Jin Hai dab74700b2
Delete and WAL feature branch merge (#1436)
* add read/write lock

* change compact to ddl queue

* add api to get vector data

* add flush / merge / compact lock

* add api to get vector data

* add data size for table info

* add db recovery test

* add data_size check

* change file name to uppercase

Signed-off-by: jinhai <hai.jin@zilliz.com>

* update wal flush_merge_compact_mutex_

* update wal flush_merge_compact_mutex_

* change requirement

* change requirement

* upd requirement

* add logging

* add logging

* add logging

* add logging

* add logging

* add logging

* add logging

* add logging

* add logging

* delete part

* add all size checks

* fix bug

* update faiss get_vector_by_id

* add get_vector case

* update get vector by id

* update server

* fix DBImpl

* attempting to fix #1268

* lint

* update unit test

* fix #1259

* issue 1271 fix wal config

* update

* fix cases

Signed-off-by: del.zhenwu <zhenxiang.li@zilliz.com>

* update read / write error message

* update read / write error message

* [skip ci] get vectors by id from raw files instead faiss

* [skip ci] update FilesByType meta

* update

* fix ci error

* update

* lint

* Hide partition_name parameter

* Remove douban pip source

Signed-off-by: zhenwu <zw@zilliz.com>

* Update epsilon value in test cases

Signed-off-by: zhenwu <zw@zilliz.com>

* Add default partition

* Caiyd crud (#1313)

* fix clang format

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix unittest build error

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* add faiss_bitset_test

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* avoid user directly operate partition table

* fix has table bug

* Caiyd crud (#1323)

* fix clang format

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix unittest build error

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* use compile option -O3

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* update faiss_bitset_test.cpp

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* change open flags

* change OngoingFileChecker to static instance

* mark ongoing files when applying deletes

* update clean up with ttl

* fix centos ci

* update

* lint

* update partition

Signed-off-by: zhenwu <zw@zilliz.com>

* update delete and flush to include partitions

* update

* Update cases

Signed-off-by: zhenwu <zw@zilliz.com>

* Fix test cases crud (#1350)

* fix order

* add wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix invalid operation issue

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix invalid operation issue

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix bug

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix bug

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* crud fix

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* crud fix

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* add table info test cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>
Signed-off-by: JinHai-CN <hai.jin@zilliz.com>

* merge cases

Signed-off-by: zhenwu <zw@zilliz.com>

* Shengjun (#1349)

* Add GPU sharing solution on native Kubernetes  (#1102)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Fix http server bug (#1096)

* refactoring(create_table done)

* refactoring

* refactor server delivery (insert done)

* refactoring server module (count_table done)

* server refactor done

* cmake pass

* refactor server module done.

* set grpc response status correctly

* format done.

* fix redefine ErrorMap()

* optimize insert reducing ids data copy

* optimize grpc request with reducing data copy

* clang format

* [skip ci] Refactor server module done. update changlog. prepare for PR

* remove explicit and change int32_t to int64_t

* add web server

* [skip ci] add license in web module

* modify header include & comment oatpp environment config

* add port configure & create table in handler

* modify web url

* simple url complation done & add swagger

* make sure web url

* web functionality done. debuging

* add web unittest

* web test pass

* add web server port

* add web server port in template

* update unittest cmake file

* change web server default port to 19121

* rename method in web module & unittest pass

* add search case in unittest for web module

* rename some variables

* fix bug

* unittest pass

* web prepare

* fix cmd bug(check server status)

* update changlog

* add web port validate & default set

* clang-format pass

* add web port test in unittest

* add CORS & redirect root to swagger ui

* add web status

* web table method func cascade test pass

* add config url in web module

* modify thirdparty cmake to avoid building oatpp test

* clang format

* update changlog

* add constants in web module

* reserve Config.cpp

* fix constants reference bug

* replace web server with async module

* modify component to support async

* format

* developing controller & add test clent into unittest

* add web port into demo/server_config

* modify thirdparty cmake to allow build test

* remove  unnecessary comment

* add endpoint info in controller

* finish web test(bug here)

* clang format

* add web test cpp to lint exclusions

* check null field in GetConfig

* add macro RETURN STATUS DTo

* fix cmake conflict

* fix crash when exit server

* remove surplus comments & add http param check

* add uri /docs to direct swagger

* format

* change cmd to system

* add default value & unittest in web module

* add macros to judge if GPU supported

* add macros in unit & add default in index dto & print error message when bind http port fail

* format (fix #788)

* fix cors bug (not completed)

* comment cors

* change web framework to simple api

* comments optimize

* change to simple API

* remove comments in controller.hpp

* remove EP_COMMON_CMAKE_ARGS in oatpp and oatpp-swagger

* add ep cmake args to sqlite

* clang-format

* change a format

* test pass

* change name to

* fix compiler issue(oatpp-swagger depend on oatpp)

* add & in start_server.h

* specify lib location with oatpp and oatpp-swagger

* add comments

* add swagger definition

* [skip ci] change http method options status code

* remove oatpp swagger(fix #970)

* remove comments

* check Start web behavior

* add default to cpu_cache_capacity

* remove swagger component.hpp & /docs url

* remove /docs info

* remove /docs in unittest

* remove space in test rpc

* remove repeate info in CHANGLOG

* change cache_insert_data default value as a constant

* [skip ci] Fix some broken links (#960)

* [skip ci] Fix broken link

* [skip ci] Fix broken link

* [skip ci] Fix broken link

* [skip ci] Fix broken links

* fix issue 373 (#964)

* fix issue 373

* Adjustment format

* Adjustment format

* Adjustment format

* change readme

* #966 update NOTICE.md (#967)

* remove comments

* check Start web behavior

* add default to cpu_cache_capacity

* remove swagger component.hpp & /docs url

* remove /docs info

* remove /docs in unittest

* remove space in test rpc

* remove repeate info in CHANGLOG

* change cache_insert_data default value as a constant

* adjust web port cofig place

* rename web_port variable

* change gpu resources invoke way to cmd()

* set advanced config name add DEFAULT

* change config setting to cmd

* modify ..

* optimize code

* assign TableDto' count default value 0 (fix #995)

* check if table exists when show partitions (fix #1028)

* check table exists when drop partition (fix #1029)

* check if partition name is legal (fix #1022)

* modify status code when partition tag is illegal

* update changlog

* add info to /system url

* add binary index and add bin uri & handler method(not completed)

* optimize http insert and search time(fix #1066) | add binary vectors support(fix #1067)

* fix test partition bug

* fix test bug when check insert records

* add binary vectors test

* add default for offset and page_size

* fix uinttest bug

* [skip ci] remove comments

* optimize web code for PR comments

* add new folder named utils

* check offset and pagesize (fix #1082)

* improve error message if offset or page_size is not legal (fix #1075)

* add log into web module

* update changlog

* check gpu sources setting when assign repeated value (fix #990)

* update changlog

* clang-format pass

* add default handler in http handler

* [skip ci] improve error msg when check gpu resources

* change check offset way

* remove func IsIntStr

* add case

* change int32 to int64 when check number str

* add log in we module(doing)

* update test case

* add log in web controller

Co-authored-by: jielinxu <52057195+jielinxu@users.noreply.github.com>
Co-authored-by: JackLCL <53512883+JackLCL@users.noreply.github.com>
Co-authored-by: Cai Yudong <yudong.cai@zilliz.com>

* Filtering for specific paths in Jenkins CI  (#1107)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Fix Filtering for specific paths in Jenkins CI bug (#1109)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Fix Filtering for specific paths in Jenkins CI bug (#1110)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Don't skip ci when triggered by a time (#1113)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Don't skip ci when triggered by a time

* Don't skip ci when triggered by a time

* Set default sending to Milvus Dev mail group  (#1121)

* run hadolint with reviewdog

* add LINCENSE in Dockerfile

* run hadolint with reviewdog

* Reporter of reviewdog command is "github-pr-check"

* format Dockerfile

* ignore DL3007 in hadolint

* clean up old docker images

* Add GPU sharing solution on native Kubernetes

* nightly test mailer

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Test filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* Filtering for specific paths in Jenkins CI

* No skip ci when triggered by a time

* Don't skip ci when triggered by a time

* Set default sending to Milvus Dev

* Support hnsw (#1131)

* add hnsw

* add config

* format...

* format..

* Remove test.template (#1129)

* Update framework

* remove files

* Remove files

* Remove ann-acc cases && Update java-sdk cases

* change cn to en

* [skip ci] remove doc test

* [skip ci] change cn to en

* Case stability

* Add mail notification when test failed

* Add main notification

* Add main notification

* gen milvus instance from utils

* Distable case with multiprocess

* Add mail notification when nightly test failed

* add milvus handler param

* add http handler

* Remove test.template

Co-authored-by: quicksilver <zhifeng.zhang@zilliz.com>

* Add doc for the RESTful API / Update contributor number in Milvus readme (#1100)

* [skip ci] Update contributor number.

* [skip ci] Add RESTful API doc.

* [skip ci] Some updates.

* [skip ci] Change port to 19121.

* [skip ci] Update README.md.

Update the descriptions for OPTIONS.

* Update README.md

Fix a typo.

* #1105 update error message when creating IVFSQ8H index without GPU resources (#1117)

* [skip ci] Update README (#1104)

* remove Nvidia owned files from faiss (#1136)

* #1135 remove Nvidia owned files from faiss

* Revert "#1135 remove Nvidia owned files from faiss"

This reverts commit 3bc007c28c8df5861fdd0452fd64c0e2e719eda2.

* #1135 remove Nvidia API implementation

* #1135 remove Nvidia owned files from faiss

* Update CODE_OF_CONDUCT.md (#1163)

* Improve codecov (#1095)

* Optimize config test. Dir src/config 99% lines covered

* add unittest coverage

* optimize cache&config unittest

* code format

* format

* format code

* fix merge conflict

* cover src/utils unittest

*  '#831 fix exe_path judge error'

* #831 fix exe_path judge error

* add some unittest coverage

* add some unittest coverage

* improve coverage of src/wrapper

* improve src/wrapper coverage

* *test optimize db/meta unittest

* fix bug

* *test optimize mysqlMetaImpl unittest

* *style: format code

* import server& scheduler unittest coverage

* handover next work

* *test: add some test_meta test case

* *format code

* *fix: fix typo

* feat(codecov): improve code coverage for src/db(#872)

* feat(codecov): improve code coverage for src/db/engine(#872)

* feat(codecov): improve code coverage(#872)

* fix config unittest bug

* feat(codecov): improve code coverage core/db/engine(#872)

* feat(codecov): improve code coverage core/knowhere

* feat(codecov): improve code coverage core/knowhere

* feat(codecov): improve code coverage

* feat(codecov): fix cpu test some error

* feat(codecov): improve code coverage

* feat(codecov): rename some fiu

* fix(db/meta): fix switch/case default action

* feat(codecov): improve code coverage(#872)
* fix error caused by merge code
* format code

* feat(codecov): improve code coverage & format code(#872)

* feat(codecov): fix test error(#872)

* feat(codecov): fix unittest test_mem(#872)

* feat(codecov): fix unittest(#872)

* feat(codecov): fix unittest for resource manager(#872)

* feat(codecov): code format (#872)

* feat(codecov): trigger ci(#872)

* fix(RequestScheduler): remove a wrong sleep statement

* test(test_rpc): fix rpc test

* Fix format issue

* Remove unused comments

* Fix unit test error

Co-authored-by: ABNER-1 <ABNER-1@users.noreply.github.com>
Co-authored-by: Jin Hai <hai.jin@zilliz.com>

* Support run dev test with http handler in python SDK (#1116)

* refactoring(create_table done)

* refactoring

* refactor server delivery (insert done)

* refactoring server module (count_table done)

* server refactor done

* cmake pass

* refactor server module done.

* set grpc response status correctly

* format done.

* fix redefine ErrorMap()

* optimize insert reducing ids data copy

* optimize grpc request with reducing data copy

* clang format

* [skip ci] Refactor server module done. update changlog. prepare for PR

* remove explicit and change int32_t to int64_t

* add web server

* [skip ci] add license in web module

* modify header include & comment oatpp environment config

* add port configure & create table in handler

* modify web url

* simple url complation done & add swagger

* make sure web url

* web functionality done. debuging

* add web unittest

* web test pass

* add web server port

* add web server port in template

* update unittest cmake file

* change web server default port to 19121

* rename method in web module & unittest pass

* add search case in unittest for web module

* rename some variables

* fix bug

* unittest pass

* web prepare

* fix cmd bug(check server status)

* update changlog

* add web port validate & default set

* clang-format pass

* add web port test in unittest

* add CORS & redirect root to swagger ui

* add web status

* web table method func cascade test pass

* add config url in web module

* modify thirdparty cmake to avoid building oatpp test

* clang format

* update changlog

* add constants in web module

* reserve Config.cpp

* fix constants reference bug

* replace web server with async module

* modify component to support async

* format

* developing controller & add test clent into unittest

* add web port into demo/server_config

* modify thirdparty cmake to allow build test

* remove  unnecessary comment

* add endpoint info in controller

* finish web test(bug here)

* clang format

* add web test cpp to lint exclusions

* check null field in GetConfig

* add macro RETURN STATUS DTo

* fix cmake conflict

* fix crash when exit server

* remove surplus comments & add http param check

* add uri /docs to direct swagger

* format

* change cmd to system

* add default value & unittest in web module

* add macros to judge if GPU supported

* add macros in unit & add default in index dto & print error message when bind http port fail

* format (fix #788)

* fix cors bug (not completed)

* comment cors

* change web framework to simple api

* comments optimize

* change to simple API

* remove comments in controller.hpp

* remove EP_COMMON_CMAKE_ARGS in oatpp and oatpp-swagger

* add ep cmake args to sqlite

* clang-format

* change a format

* test pass

* change name to

* fix compiler issue(oatpp-swagger depend on oatpp)

* add & in start_server.h

* specify lib location with oatpp and oatpp-swagger

* add comments

* add swagger definition

* [skip ci] change http method options status code

* remove oatpp swagger(fix #970)

* remove comments

* check Start web behavior

* add default to cpu_cache_capacity

* remove swagger component.hpp & /docs url

* remove /docs info

* remove /docs in unittest

* remove space in test rpc

* remove repeate info in CHANGLOG

* change cache_insert_data default value as a constant

* [skip ci] Fix some broken links (#960)

* [skip ci] Fix broken link

* [skip ci] Fix broken link

* [skip ci] Fix broken link

* [skip ci] Fix broken links

* fix issue 373 (#964)

* fix issue 373

* Adjustment format

* Adjustment format

* Adjustment format

* change readme

* #966 update NOTICE.md (#967)

* remove comments

* check Start web behavior

* add default to cpu_cache_capacity

* remove swagger component.hpp & /docs url

* remove /docs info

* remove /docs in unittest

* remove space in test rpc

* remove repeate info in CHANGLOG

* change cache_insert_data default value as a constant

* adjust web port cofig place

* rename web_port variable

* change gpu resources invoke way to cmd()

* set advanced config name add DEFAULT

* change config setting to cmd

* modify ..

* optimize code

* assign TableDto' count default value 0 (fix #995)

* check if table exists when show partitions (fix #1028)

* check table exists when drop partition (fix #1029)

* check if partition name is legal (fix #1022)

* modify status code when partition tag is illegal

* update changlog

* add info to /system url

* add binary index and add bin uri & handler method(not completed)

* optimize http insert and search time(fix #1066) | add binary vectors support(fix #1067)

* fix test partition bug

* fix test bug when check insert records

* add binary vectors test

* add default for offset and page_size

* fix uinttest bug

* [skip ci] remove comments

* optimize web code for PR comments

* add new folder named utils

* check offset and pagesize (fix #1082)

* improve error message if offset or page_size is not legal (fix #1075)

* add log into web module

* update changlog

* check gpu sources setting when assign repeated value (fix #990)

* update changlog

* clang-format pass

* add default handler in http handler

* [skip ci] improve error msg when check gpu resources

* change check offset way

* remove func IsIntStr

* add case

* change int32 to int64 when check number str

* add log in we module(doing)

* update test case

* add log in web controller

* remove surplus dot

* add preload into /system/

* change get_milvus() to get_milvus(args['handler'])

* support load table into memory with http server (fix #1115)

* [skip ci] comment surplus dto in VectorDto

Co-authored-by: jielinxu <52057195+jielinxu@users.noreply.github.com>
Co-authored-by: JackLCL <53512883+JackLCL@users.noreply.github.com>
Co-authored-by: Cai Yudong <yudong.cai@zilliz.com>

* Fix #1140 (#1162)

* fix

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* update...

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* fix2

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* fix3

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* update changelog

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* Update INSTALL.md (#1175)

* Update INSTALL.md

1. Change image tag and Milvus source code to latest.
2. Fix a typo

Signed-off-by: Lu Wang <yamasite@qq.com>

* Update INSTALL.md

Signed-off-by: lu.wang <yamasite@qq.com>

* add Tanimoto ground truth (#1138)

* add milvus ground truth

* add milvus groundtruth

* [skip ci] add milvus ground truth

* [skip ci]add tanimoto ground truth

* fix mix case bug (#1208)

* fix mix case bug

Signed-off-by: del.zhenwu <zhenxiang.li@zilliz.com>

* Remove case.md

Signed-off-by: del.zhenwu <zhenxiang.li@zilliz.com>

* Update README.md (#1206)

Add LFAI mailing lists.

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Add design.md to store links to design docs (#1219)

* Update README.md

Add link to Milvus design docs

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Create design.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update design.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Add troubleshooting info about libmysqlpp.so.3 error (#1225)

* Update INSTALL.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update INSTALL.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update README.md (#1233)

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* #1240 Update license declaration of each file (#1241)

* #1240 Update license declaration of each files

Signed-off-by: jinhai <hai.jin@zilliz.com>

* #1240 Update CHANGELOG

Signed-off-by: jinhai <hai.jin@zilliz.com>

* Update README.md (#1258)

Add Jenkins master badge.

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update INSTALL.md (#1265)

Fix indentation.

* support CPU profiling (#1251)

* #1250 support CPU profiling

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* #1250 fix code coverage

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* Fix HNSW crash (#1262)

* fix

Signed-off-by: xiaojun.lin <xiaojun.lin@zilliz.com>

* update.

Signed-off-by: xiaojun.lin <xiaojun.lin@zilliz.com>

* Add troubleshooting information for INSTALL.md and enhance readability (#1274)

* Update INSTALL.md

1. Add new troubleshooting message;
2. Enhance readability.

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update INSTALL.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update INSTALL.md

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Update INSTALL.md

Add CentOS link.

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* Create COMMUNITY.md (#1292)

Signed-off-by: Lutkin Wang <yamasite@qq.com>

* fix gtest

* add copyright

* fix gtest

* MERGE_NOT_YET

* fix lint

Co-authored-by: quicksilver <zhifeng.zhang@zilliz.com>
Co-authored-by: BossZou <40255591+BossZou@users.noreply.github.com>
Co-authored-by: jielinxu <52057195+jielinxu@users.noreply.github.com>
Co-authored-by: JackLCL <53512883+JackLCL@users.noreply.github.com>
Co-authored-by: Cai Yudong <yudong.cai@zilliz.com>
Co-authored-by: Tinkerrr <linxiaojun.cn@outlook.com>
Co-authored-by: del-zhenwu <56623710+del-zhenwu@users.noreply.github.com>
Co-authored-by: Lutkin Wang <yamasite@qq.com>
Co-authored-by: shengjh <46514371+shengjh@users.noreply.github.com>
Co-authored-by: ABNER-1 <ABNER-1@users.noreply.github.com>
Co-authored-by: Jin Hai <hai.jin@zilliz.com>
Co-authored-by: shiyu22 <cshiyu22@gmail.com>

* #1302 Get all record IDs in a segment by given a segment id

* Remove query time ranges

Signed-off-by: zhenwu <zw@zilliz.com>

* #1295 let wal enable by default

* fix cases

Signed-off-by: zhenwu <zw@zilliz.com>

* fix partition cases

Signed-off-by: zhenwu <zw@zilliz.com>

* [skip ci] update test_db

* update

* fix case bug

Signed-off-by: zhenwu <zw@zilliz.com>

* lint

* fix test case failures

* remove some code

* Caiyd crud 1 (#1377)

* fix clang format

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix unittest build error

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix build issue when enable profiling

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix hastable bug

* update bloom filter

* update

* benchmark

* update benchmark

* update

* update

* remove wal record size

Signed-off-by: shengjun.li <shengjun.li@zilliz.com>

* remove wal record size config

Signed-off-by: shengjun.li <shengjun.li@zilliz.com>

* update apply deletes: switch to binary search

* update sdk_simple

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* update apply deletes: switch to binary search

* add test_search_by_id

Signed-off-by: zhenwu <zw@zilliz.com>

* add more log

* flush error with multi same ids

Signed-off-by: zhenwu <zw@zilliz.com>

* modify wal config

Signed-off-by: shengjun.li <shengjun.li@zilliz.com>

* update

* add binary search_by_id

* fix case bug

Signed-off-by: zhenwu <zw@zilliz.com>

* update cases

Signed-off-by: zhenwu <zw@zilliz.com>

* fix unit test #1395

* improve merge performance

* add uids_ for VectorIndex to improve search performance

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix error

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* update

* fix search

* fix record num

Signed-off-by: shengjun.li <shengjun.li@zilliz.com>

* refine code

* refine code

* Add get_vector_ids test cases (#1407)

* fix order

* add wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix wal case

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix invalid operation issue

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix invalid operation issue

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix bug

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* fix bug

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* crud fix

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* crud fix

Signed-off-by: sahuang <xiaohaix@student.unimelb.edu.au>

* add table info test cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>
Signed-off-by: JinHai-CN <hai.jin@zilliz.com>

* add to compact case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* add to compact case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* add to compact case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* add case and debug compact

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* test pdb

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* test pdb

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* test pdb

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix cases

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update table_info case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update table_info case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update table_info case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update get vector ids case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update get vector ids case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update get vector ids case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update get vector ids case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* update case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* pdb test

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* pdb test

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* add tests for get_vector_ids

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix case

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* add binary and ip

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix binary index

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* fix pdb

Signed-off-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>

* #1408 fix search result in-correct after DeleteById

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* add one case

* delete failed segment

* update serialize

* update serialize

* fix case

Signed-off-by: zhenwu <zw@zilliz.com>

* update

* update case assertion

Signed-off-by: zhenwu <zw@zilliz.com>

* [skip ci] update config

* change bloom filter msync flag to async

* #1319 add more timing debug info

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* update

* update

* add normalize

Signed-off-by: zhenwu <zw@zilliz.com>

* add normalize

Signed-off-by: zhenwu <zw@zilliz.com>

* add normalize

Signed-off-by: zhenwu <zw@zilliz.com>

* Fix compiling error

Signed-off-by: jinhai <hai.jin@zilliz.com>

* support ip (#1383)

* support ip

Signed-off-by: xiaojun.lin <xiaojun.lin@zilliz.com>

* IP result distance sort by descend

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* update

Signed-off-by: Nicky <nicky.xj.lin@gmail.com>

* format

Signed-off-by: xiaojun.lin <xiaojun.lin@zilliz.com>

* get table lsn

* Remove unused third party

Signed-off-by: jinhai <hai.jin@zilliz.com>

* Refine code

Signed-off-by: jinhai <hai.jin@zilliz.com>

* #1319 fix clang format

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* fix wal applied lsn

Signed-off-by: shengjun.li <shengjun.li@zilliz.com>

* validate partition tag

* #1319 improve search performance

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

* build error

Co-authored-by: Zhiru Zhu <youny626@hotmail.com>
Co-authored-by: groot <yihua.mo@zilliz.com>
Co-authored-by: Xiaohai Xu <xiaohaix@student.unimelb.edu.au>
Co-authored-by: shengjh <46514371+shengjh@users.noreply.github.com>
Co-authored-by: del-zhenwu <56623710+del-zhenwu@users.noreply.github.com>
Co-authored-by: shengjun.li <49774184+shengjun1985@users.noreply.github.com>
Co-authored-by: Cai Yudong <yudong.cai@zilliz.com>
Co-authored-by: quicksilver <zhifeng.zhang@zilliz.com>
Co-authored-by: BossZou <40255591+BossZou@users.noreply.github.com>
Co-authored-by: jielinxu <52057195+jielinxu@users.noreply.github.com>
Co-authored-by: JackLCL <53512883+JackLCL@users.noreply.github.com>
Co-authored-by: Tinkerrr <linxiaojun.cn@outlook.com>
Co-authored-by: Lutkin Wang <yamasite@qq.com>
Co-authored-by: ABNER-1 <ABNER-1@users.noreply.github.com>
Co-authored-by: shiyu22 <cshiyu22@gmail.com>
2020-02-29 16:11:31 +08:00

1566 lines
66 KiB
Python

"""
For testing index operations, including `create_index`, `describe_index` and `drop_index` interfaces
"""
import logging
import pytest
import time
import pdb
import threading
from multiprocessing import Pool, Process
import numpy
import sklearn.preprocessing
from milvus import IndexType, MetricType
from utils import *
nb = 6000
dim = 128
index_file_size = 10
vectors = gen_vectors(nb, dim)
vectors = sklearn.preprocessing.normalize(vectors, axis=1, norm='l2')
vectors = vectors.tolist()
BUILD_TIMEOUT = 300
nprobe = 1
tag = "1970-01-01"
NLIST = 16384
class TestIndexBase:
@pytest.fixture(
scope="function",
params=gen_index_params()
)
def get_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_simple_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip PQ Temporary")
return request.param
"""
******************************************************************
The following cases are used to test `create_index` function
******************************************************************
"""
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index(self, connect, table, get_simple_index_params):
'''
target: test create index interface
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
if index_params["index_type"] == IndexType.IVF_SQ8:
status, ids = connect.add_vectors(table, vectors)
status = connect.create_index(table, index_params)
assert status.OK()
else:
pytest.skip("skip other index types")
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_no_vectors(self, connect, table, get_simple_index_params):
'''
target: test create index interface
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
status = connect.create_index(table, index_params)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_partition(self, connect, table, get_simple_index_params):
'''
target: test create index interface
method: create table, create partition, and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status = connect.create_partition(table, tag)
status, ids = connect.add_vectors(table, vectors, partition_tag=tag)
status = connect.create_index(table, index_params)
assert status.OK()
@pytest.mark.level(2)
def test_create_index_without_connect(self, dis_connect, table):
'''
target: test create index without connection
method: create table and add vectors in it, check if added successfully
expected: raise exception
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
with pytest.raises(Exception) as e:
status = dis_connect.create_index(table, index_param)
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_search_with_query_vectors(self, connect, table, get_simple_index_params):
'''
target: test create index interface, search with more query vectors
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(table, vectors)
status = connect.create_index(table, index_params)
logging.getLogger().info(connect.describe_index(table))
query_vecs = [vectors[0], vectors[1], vectors[2]]
top_k = 5
status, result = connect.search_vectors(table, top_k, nprobe, query_vecs)
assert status.OK()
assert len(result) == len(query_vecs)
logging.getLogger().info(result)
# TODO: enable
@pytest.mark.timeout(BUILD_TIMEOUT)
@pytest.mark.level(2)
def _test_create_index_multiprocessing(self, connect, table, args):
'''
target: test create index interface with multiprocess
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
status, ids = connect.add_vectors(table, vectors)
def build(connect):
status = connect.create_index(table)
assert status.OK()
process_num = 8
processes = []
uri = "tcp://%s:%s" % (args["ip"], args["port"])
for i in range(process_num):
m = get_milvus(args["handler"])
m.connect(uri=uri)
p = Process(target=build, args=(m,))
processes.append(p)
p.start()
time.sleep(0.2)
for p in processes:
p.join()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(table, top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
assert result[0][0].distance == 0.0
# TODO: enable
@pytest.mark.timeout(BUILD_TIMEOUT)
def _test_create_index_multiprocessing_multitable(self, connect, args):
'''
target: test create index interface with multiprocess
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
process_num = 8
loop_num = 8
processes = []
table = []
j = 0
while j < (process_num*loop_num):
table_name = gen_unique_str("test_create_index_multiprocessing")
table.append(table_name)
param = {'table_name': table_name,
'dimension': dim,
'index_type': IndexType.FLAT,
'store_raw_vector': False}
connect.create_table(param)
j = j + 1
def create_index():
i = 0
while i < loop_num:
# assert connect.has_table(table[ids*process_num+i])
status, ids = connect.add_vectors(table[ids*process_num+i], vectors)
status = connect.create_index(table[ids*process_num+i])
assert status.OK()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(table[ids*process_num+i], top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
assert result[0][0].distance == 0.0
i = i + 1
uri = "tcp://%s:%s" % (args["ip"], args["port"])
for i in range(process_num):
m = get_milvus(args["handler"])
m.connect(uri=uri)
ids = i
p = Process(target=create_index, args=(m,ids))
processes.append(p)
p.start()
time.sleep(0.2)
for p in processes:
p.join()
def test_create_index_table_not_existed(self, connect):
'''
target: test create index interface when table name not existed
method: create table and add vectors in it, create index
, make sure the table name not in index
expected: return code not equals to 0, create index failed
'''
table_name = gen_unique_str(self.__class__.__name__)
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status = connect.create_index(table_name, index_param)
assert not status.OK()
def test_create_index_table_None(self, connect):
'''
target: test create index interface when table name is None
method: create table and add vectors in it, create index with an table_name: None
expected: return code not equals to 0, create index failed
'''
table_name = None
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
with pytest.raises(Exception) as e:
status = connect.create_index(table_name, index_param)
def test_create_index_no_vectors(self, connect, table):
'''
target: test create index interface when there is no vectors in table
method: create table and add no vectors in it, and then create index
expected: return code equals to 0
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status = connect.create_index(table, index_param)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_no_vectors_then_add_vectors(self, connect, table, get_simple_index_params):
'''
target: test create index interface when there is no vectors in table, and does not affect the subsequent process
method: create table and add no vectors in it, and then create index, add vectors in it
expected: return code equals to 0
'''
index_param = get_simple_index_params
if index_param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status = connect.create_index(table, index_param)
status, ids = connect.add_vectors(table, vectors)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_same_index_repeatedly(self, connect, table, get_simple_index_params):
'''
target: check if index can be created repeatedly, with the same create_index params
method: create index after index have been built
expected: return code success, and search ok
'''
status, ids = connect.add_vectors(table, vectors)
index_param = get_simple_index_params
if index_param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status = connect.create_index(table, index_param)
status = connect.create_index(table, index_param)
assert status.OK()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(table, top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_different_index_repeatedly(self, connect, table):
'''
target: check if index can be created repeatedly, with the different create_index params
method: create another index with different index_params after index have been built
expected: return code 0, and describe index result equals with the second index params
'''
nlist = NLIST
status, ids = connect.add_vectors(table, vectors)
index_type_1 = IndexType.IVF_SQ8
index_type_2 = IndexType.IVFLAT
index_params = [{"index_type": index_type_1, "nlist": nlist}, {"index_type": index_type_2, "nlist": nlist}]
logging.getLogger().info(index_params)
for index_param in index_params:
status = connect.create_index(table, index_param)
assert status.OK()
status, result = connect.describe_index(table)
assert result._nlist == nlist
assert result._table_name == table
assert result._index_type == index_type_2
"""
******************************************************************
The following cases are used to test `describe_index` function
******************************************************************
"""
def test_describe_index(self, connect, table, get_index_params):
'''
target: test describe index interface
method: create table and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_index_params
logging.getLogger().info(index_params)
# status, ids = connect.add_vectors(table, vectors)
status = connect.create_index(table, index_params)
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == table
assert result._index_type == index_params["index_type"]
def test_describe_and_drop_index_multi_tables(self, connect, get_simple_index_params):
'''
target: test create, describe and drop index interface with multiple tables of L2
method: create tables and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
nq = 100
vectors = gen_vectors(nq, dim)
table_list = []
for i in range(10):
table_name = gen_unique_str()
table_list.append(table_name)
param = {'table_name': table_name,
'dimension': dim,
'index_file_size': index_file_size,
'metric_type': MetricType.L2}
connect.create_table(param)
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(table_name=table_name, records=vectors)
status = connect.create_index(table_name, index_params)
assert status.OK()
for i in range(10):
status, result = connect.describe_index(table_list[i])
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == table_list[i]
assert result._index_type == index_params["index_type"]
for i in range(10):
status = connect.drop_index(table_list[i])
assert status.OK()
status, result = connect.describe_index(table_list[i])
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table_list[i]
assert result._index_type == IndexType.FLAT
@pytest.mark.level(2)
def test_describe_index_without_connect(self, dis_connect, table):
'''
target: test describe index without connection
method: describe index, and check if describe successfully
expected: raise exception
'''
with pytest.raises(Exception) as e:
status = dis_connect.describe_index(table)
def test_describe_index_table_not_existed(self, connect):
'''
target: test describe index interface when table name not existed
method: create table and add vectors in it, create index
, make sure the table name not in index
expected: return code not equals to 0, describe index failed
'''
table_name = gen_unique_str(self.__class__.__name__)
status, result = connect.describe_index(table_name)
assert not status.OK()
def test_describe_index_table_None(self, connect):
'''
target: test describe index interface when table name is None
method: create table and add vectors in it, create index with an table_name: None
expected: return code not equals to 0, describe index failed
'''
table_name = None
with pytest.raises(Exception) as e:
status = connect.describe_index(table_name)
def test_describe_index_not_create(self, connect, table):
'''
target: test describe index interface when index not created
method: create table and add vectors in it, create index
, make sure the table name not in index
expected: return code not equals to 0, describe index failed
'''
status, ids = connect.add_vectors(table, vectors)
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert status.OK()
# assert result._nlist == index_params["nlist"]
# assert result._table_name == table
# assert result._index_type == index_params["index_type"]
"""
******************************************************************
The following cases are used to test `drop_index` function
******************************************************************
"""
def test_drop_index(self, connect, table, get_simple_index_params):
'''
target: test drop index interface
method: create table and add vectors in it, create index, call drop index
expected: return code 0, and default index param
'''
index_param = get_simple_index_params
# status, ids = connect.add_vectors(table, vectors)
status = connect.create_index(table, index_param)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
status = connect.drop_index(table)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table
assert result._index_type == IndexType.FLAT
def test_drop_index_repeatly(self, connect, table, get_simple_index_params):
'''
target: test drop index repeatly
method: create index, call drop index, and drop again
expected: return code 0
'''
index_param = get_simple_index_params
# status, ids = connect.add_vectors(table, vectors)
status = connect.create_index(table, index_param)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
status = connect.drop_index(table)
assert status.OK()
status = connect.drop_index(table)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table
assert result._index_type == IndexType.FLAT
@pytest.mark.level(2)
def test_drop_index_without_connect(self, dis_connect, table):
'''
target: test drop index without connection
method: drop index, and check if drop successfully
expected: raise exception
'''
with pytest.raises(Exception) as e:
status = dis_connect.drop_index(table)
def test_drop_index_table_not_existed(self, connect):
'''
target: test drop index interface when table name not existed
method: create table and add vectors in it, create index
, make sure the table name not in index, and then drop it
expected: return code not equals to 0, drop index failed
'''
table_name = gen_unique_str(self.__class__.__name__)
status = connect.drop_index(table_name)
assert not status.OK()
def test_drop_index_table_None(self, connect):
'''
target: test drop index interface when table name is None
method: create table and add vectors in it, create index with an table_name: None
expected: return code not equals to 0, drop index failed
'''
table_name = None
with pytest.raises(Exception) as e:
status = connect.drop_index(table_name)
def test_drop_index_table_not_create(self, connect, table):
'''
target: test drop index interface when index not created
method: create table and add vectors in it, create index
expected: return code not equals to 0, drop index failed
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status, ids = connect.add_vectors(table, vectors)
status, result = connect.describe_index(table)
logging.getLogger().info(result)
# no create index
status = connect.drop_index(table)
logging.getLogger().info(status)
assert status.OK()
def test_create_drop_index_repeatly(self, connect, table, get_simple_index_params):
'''
target: test create / drop index repeatly, use the same index params
method: create index, drop index, four times
expected: return code 0
'''
index_params = get_simple_index_params
# status, ids = connect.add_vectors(table, vectors)
for i in range(2):
status = connect.create_index(table, index_params)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
status = connect.drop_index(table)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table
assert result._index_type == IndexType.FLAT
def test_create_drop_index_repeatly_different_index_params(self, connect, table):
'''
target: test create / drop index repeatly, use the different index params
method: create index, drop index, four times, each tme use different index_params to create index
expected: return code 0
'''
nlist = NLIST
index_params = [{"index_type": IndexType.IVFLAT, "nlist": nlist}, {"index_type": IndexType.IVF_SQ8, "nlist": nlist}]
# status, ids = connect.add_vectors(table, vectors)
for i in range(2):
status = connect.create_index(table, index_params[i])
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
status = connect.drop_index(table)
assert status.OK()
status, result = connect.describe_index(table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table
assert result._index_type == IndexType.FLAT
class TestIndexIP:
@pytest.fixture(
scope="function",
params=gen_index_params()
)
def get_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_simple_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip PQ Temporary")
return request.param
"""
******************************************************************
The following cases are used to test `create_index` function
******************************************************************
"""
@pytest.mark.level(2)
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index(self, connect, ip_table, get_simple_index_params):
'''
target: test create index interface
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ip_table, index_params)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_table(self, connect, ip_table, get_simple_index_params):
'''
target: test create index interface
method: create table, create partition, and add vectors in it, create index on table
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status = connect.create_partition(ip_table, tag)
status, ids = connect.add_vectors(ip_table, vectors, partition_tag=tag)
status = connect.create_index(ip_table, index_params)
assert status.OK()
@pytest.mark.level(2)
def test_create_index_without_connect(self, dis_connect, ip_table):
'''
target: test create index without connection
method: create table and add vectors in it, check if added successfully
expected: raise exception
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
with pytest.raises(Exception) as e:
status = dis_connect.create_index(ip_table, index_param)
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_search_with_query_vectors(self, connect, ip_table, get_simple_index_params):
'''
target: test create index interface, search with more query vectors
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ip_table, index_params)
logging.getLogger().info(connect.describe_index(ip_table))
query_vecs = [vectors[0], vectors[1], vectors[2]]
top_k = 5
status, result = connect.search_vectors(ip_table, top_k, nprobe, query_vecs)
logging.getLogger().info(result)
assert status.OK()
assert len(result) == len(query_vecs)
# TODO: enable
@pytest.mark.timeout(BUILD_TIMEOUT)
@pytest.mark.level(2)
def _test_create_index_multiprocessing(self, connect, ip_table, args):
'''
target: test create index interface with multiprocess
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
status, ids = connect.add_vectors(ip_table, vectors)
def build(connect):
status = connect.create_index(ip_table)
assert status.OK()
process_num = 8
processes = []
uri = "tcp://%s:%s" % (args["ip"], args["port"])
for i in range(process_num):
m = get_milvus(args["handler"])
m.connect(uri=uri)
p = Process(target=build, args=(m,))
processes.append(p)
p.start()
time.sleep(0.2)
for p in processes:
p.join()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(ip_table, top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
assert result[0][0].distance == 0.0
# TODO: enable
@pytest.mark.timeout(BUILD_TIMEOUT)
def _test_create_index_multiprocessing_multitable(self, connect, args):
'''
target: test create index interface with multiprocess
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
process_num = 8
loop_num = 8
processes = []
table = []
j = 0
while j < (process_num*loop_num):
table_name = gen_unique_str("test_create_index_multiprocessing")
table.append(table_name)
param = {'table_name': table_name,
'dimension': dim}
connect.create_table(param)
j = j + 1
def create_index():
i = 0
while i < loop_num:
# assert connect.has_table(table[ids*process_num+i])
status, ids = connect.add_vectors(table[ids*process_num+i], vectors)
status = connect.create_index(table[ids*process_num+i])
assert status.OK()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(table[ids*process_num+i], top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
assert result[0][0].distance == 0.0
i = i + 1
uri = "tcp://%s:%s" % (args["ip"], args["port"])
for i in range(process_num):
m = get_milvus(args["handler"])
m.connect(uri=uri)
ids = i
p = Process(target=create_index, args=(m,ids))
processes.append(p)
p.start()
time.sleep(0.2)
for p in processes:
p.join()
def test_create_index_no_vectors(self, connect, ip_table):
'''
target: test create index interface when there is no vectors in table
method: create table and add no vectors in it, and then create index
expected: return code equals to 0
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status = connect.create_index(ip_table, index_param)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_no_vectors_then_add_vectors(self, connect, ip_table, get_simple_index_params):
'''
target: test create index interface when there is no vectors in table, and does not affect the subsequent process
method: create table and add no vectors in it, and then create index, add vectors in it
expected: return code equals to 0
'''
index_param = get_simple_index_params
if index_param["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status = connect.create_index(ip_table, index_param)
status, ids = connect.add_vectors(ip_table, vectors)
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_same_index_repeatedly(self, connect, ip_table):
'''
target: check if index can be created repeatedly, with the same create_index params
method: create index after index have been built
expected: return code success, and search ok
'''
nlist = NLIST
status, ids = connect.add_vectors(ip_table, vectors)
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status = connect.create_index(ip_table, index_param)
status = connect.create_index(ip_table, index_param)
assert status.OK()
query_vec = [vectors[0]]
top_k = 1
status, result = connect.search_vectors(ip_table, top_k, nprobe, query_vec)
assert len(result) == 1
assert len(result[0]) == top_k
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_different_index_repeatedly(self, connect, ip_table):
'''
target: check if index can be created repeatedly, with the different create_index params
method: create another index with different index_params after index have been built
expected: return code 0, and describe index result equals with the second index params
'''
nlist = NLIST
status, ids = connect.add_vectors(ip_table, vectors)
index_type_1 = IndexType.IVF_SQ8
index_type_2 = IndexType.IVFLAT
index_params = [{"index_type": index_type_1, "nlist": nlist}, {"index_type": index_type_2, "nlist": nlist}]
logging.getLogger().info(index_params)
for index_param in index_params:
status = connect.create_index(ip_table, index_param)
assert status.OK()
status, result = connect.describe_index(ip_table)
assert result._nlist == nlist
assert result._table_name == ip_table
assert result._index_type == index_type_2
"""
******************************************************************
The following cases are used to test `describe_index` function
******************************************************************
"""
def test_describe_index(self, connect, ip_table, get_simple_index_params):
'''
target: test describe index interface
method: create table and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_simple_index_params
logging.getLogger().info(index_params)
# status, ids = connect.add_vectors(ip_table, vectors[:5000])
status = connect.create_index(ip_table, index_params)
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._table_name == ip_table
status, mode = connect._cmd("mode")
if str(mode) == "GPU" and index_params["index_type"] == IndexType.IVF_PQ:
assert result._index_type == IndexType.FLAT
assert result._nlist == NLIST
else:
assert result._index_type == index_params["index_type"]
assert result._nlist == index_params["nlist"]
def test_describe_index_partition(self, connect, ip_table, get_simple_index_params):
'''
target: test describe index interface
method: create table, create partition and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status = connect.create_partition(ip_table, tag)
status, ids = connect.add_vectors(ip_table, vectors, partition_tag=tag)
status = connect.create_index(ip_table, index_params)
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == ip_table
assert result._index_type == index_params["index_type"]
def test_describe_index_partition_A(self, connect, ip_table, get_simple_index_params):
'''
target: test describe index interface
method: create table, create partitions and add vectors in it, create index on partitions, call describe index
expected: return code 0, and index instructure
'''
new_tag = "new_tag"
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status = connect.create_partition(ip_table, tag)
status = connect.create_partition(ip_table, new_tag)
status, ids = connect.add_vectors(ip_table, vectors, partition_tag=tag)
status, ids = connect.add_vectors(ip_table, vectors, partition_tag=new_tag)
status = connect.create_index(ip_table, index_params)
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == ip_table
assert result._index_type == index_params["index_type"]
def test_describe_and_drop_index_multi_tables(self, connect, get_simple_index_params):
'''
target: test create, describe and drop index interface with multiple tables of IP
method: create tables and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
nq = 100
vectors = gen_vectors(nq, dim)
table_list = []
for i in range(10):
table_name = gen_unique_str()
table_list.append(table_name)
param = {'table_name': table_name,
'dimension': dim,
'index_file_size': index_file_size,
'metric_type': MetricType.IP}
connect.create_table(param)
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(table_name=table_name, records=vectors)
status = connect.create_index(table_name, index_params)
assert status.OK()
for i in range(10):
status, result = connect.describe_index(table_list[i])
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == table_list[i]
assert result._index_type == index_params["index_type"]
for i in range(10):
status = connect.drop_index(table_list[i])
assert status.OK()
status, result = connect.describe_index(table_list[i])
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == table_list[i]
assert result._index_type == IndexType.FLAT
@pytest.mark.level(2)
def test_describe_index_without_connect(self, dis_connect, ip_table):
'''
target: test describe index without connection
method: describe index, and check if describe successfully
expected: raise exception
'''
with pytest.raises(Exception) as e:
status = dis_connect.describe_index(ip_table)
def test_describe_index_not_create(self, connect, ip_table):
'''
target: test describe index interface when index not created
method: create table and add vectors in it, create index
, make sure the table name not in index
expected: return code not equals to 0, describe index failed
'''
status, ids = connect.add_vectors(ip_table, vectors)
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert status.OK()
# assert result._nlist == index_params["nlist"]
# assert result._table_name == table
# assert result._index_type == index_params["index_type"]
"""
******************************************************************
The following cases are used to test `drop_index` function
******************************************************************
"""
def test_drop_index(self, connect, ip_table, get_simple_index_params):
'''
target: test drop index interface
method: create table and add vectors in it, create index, call drop index
expected: return code 0, and default index param
'''
index_params = get_simple_index_params
status, mode = connect._cmd("mode")
assert status.OK()
# status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ip_table, index_params)
if str(mode) == "GPU" and (index_params["index_type"] == IndexType.IVF_PQ):
assert not status.OK()
else:
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
def test_drop_index_partition(self, connect, ip_table, get_simple_index_params):
'''
target: test drop index interface
method: create table, create partition and add vectors in it, create index on table, call drop table index
expected: return code 0, and default index param
'''
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status = connect.create_partition(ip_table, tag)
status, ids = connect.add_vectors(ip_table, vectors, partition_tag=tag)
status = connect.create_index(ip_table, index_params)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
def test_drop_index_partition_C(self, connect, ip_table, get_simple_index_params):
'''
target: test drop index interface
method: create table, create partitions and add vectors in it, create index on partitions, call drop partition index
expected: return code 0, and default index param
'''
new_tag = "new_tag"
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status = connect.create_partition(ip_table, tag)
status = connect.create_partition(ip_table, new_tag)
status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ip_table, index_params)
assert status.OK()
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
def test_drop_index_repeatly(self, connect, ip_table, get_simple_index_params):
'''
target: test drop index repeatly
method: create index, call drop index, and drop again
expected: return code 0
'''
index_params = get_simple_index_params
# status, ids = connect.add_vectors(ip_table, vectors)
status, mode = connect._cmd("mode")
assert status.OK()
# status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ip_table, index_params)
if str(mode) == "GPU" and (index_params["index_type"] == IndexType.IVF_PQ):
assert not status.OK()
else:
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
status = connect.drop_index(ip_table)
assert status.OK()
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
@pytest.mark.level(2)
def test_drop_index_without_connect(self, dis_connect, ip_table):
'''
target: test drop index without connection
method: drop index, and check if drop successfully
expected: raise exception
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVFLAT, "nlist": nlist}
with pytest.raises(Exception) as e:
status = dis_connect.drop_index(ip_table, index_param)
def test_drop_index_table_not_create(self, connect, ip_table):
'''
target: test drop index interface when index not created
method: create table and add vectors in it, create index
expected: return code not equals to 0, drop index failed
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
logging.getLogger().info(index_param)
status, ids = connect.add_vectors(ip_table, vectors)
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
# no create index
status = connect.drop_index(ip_table)
logging.getLogger().info(status)
assert status.OK()
def test_create_drop_index_repeatly(self, connect, ip_table, get_simple_index_params):
'''
target: test create / drop index repeatly, use the same index params
method: create index, drop index, four times
expected: return code 0
'''
index_params = get_simple_index_params
if index_params["index_type"] == IndexType.IVF_PQ:
pytest.skip("Skip some PQ cases")
status, ids = connect.add_vectors(ip_table, vectors)
for i in range(2):
status = connect.create_index(ip_table, index_params)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
def test_create_drop_index_repeatly_different_index_params(self, connect, ip_table):
'''
target: test create / drop index repeatly, use the different index params
method: create index, drop index, four times, each tme use different index_params to create index
expected: return code 0
'''
nlist = NLIST
index_params = [{"index_type": IndexType.IVFLAT, "nlist": nlist}, {"index_type": IndexType.IVF_SQ8, "nlist": nlist}]
status, ids = connect.add_vectors(ip_table, vectors)
for i in range(2):
status = connect.create_index(ip_table, index_params[i])
assert status.OK()
status, result = connect.describe_index(ip_table)
assert result._nlist == index_params[i]["nlist"]
assert result._table_name == ip_table
assert result._index_type == index_params[i]["index_type"]
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
status = connect.drop_index(ip_table)
assert status.OK()
status, result = connect.describe_index(ip_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ip_table
assert result._index_type == IndexType.FLAT
class TestIndexJAC:
tmp, vectors = gen_binary_vectors(nb, dim)
@pytest.fixture(
scope="function",
params=gen_index_params()
)
def get_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ or request.param["index_type"] == IndexType.HNSW:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_simple_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ or request.param["index_type"] == IndexType.HNSW:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_jaccard_index_params(self, request, connect):
logging.getLogger().info(request.param)
if request.param["index_type"] == IndexType.IVFLAT or request.param["index_type"] == IndexType.FLAT:
return request.param
else:
pytest.skip("Skip index Temporary")
"""
******************************************************************
The following cases are used to test `create_index` function
******************************************************************
"""
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index(self, connect, jac_table, get_jaccard_index_params):
'''
target: test create index interface
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_jaccard_index_params
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(jac_table, self.vectors)
status = connect.create_index(jac_table, index_params)
if index_params["index_type"] != IndexType.FLAT and index_params["index_type"] != IndexType.IVFLAT:
assert not status.OK()
else:
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_partition(self, connect, jac_table, get_jaccard_index_params):
'''
target: test create index interface
method: create table, create partition, and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_jaccard_index_params
logging.getLogger().info(index_params)
status = connect.create_partition(jac_table, tag)
status, ids = connect.add_vectors(jac_table, self.vectors, partition_tag=tag)
status = connect.create_index(jac_table, index_params)
assert status.OK()
@pytest.mark.level(2)
def test_create_index_without_connect(self, dis_connect, jac_table):
'''
target: test create index without connection
method: create table and add vectors in it, check if added successfully
expected: raise exception
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
with pytest.raises(Exception) as e:
status = dis_connect.create_index(jac_table, index_param)
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_search_with_query_vectors(self, connect, jac_table, get_jaccard_index_params):
'''
target: test create index interface, search with more query vectors
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_jaccard_index_params
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(jac_table, self.vectors)
status = connect.create_index(jac_table, index_params)
logging.getLogger().info(connect.describe_index(jac_table))
query_vecs = [self.vectors[0], self.vectors[1], self.vectors[2]]
top_k = 5
status, result = connect.search_vectors(jac_table, top_k, nprobe, query_vecs)
logging.getLogger().info(result)
assert status.OK()
assert len(result) == len(query_vecs)
"""
******************************************************************
The following cases are used to test `describe_index` function
******************************************************************
"""
def test_describe_index(self, connect, jac_table, get_jaccard_index_params):
'''
target: test describe index interface
method: create table and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_jaccard_index_params
logging.getLogger().info(index_params)
# status, ids = connect.add_vectors(jac_table, vectors[:5000])
status = connect.create_index(jac_table, index_params)
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
assert result._table_name == jac_table
assert result._index_type == index_params["index_type"]
assert result._nlist == index_params["nlist"]
def test_describe_index_partition(self, connect, jac_table, get_jaccard_index_params):
'''
target: test describe index interface
method: create table, create partition and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_jaccard_index_params
logging.getLogger().info(index_params)
status = connect.create_partition(jac_table, tag)
status, ids = connect.add_vectors(jac_table, vectors, partition_tag=tag)
status = connect.create_index(jac_table, index_params)
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == jac_table
assert result._index_type == index_params["index_type"]
"""
******************************************************************
The following cases are used to test `drop_index` function
******************************************************************
"""
def test_drop_index(self, connect, jac_table, get_jaccard_index_params):
'''
target: test drop index interface
method: create table and add vectors in it, create index, call drop index
expected: return code 0, and default index param
'''
index_params = get_jaccard_index_params
status, mode = connect._cmd("mode")
assert status.OK()
# status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(jac_table, index_params)
assert status.OK()
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
status = connect.drop_index(jac_table)
assert status.OK()
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == jac_table
assert result._index_type == IndexType.FLAT
def test_drop_index_partition(self, connect, jac_table, get_jaccard_index_params):
'''
target: test drop index interface
method: create table, create partition and add vectors in it, create index on table, call drop table index
expected: return code 0, and default index param
'''
index_params = get_jaccard_index_params
status = connect.create_partition(jac_table, tag)
status, ids = connect.add_vectors(jac_table, vectors, partition_tag=tag)
status = connect.create_index(jac_table, index_params)
assert status.OK()
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
status = connect.drop_index(jac_table)
assert status.OK()
status, result = connect.describe_index(jac_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == jac_table
assert result._index_type == IndexType.FLAT
class TestIndexHAM:
tmp, vectors = gen_binary_vectors(nb, dim)
@pytest.fixture(
scope="function",
params=gen_index_params()
)
def get_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ or request.param["index_type"] == IndexType.HNSW:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_simple_index_params(self, request, connect):
if str(connect._cmd("mode")[1]) == "CPU":
if request.param["index_type"] == IndexType.IVF_SQ8H:
pytest.skip("sq8h not support in CPU mode")
if request.param["index_type"] == IndexType.IVF_PQ or request.param["index_type"] == IndexType.HNSW:
pytest.skip("Skip PQ Temporary")
return request.param
@pytest.fixture(
scope="function",
params=gen_simple_index_params()
)
def get_hamming_index_params(self, request, connect):
logging.getLogger().info(request.param)
if request.param["index_type"] == IndexType.IVFLAT or request.param["index_type"] == IndexType.FLAT:
return request.param
else:
pytest.skip("Skip index Temporary")
"""
******************************************************************
The following cases are used to test `create_index` function
******************************************************************
"""
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index(self, connect, ham_table, get_hamming_index_params):
'''
target: test create index interface
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_hamming_index_params
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(ham_table, self.vectors)
status = connect.create_index(ham_table, index_params)
if index_params["index_type"] != IndexType.FLAT and index_params["index_type"] != IndexType.IVFLAT:
assert not status.OK()
else:
assert status.OK()
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_partition(self, connect, ham_table, get_hamming_index_params):
'''
target: test create index interface
method: create table, create partition, and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_hamming_index_params
logging.getLogger().info(index_params)
status = connect.create_partition(ham_table, tag)
status, ids = connect.add_vectors(ham_table, self.vectors, partition_tag=tag)
status = connect.create_index(ham_table, index_params)
assert status.OK()
status, res = connect.get_table_row_count(ham_table)
assert res == len(self.vectors)
@pytest.mark.level(2)
def test_create_index_without_connect(self, dis_connect, ham_table):
'''
target: test create index without connection
method: create table and add vectors in it, check if added successfully
expected: raise exception
'''
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
with pytest.raises(Exception) as e:
status = dis_connect.create_index(ham_table, index_param)
@pytest.mark.timeout(BUILD_TIMEOUT)
def test_create_index_search_with_query_vectors(self, connect, ham_table, get_hamming_index_params):
'''
target: test create index interface, search with more query vectors
method: create table and add vectors in it, create index
expected: return code equals to 0, and search success
'''
index_params = get_hamming_index_params
logging.getLogger().info(index_params)
status, ids = connect.add_vectors(ham_table, self.vectors)
status = connect.create_index(ham_table, index_params)
logging.getLogger().info(connect.describe_index(ham_table))
query_vecs = [self.vectors[0], self.vectors[1], self.vectors[2]]
top_k = 5
status, result = connect.search_vectors(ham_table, top_k, nprobe, query_vecs)
logging.getLogger().info(result)
assert status.OK()
assert len(result) == len(query_vecs)
"""
******************************************************************
The following cases are used to test `describe_index` function
******************************************************************
"""
def test_describe_index(self, connect, ham_table, get_hamming_index_params):
'''
target: test describe index interface
method: create table and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_hamming_index_params
logging.getLogger().info(index_params)
# status, ids = connect.add_vectors(jac_table, vectors[:5000])
status = connect.create_index(ham_table, index_params)
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
assert result._table_name == ham_table
assert result._index_type == index_params["index_type"]
assert result._nlist == index_params["nlist"]
def test_describe_index_partition(self, connect, ham_table, get_hamming_index_params):
'''
target: test describe index interface
method: create table, create partition and add vectors in it, create index, call describe index
expected: return code 0, and index instructure
'''
index_params = get_hamming_index_params
logging.getLogger().info(index_params)
status = connect.create_partition(ham_table, tag)
status, ids = connect.add_vectors(ham_table, vectors, partition_tag=tag)
status = connect.create_index(ham_table, index_params)
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
assert result._nlist == index_params["nlist"]
assert result._table_name == ham_table
assert result._index_type == index_params["index_type"]
"""
******************************************************************
The following cases are used to test `drop_index` function
******************************************************************
"""
def test_drop_index(self, connect, ham_table, get_hamming_index_params):
'''
target: test drop index interface
method: create table and add vectors in it, create index, call drop index
expected: return code 0, and default index param
'''
index_params = get_hamming_index_params
status, mode = connect._cmd("mode")
assert status.OK()
# status, ids = connect.add_vectors(ip_table, vectors)
status = connect.create_index(ham_table, index_params)
assert status.OK()
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
status = connect.drop_index(ham_table)
assert status.OK()
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ham_table
assert result._index_type == IndexType.FLAT
def test_drop_index_partition(self, connect, ham_table, get_hamming_index_params):
'''
target: test drop index interface
method: create table, create partition and add vectors in it, create index on table, call drop table index
expected: return code 0, and default index param
'''
index_params = get_hamming_index_params
status = connect.create_partition(ham_table, tag)
status, ids = connect.add_vectors(ham_table, vectors, partition_tag=tag)
status = connect.create_index(ham_table, index_params)
assert status.OK()
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
status = connect.drop_index(ham_table)
assert status.OK()
status, result = connect.describe_index(ham_table)
logging.getLogger().info(result)
assert result._nlist == NLIST
assert result._table_name == ham_table
assert result._index_type == IndexType.FLAT
class TestIndexTableInvalid(object):
"""
Test create / describe / drop index interfaces with invalid table names
"""
@pytest.fixture(
scope="function",
params=gen_invalid_table_names()
)
def get_table_name(self, request):
yield request.param
@pytest.mark.level(1)
def test_create_index_with_invalid_tablename(self, connect, get_table_name):
table_name = get_table_name
nlist = NLIST
index_param = {"index_type": IndexType.IVF_SQ8, "nlist": nlist}
status = connect.create_index(table_name, index_param)
assert not status.OK()
@pytest.mark.level(1)
def test_describe_index_with_invalid_tablename(self, connect, get_table_name):
table_name = get_table_name
status, result = connect.describe_index(table_name)
assert not status.OK()
@pytest.mark.level(1)
def test_drop_index_with_invalid_tablename(self, connect, get_table_name):
table_name = get_table_name
status = connect.drop_index(table_name)
assert not status.OK()
class TestCreateIndexParamsInvalid(object):
"""
Test Building index with invalid table names, table names not in db
"""
@pytest.fixture(
scope="function",
params=gen_invalid_index_params()
)
def get_index_params(self, request):
yield request.param
@pytest.mark.level(1)
def test_create_index_with_invalid_index_params(self, connect, table, get_index_params):
index_params = get_index_params
index_type = index_params["index_type"]
nlist = index_params["nlist"]
logging.getLogger().info(index_params)
# status, ids = connect.add_vectors(table, vectors)
if (not index_type) or (not nlist) or (not isinstance(index_type, IndexType)) or (not isinstance(nlist, int)):
with pytest.raises(Exception) as e:
status = connect.create_index(table, index_params)
else:
status = connect.create_index(table, index_params)
assert not status.OK()