Unique index validation stage #20133
Conversation
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ DetailsTest history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
🟢 |
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ |
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
UgnineSirdis
force-pushed
the
unique-index-validation-stage
branch
from
b2e1d86 to
ce1b363
Compare
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
snaury
left a comment
snaury
left a comment
There was a problem hiding this comment.
По даташардам от меня окей (изменение на вид минимально инвазивное). Однако учитывайте, что дизайн уникальных индексов, которые не уникальные на уровне реализации в даташарде / локальной базе может быть ошибкой с точки зрения эффективности. Дело в том, что тогда любой point read в таком индексе на самом деле выливается в range read, который дороже в реализации. Но кажется я это уже говорил на каком-то из прошлых обсуждений.
vitalif
left a comment
vitalif
left a comment
There was a problem hiding this comment.
Ну в общем я за то, чтобы ПР разделить, а статусы вторые убрать :-)
UgnineSirdis
force-pushed
the
unique-index-validation-stage
branch
from
ce1b363 to
362c4cc
Compare
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
| Y_ENSURE(found); | ||
| } | ||
|
|
||
| // Arrange shards in ascending order by shard index |
There was a problem hiding this comment.
а чего PerformCrossShardUniqIndexValidation сам внутри не посортирует как ему надо
There was a problem hiding this comment.
и это не by shard index, а по ключу видимо
There was a problem hiding this comment.
и что-то не пойму где оно сортируется вообще сейчас
There was a problem hiding this comment.
Я спрашивал у Ильназа и он сказал мне, что нужный мне порядок (по ключу) содержится в table->GetPartitions(). Сортировать отдельно смысла нет. Написал коммент, что "in ascending order by shard index"
|
⚪ Test history | Ya make output | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
UgnineSirdis
dismissed
vitalif’s stale review
I refactored according to the request
(cherry picked from commit 6771cfb)
(cherry picked from commit 6771cfb)
6 participants
Changelog entry
...
Changelog category
Description for reviewers
In this PR I add a special stage to index build process: validation of built index. This is used only as a unique index build stage.
Also here is the fix of the bug: in KQP index build case we don't fail query if index build process fails.