-
Notifications
You must be signed in to change notification settings - Fork 114
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix creation of replicated tables when using legacy strategy and unique_key #208
Fix creation of replicated tables when using legacy strategy and unique_key #208
Conversation
Thanks for the PR. I have mixed feelings about this because |
Ah seems like you're right, it's just the However, moving the
The engine for this model is set to |
What you're seeing might be a result of the behavior described here: ClickHouse/ClickHouse#12135. If you are running the model repeatedly within 8 minutes the old table path might still be in zookeeper. I'm fairly certain the UUID for "CREATE TABLE AS" would be based on either a timestamp, the "different" table name, or both, so I don't think it's reusing the source table one. It's likely we should be applying the Unfortunately this sort of behavior is one of the reasons that dbt is problematic on ClickHouse clusters. |
c91f961
to
a75eeb9
Compare
You're absolutely right, not sure why I didn't catch that before. The UUID does change when doing a CTAS, but somehow the UUID was hardcoded for some of my tables. I set up a fresh instance and using CTAS with |
* fix: SYSTEM SYNC REPLICA for on_cluster_clause (ClickHouse#156) * fix SYSTEM SYNC REPLICA * add schema * Update version and pypi job * Fix incompatible return type (ClickHouse#162) * Distributed table materialization (ClickHouse#163) * distributed table materialization * fix rebase * PR fixes * Bump version * Tweak PyPI build Python release * Add space to exchange_tables_atomic macro (ClickHouse#168) * Add space to exchange_tables_atomic macro This changes the SYSTEM SYNC REPLICA query to have a space between the ON CLUSTER clause and the table name. * Move whitespace to on_cluster_clause * Fix bad logging/error handling (ClickHouse#170) * Distributed incremental materialization (ClickHouse#172) * distributed table materialization * fix rebase * PR fixes * distributed incremental materialization * fix * fix * add insert_distributed_sync to README.md * add checks on insert_distributed_sync * add checks on insert_distributed_sync * review fixes * Update version and tweak docs * Lw delete set fix (ClickHouse#174) * Move lightweight delete settings to per query for HTTP stickiness fix * Minor cleanup and doc updates * Fix legacy incremental materialization (ClickHouse#178) * fix: distributed_table materialization issue (ClickHouse#184) * Bump version and changelog (ClickHouse#185) * cluster names containing dash characters (ClickHouse#198) (ClickHouse#200) Co-authored-by: the4thamigo-uk <the4thamigo-uk> * Add basic error test, fix minor merge conflict (ClickHouse#202) * Cluster setting and Distributed Table tests (ClickHouse#186) * added can_on_cluster var in ClickhouseRelation * add tests for cluster * fix lint issue * debug set cluster env variable * debug test * debug and add tests * skip distributed table grant test * debug workflow * debug workflow * debug test * add tests fro distributed_incremental * fix zk path error * fix wrong alias for distributed materializations update aliase test * update base on review * Update version and CHANGELOG, incorporate cluster name fix (ClickHouse#203) * Release 1 5 0 (ClickHouse#210) * Initial 1.5.0 commit * Reorganize basic tests * Fix lint * Add case sensitive cache * Fix s3 bucket bug * Checkpoint for constraints/contracts * Fix native column query * Loosen replication test * Checkpoint for constraints tests * Checkpoint for constraints tests * Add rendering of model level CHECK constraints * Fix lint * Reorganize test files * Add one hooks test * Fix lint * Update test and dependency versions. (ClickHouse#211) * Adjust the wrapper parenthesis around the table materialization sql code (ClickHouse#212) * Update for 1.5.1 bug fix * Fix creation of replicated tables when using legacy materialization (ClickHouse#208) * On cluster sync cleanup * Bug fixes related to model settings. (ClickHouse#214) * Add materialization macro for materialized view (ClickHouse#207) * Add materialization macro for materialized view * fix isort issues in materialized view test * Release 1 6 0 (ClickHouse#215) * Initial dbt 1.6 update * Add skipped clone test * Clean up MV PR * Release 1 6 1 (ClickHouse#217) * Identifier quoting checkpoint * Identifier quoting checkpoint * Fix distributed table local quoting * Fix issues with deduplication settings * Release 1 6 2 (ClickHouse#219) * Limited fix to completely broken `on_schema_change` * Tweak changelog * Release 1 7 0 (ClickHouse#220) * Initial dependency updates for 1.7.x * Initial dependency updates for 1.7.x * Correctly warn or error if light weight deletes not available * Wrap columns_in_query query in select statement (ClickHouse#222) * Wrap columns_in_query query in select statement * formatting * Update changelog * allows to add a comment in table's or view's metadata * add settings_section flag as comment for code using settings * override test sql macro and add limit-placer macro * update CHANGELOG.md * fix: use correct schema for MV target tables (ClickHouse#244) * fix: use correct schema when updating MVs The existing implementation passes just the name for `target_table`, which ultimately means that the target schema is not included when the final SQL is generated. By passing the entire relation object, the correct target schema will be present in the final SQL. * update MV tests Provide a custom schema to make sure that the full target table name (schema + relation name) is included in the CREATE MATERIALIZED VIEW statement * Update changelog * rename end of query flag * Bug/223 relationship test with limit (ClickHouse#245) * add settings_section flag as comment for code using settings * override test sql macro and add limit-placer macro * update CHANGELOG.md * rename end of query flag * Revert "Bug/223 relationship test with limit (ClickHouse#245)" (ClickHouse#247) This reverts commit d8afb93. * always return --end_of_sql when asking for settings * Add model settings based on materialization type * support setting clause on view creation * edit CHANGELOG.md * Bump version and tweak changelog * change list syntax to satisfy lint test * change list syntax to satisfy lint test * change imports order to satisfy lint test * Add typing to satisfy lint * Add snapshot materialization to default settings * Fix tests - add distributed_table and distributed_incremental materializations * Fix tests - make sure to call the get_model_settings only when materialization is view * clean up recent changelog * Add materialization macro for dictionaries * address lint issue in dictionary test * address lint issue with enum * Fix model settings with custom materialization * Release 1.7.4 housekeeping (ClickHouse#261) * Bump black from 23.11.0 to 24.3.0 (ClickHouse#259) Bumps [black](https://github.com/psf/black) from 23.11.0 to 24.3.0. - [Release notes](https://github.com/psf/black/releases) - [Changelog](https://github.com/psf/black/blob/main/CHANGES.md) - [Commits](psf/black@23.11.0...24.3.0) --- updated-dependencies: - dependency-name: black dependency-type: direct:development ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Release 1 7 5 (ClickHouse#265) * Release 1.7.5 housekeeping * Upgrade setuptools requirement for clickhouse_driver install * Remove flake8 checks for the moment * Update workflow actions * Fix black comma * fix(clients): add newlines around subquery when retrieving columns to avoid a syntax error (ClickHouse#262) * Fix lint * lazy load agate (ClickHouse#263) * feat: add TTL support (ClickHouse#254) * Fix lint * Update table relation after exchange command (ClickHouse#230) Related to ClickHouse#226 * feat: allow to add connection overrides for dictionaries (ClickHouse#267) * Housekeeping for 1.7.6 release (ClickHouse#268) * Revert "allows to add a comment in table's or view's metadata" * Fix bool_or behavior (ClickHouse#270) * feat: support column codecs * Use Column.data_type in ClickHouseAdapter.format_columns * Always apply query_settings in clickhouse__insert_into macro * Add ClickHouseColumn.is_low_cardinality * Update column type test cases for LowCardinality * Omit empty dictionary connection_overrides from materialization DDL --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Schum <68906108+Schum-io@users.noreply.github.com> Co-authored-by: Geoff Genz <geoff@clickhouse.com> Co-authored-by: Sergey Reshetnikov <resh.sersh@gmail.com> Co-authored-by: gladkikhtutu <88535677+gladkikhtutu@users.noreply.github.com> Co-authored-by: Damir Basic Knezevic <damirbasicknezevic@gmail.com> Co-authored-by: Zhenbang <122523068+zli06160@users.noreply.github.com> Co-authored-by: Andy <email@elevatesystems.co.uk> Co-authored-by: gfunc <fcjchaojian@gmail.com> Co-authored-by: Kristof Szaloki <szalokikristof@gmail.com> Co-authored-by: Steven Reitsma <4895139+StevenReitsma@users.noreply.github.com> Co-authored-by: Rory Sawyer <rory@sawyer.dev> Co-authored-by: ptemarvelde <45282601+ptemarvelde@users.noreply.github.com> Co-authored-by: Dmitrii Tcimokha <dstsimokha@gmail.com> Co-authored-by: bentsileviav <bentsi.leviav@clickhouse.com> Co-authored-by: Dmitriy Sokolov <silentsokolov@gmail.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: triou <teva.riou@gmail.com> Co-authored-by: Daniel Reeves <31971762+dwreeves@users.noreply.github.com> Co-authored-by: Cristhian Garcia <crisgarta8@gmail.com> Co-authored-by: Thomas Schmidt <somtom91@gmail.com> Co-authored-by: scrawfor <scrawfor@users.noreply.github.com> Co-authored-by: Robin Norgren <68205730+rjoelnorgren@users.noreply.github.com>
Problem
Table creation fails in the situation where:
ReplicatedMergeTree
table enginescluster_mode: true
unique_key
for a modelCause
The cause is twofold:
create table [table1] as [table2] on cluster [cluster]
is invalid SQL for ClickHouse (at least for OSS). Theon cluster [cluster]
clause is not supported forcreate table .. as
statements.on_cluster_clause
, table creation fails onCode: 36. DB::Exception: Received from localhost:9000. DB::Exception: Macro 'uuid' and empty arguments of ReplicatedMergeTree are supported only for ON CLUSTER queries with Atomic database engine. (BAD_ARGUMENTS)
.So in this case,
on cluster
is both required and not supported at the same time.Solution
This PR changes
create table [table1] as [table2]
tocreate table [table1] empty as ([sql for table1])
. This lets dbt supply ClickHouse with the SQL that was used to create[table1]
for[table2]
, instead of letting ClickHouse look at[table2]
itself to copy the schema.Tests
All tests pass locally.