directory name is created as such because if we end it with either stress or test, .gitignore will ignore any file under it and makes it prone to issues in developements.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6134

Test Plan: Build under GCC7 with and without LITE on using GNU Make. Build with GCC 4.8. Build with cmake with -DWITH_TOOL=1

Differential Revision: D18876064

fbshipit-source-id: b25d0a7451840f31ac0f5ebb0068785f783fdf7d
2019-12-08 23:51:01 -08:00