tests: More shared code based on the patterns we have by def- · Pull Request #35021 · MaterializeInc/materialize

def- · 2026-02-15T22:35:45Z

No description provided.

github-actions · 2026-02-15T22:35:53Z

Pre-merge checklist

The PR title is descriptive and will make sense in the git log.
This PR has adequate test coverage / QA involvement has been duly considered. (trigger-ci for additional test/nightly runs)
If this PR includes major user-facing behavior changes, I have pinged the relevant PM to schedule a changelog post.
This PR has an associated up-to-date design doc, is a design doc (template), or is sufficiently small to not require a design.
If this PR evolves an existing $T ⇔ Proto$T mapping (possibly in a backwards-incompatible way), then it is tagged with a T-proto label.
If this PR will require changes to cloud orchestration or tests, there is a companion cloud PR to account for those changes that is tagged with the release-blocker label (example).

…l_workflows() 8 mzcompose.py files had an identical ~8-line workflow_default body that iterates all workflows as test parts. Extract this into a single method on Composition and replace the boilerplate with a one-liner call. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ws() Extend run_all_workflows() to support skipping additional workflows (beyond 'default') and optionally sharding via Buildkite parallelism. This deduplicates workflow_default boilerplate in 4 more test files. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ters Add internally_sharded, args, and between_workflows parameters to Composition.run_all_workflows() and use them to replace the repeated "internal sharding + buildkite sharding" boilerplate in 5 CDC test files. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…files() Add `shard_test_files()` and `glob_test_files()` methods to Composition that encapsulate the repeated pattern of parsing a filter argument, globbing for .td files in the test directory, sharding via Buildkite parallelism, and printing the result. Replace 8 occurrences of this ~10-15 line boilerplate across 6 CDC test files. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…run_scenarios() Deduplicate the repeated buildkite shard + scenario loop pattern across 5 CDC resumption test files (10 occurrences) into a single reusable method on Composition. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Move duplicated helper functions (create_mysql, create_mysql_replica, get_targeted_mysql_version, PostgresRecvlogical, create_postgres, get_targeted_pg_version) from test/mysql-cdc and test/pg-cdc pairs into shared utility modules, following the existing mysql_util.py pattern. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…orkflows() Replace the 29-line workflow_default boilerplate in both mysql-cdc-resumption and mysql-cdc-resumption-old-syntax with calls to c.run_all_workflows(), using the exclude, internally_sharded, and between_workflows parameters added in earlier sessions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…estdrive_size_args() Replace repeated two-line f-string pairs for default replica/storage size testdrive vars with a single static method call, improving readability across 6 CDC and kafka-matrix test files. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ows() Replace manual workflow_default boilerplate in 3 CDC resumption test files with c.run_all_workflows(between_workflows=...). Also extract a run_testdrive_files helper in sql-server-resumption-old-syntax to eliminate 13 repeated SQL Server credential var pairs. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…l_cdc_resumption.py The two mysql-cdc-resumption mzcompose.py files (675 lines each) differed by only 1 import line. Extract all 24 shared helper and scenario functions into a new misc/python/materialize/mysql_cdc_resumption.py module, reducing each file from 675 to 313 lines (-324 lines net). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…resumption.py Extract 14 shared functions (helpers + scenarios) from pg-cdc-resumption and pg-cdc-resumption-old-syntax into a new shared module, eliminating ~280 lines of duplicated code. The new-syntax file retains its 3 extra scenarios and pg8000 helpers; the old-syntax file is reduced from 281 to 95 lines. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

The SSH setup boilerplate (run setup.td, query public key, authorize on bastion host) was repeated in 10 workflows. Extract three helpers: _get_ssh_public_key, _authorize_ssh_key, _setup_and_authorize_ssh. Net -139 lines (42 added, 181 removed). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Add `MySql.default_testdrive_args()` and `SqlServer.default_testdrive_args()` static methods that return the standard testdrive --var args for MySQL root password and SQL Server credentials. Replace all 24 inline occurrences across 19 files with calls to these methods, consistent with the existing `Materialized.default_testdrive_size_args()` pattern. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…s_util.py Move 5 identical replication slot helper functions from both pg-cdc/mzcompose.py and pg-cdc-old-syntax/mzcompose.py into the shared postgres_util.py module. The helpers use the common cursor/execute/fetchall DB interface and work with both psycopg and pg8000 connection objects. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Move 4 identical functions (workflow_cdc, workflow_replica_connection, workflow_schema_change_restart, _make_inserts) from both mysql-cdc and mysql-cdc-old-syntax mzcompose.py files into a shared mysql_cdc.py module. The mzcompose framework discovers imported workflow_* functions via getmembers(module, isfunction), so no wrapper functions are needed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…rive_ssl_args to postgres_util.py Move workflow_wal_level, workflow_replication_disabled, and kill_pg_and_mz into a new shared pg_cdc.py module. Move get_testdrive_ssl_args() from pg-cdc-old-syntax's local definition to postgres_util.py and use it in both pg-cdc files, replacing inline SSL cert extraction in pg-cdc. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

… in checks Add a pre-dedented `KAFKA_SCHEMA` constant and a shared `remove_target_cluster_from_explain()` function to `checks/common.py`, replacing 12 identical `schemas()`/`schema()` wrapper functions and 5 identical `remove_target_cluster_from_explain()` definitions across 20 check files. Also replaces 3 inline `dedent(KAFKA_SCHEMA_WITH_SINGLE_STRING_FIELD)` usages with the constant. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…nt_connection_drop into pg_cdc.py Move 3 shared workflow functions from pg-cdc and pg-cdc-old-syntax mzcompose.py files into the shared pg_cdc.py module. Each function is parameterized to handle the minor differences between the two test suites (exclude list, max_replication_slots, DB connection library). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Replace 27 occurrences of the 4-line ALTER SYSTEM SET boilerplate across 9 test files with a single method call on Composition. This makes each call site clearly express intent rather than exposing port/user details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…boilerplate Replace 48 occurrences of the 4-line `c.sql("ALTER SYSTEM SET ...", port=6877, user="mz_system")` pattern across 16 files with a 1-line `c.alter_system_set(name, value)` call. The method handles the internal port and system user automatically and forwards extra kwargs like `print_statement` and `service`. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ver schema() Extract the 33-line quickstart cluster recreation boilerplate into a reusable Composition method. This replaces 3 identical occurrences in testdrive and testdrive-old-kafka-src-syntax with 4-line calls. Also replace the leftover schema() wrapper in source-sink-errors with the KAFKA_SCHEMA constant from common.py (missed in Session 17). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Replace 49 occurrences of `c.kill("materialized"); c.up("materialized")` across 15 files with `c.restart_mz()`. Remove 4 standalone `restart_mz()` helper functions from shared modules and test files. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…_system boilerplate Replace 60+ occurrences of `c.sql(..., port=6877, user="mz_system")` and `c.sql_query(..., port=6877, user="mz_system")` with dedicated `sql_as_mz_system()` and `sql_query_as_mz_system()` methods across 15 files. Also converts 3 leftover raw ALTER SYSTEM SET calls to `alter_system_set()`. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Extract two helpers to deduplicate repeated patterns: - `_run_kafka_ssh_test()`: encapsulates the full kafka SSH test setup (down, override bastion, start deps, authorize, failure/recovery cycle), replacing 3 near-identical 25-line workflows with 6-line calls each. - `_test_ssh_bastion_failure_recovery()`: encapsulates the kill/up/verify bastion failure sequence, replacing inline code in pg, mysql, and sql_server workflows. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…erplate 50 occurrences of the 4-5 line STORAGECTL/STORAGE/COMPUTECTL/COMPUTE address block across 8 test files are replaced with a single call to Clusterd.replica_addresses(). Net -164 lines. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Move the 93-line workflow_many_inserts function (testing concurrent inserts during source creation) from both mysql-cdc and mysql-cdc-old-syntax mzcompose.py files into the shared mysql_cdc.py module. The only difference was the CREATE SOURCE SQL (new vs old syntax), now passed as a parameter. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

The 0dt test file had 6 identical 20-line Materialized+Testdrive override blocks, 4 identical 10-line Testdrive-only override blocks, and 5 identical 12-line rehydrate+promote timing blocks. Extract these into _mz_new_services(), _testdrive_mz_new(), and _promote_mz_new() helpers to reduce duplication. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Both backup-restore and backup-restore-postgres test files had nearly identical workflow_default functions (~60 lines each) with an explicit TODO comment asking for extraction. The only differences were the metadata store backend (CockroachDB vs Postgres). Extract the shared logic into misc/python/materialize/backup_restore.py, using Composition.backup()/restore() (which auto-detect the store) and a new Composition.exec_metadata_store_sql() method that dispatches to the correct CLI. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

def- force-pushed the pr-test-unification branch from 25b7c96 to 0704e0f Compare February 18, 2026 08:07

def- and others added 28 commits February 18, 2026 08:09

def- force-pushed the pr-test-unification branch from 0704e0f to 9698b6f Compare February 18, 2026 08:11

def- changed the title ~~Pr test unification~~ tests: More shared code based on the patterns we have Feb 18, 2026

def- closed this Feb 18, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

tests: More shared code based on the patterns we have#35021

tests: More shared code based on the patterns we have#35021
def- wants to merge 28 commits intoMaterializeInc:mainfrom
def-:pr-test-unification

def- commented Feb 15, 2026

Uh oh!

github-actions Bot commented Feb 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

def- commented Feb 15, 2026

Uh oh!

github-actions Bot commented Feb 15, 2026

Pre-merge checklist

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant