Until now, when building the list of SharedUniqueKeys, the code was selecting only the originalUniqueKeys that existed in ghostUniqueKeys list.

However, as part of the alter table, it is possible to add more unique keys or even composed primary keys with multiple columns.

The new logic keeps the old behaviour (it matches originalUniqueKey with exactly the same ghostUniqueKey) but also supports the cases where the originalUniqueKey is now a subset of one of the new ghostUniqueKeys.

If such a case happens, we can still use the originalUniqueKey as normal because a new ghostUniqueKey with the columns of originalUniqueKey plus some new columns is inherently unique just by the columns in the originalUniqueKey, no matter the value in the new columns of the new unique key.

* WIP: MySQL 8.4 support

* Add go-version for doing version comparisons

* docker-gh-ost-replica-tests: Support MySQL 8.4+

* localtests: Support MySQL 8.4

* Remove mysql-8.4.3 from replica-tests GHA since dbdeployer / ci env does not have 8.4.3

* MySQL 8.4: Actually use caching_sha2_password pw strategy

* Commit up  vendor/github.com/hashicorp/go-version

* Add GHA job for docker-gh-ost-replica-tests

* localtests/test.sh: Fix conditional bug and replica_terminology typo

When using `-cut-over-exponential-backoff` with `-default-retries`
greater than `64`, the sleep interval will overflow and immediately
begin retrying in rapid succession. This commit corrects the backoff
algorithm and adds test for both "retry operation" functions.

Co-authored-by: meiji163 <meiji163@github.com>

)

* Update connection.DuplicateCredentials function to set correct ServerName property

Ensure the ServerName TLS property matches the new connection
instance key hostname to avoid TLS verify errors like the following:

2025-01-02 02:07:26 FATAL tls: failed to verify certificate: x509: certificate is valid for [old host], not [new host]

This is only one part of the fix for this issue. The second part,
registering TLS Config with the mysql driver, will come in
subsequent commits.

* Use connection.DuplicateCredentials in cases where the connection key changes

* Extract TLS config key name generation to GetDBTLSConfigKey function

* Extract function to register a connection's TLS config with the mysql driver

This allows us to register TLS configuration is the various places
where connection configs are created and before they're used.

* Register TLS config when setting up master connection info

This ensures that the master's TLS config has been registered with
the mysql driver before any connections are attempted.

This is the second part of resolving the following TLS verify
error:

2025-01-02 02:07:26 FATAL tls: failed to verify certificate: x509: certificate is valid for [old host], not [new host]

* Register TLS config when setting up the throttler's connection info

This ensures that the throttler's TLS config has been registered with
the mysql driver before any connections are attempted.

This is the second part of resolving the following TLS verify
error:

2025-01-02 02:07:26 FATAL tls: failed to verify certificate: x509: certificate is valid for [old host], not [new host]

---------

Co-authored-by: meiji163 <meiji163@github.com>

* Fix clear-text logging of connection config

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>

* don't log connection configs

---------

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>

* Add trigger support to gh-ost based on openark#30

* Add comprehensive test cases for trigger support functionality

* Fix trigger-basic test by adding required --trigger-suffix parameter and remove fail-trigger-unsupported test

* Update trigger suffix in local test configurations

Modify extra_args files for trigger-complex, trigger-multiple, and trigger-self-reference tests to use different trigger suffixes

* Add --remove-trigger-suffix-if-exists to local test configurations

Update extra_args files for trigger tests to include the new --remove-trigger-suffix-if-exists option, ensuring consistent trigger handling across different test scenarios

* Fix trigger-long-name test by reducing trigger name length to be valid

* Standardize trigger drop statements in local test configurations

Update create.sql files across trigger test scenarios to:
- Consistently drop both original and ghost triggers
- Ensure clean slate before creating triggers
- Align with recent trigger suffix changes

This change improves test reliability and consistency by explicitly dropping all potential trigger variations before test setup.

* Consolidate and enhance trigger test configurations

Refactor local test scenarios for triggers by:
- Merging multiple trigger test configurations into comprehensive test cases
- Updating create.sql files with more complex and diverse trigger scenarios
- Standardizing trigger and table setup across different test configurations
- Removing redundant test directories while preserving test coverage

This change simplifies the trigger testing infrastructure and provides more robust test coverage for gh-ost's trigger handling capabilities.

* Add debug logging for ghost trigger validation process

Enhance ghost trigger existence check by adding detailed debug logging to:
- Log the ghost trigger name being searched
- Log the database schema and query details
- Log when an existing ghost trigger is found

This change improves visibility into the trigger validation process, making troubleshooting easier during migration scenarios.

* Improve ghost trigger validation with enhanced logging and verification

Modify validateGhostTriggersDontExist() to:
- Add a direct query to log all triggers in the database schema
- Refactor trigger existence check to use count-based query
- Improve debug logging for trigger validation process
- Provide more detailed error reporting for existing ghost triggers

This change enhances the robustness and observability of the trigger validation mechanism in gh-ost's migration process.

* Refactor ghost trigger validation to improve logging and error detection

Simplify and enhance the validateGhostTriggersDontExist() method by:
- Removing redundant direct query logging
- Streamlining trigger existence check
- Improving debug logging for trigger validation
- Consolidating trigger existence detection logic

The changes provide more concise and focused trigger validation with clearer error reporting.

* Simplify ghost trigger validation query and reduce logging verbosity

Refactor validateGhostTriggersDontExist() to:
- Streamline trigger existence check query
- Remove redundant debug logging statements
- Use a more concise approach to detecting existing triggers

The changes reduce code complexity while maintaining the core validation logic for ghost triggers.

* Enhance ghost trigger validation query to include table name filter

Modify validateGhostTriggersDontExist() to:
- Add table name filter to trigger existence check query
- Improve specificity of ghost trigger detection
- Prevent false positives from similarly named triggers in different tables

The change ensures more precise ghost trigger validation by incorporating the original table name into the query criteria.

* Enhance trigger test configuration with advanced features and consolidated test scenarios

Update trigger-advanced-features test configuration to:
- Add new column 'color' and 'modified_count' to test table
- Implement more complex trigger logic for color and count tracking
- Consolidate trigger test scenarios with richer data transformations
- Modify event to test both numeric and color-based updates
- Remove redundant trigger-basic-features directory

The changes provide a more comprehensive and nuanced test suite for gh-ost's trigger handling capabilities, demonstrating advanced trigger behaviors and self-referencing updates.

* Fix lint errors

* Update trigger test configuration with suffix change

Modify the extra_args file to use '_ght' trigger suffix instead of '_gho', maintaining consistency with recent trigger test configuration updates.

* Remove gh-ost-ci-env submodule

Clean up repository by removing the gh-ost-ci-env submodule, which appears to be no longer needed in the project structure.

* Update create.sql

---------

Co-authored-by: Yakir Gibraltar <yakir.g@taboola.com>

* Bump golang.org/x/net in the go_modules group across 1 directory

Bumps the go_modules group with 1 update in the / directory: [golang.org/x/net](https://github.com/golang/net).


Updates `golang.org/x/net` from 0.33.0 to 0.36.0
- [Commits](golang/net@v0.33.0...v0.36.0)

---
updated-dependencies:
- dependency-name: golang.org/x/net
  dependency-type: direct:production
  dependency-group: go_modules
...

Signed-off-by: dependabot[bot] <support@github.com>

* go mod tidy

* Remove toolchain directive from go.mod

* fix go version

* try update golangci

* remove deprecated linters

* remove unused nolint comment

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: meiji163 <meiji163@github.com>

…ss (#1500)

* First pass at raise_on_warnings

* Update dev.yml

* add CLI option, ignore expected warnings, check for row count discrepancy

* Count rows in each insert range prepared query

- log insert warnings always
- terminate if row count doesn't match and the PanicOnWarnings flag is set

* Lint comments - ensure errors are handled in show warnings

* update TestBuildUniqueKeyRangeEndPreparedQuery

* Localtests for PanicOnWarnings with data loss

* Unwrap CTE (mysql 5)

* Update localtests/panic-on-warnings-duplicate-unique-values-on-column-type-change/extra_args

Co-authored-by: Bastian Bartmann <accounts@bastianbartmann.de>

* limit BuildUniqueKeyRangeEndPreparedQueryViaOffset subquery properly

* Update Applier to support all unique indices with PanicOnWarnings. Add test coverage.

* Impl code review feedback for PanicOnWarnings
- documentation
- remove dev.yml
- remove unused variable

* bump golangci-lint for local dev

* Support altering index names with PanicOnWarnings

* Fix string matching for PanicOnWarnings to correctly suppress warnings when renaming unique keys

Error message formats are different across mysql distributions and versions

---------

Co-authored-by: Bastian Bartmann <accounts@bastianbartmann.de>

)

* Remove restriction on password length

* Updating go-mysql and dependencies

* Also remove fail-password-length localtest (no longer applies)

* fix: move successful cleanup into func

* fix: ignore linter error

* fix: inline onSuccessFunc function

…_modules group across 1 directory (#1513)

* Bump github.com/containerd/containerd

Bumps the go_modules group with 1 update in the / directory: [github.com/containerd/containerd](https://github.com/containerd/containerd).


Updates `github.com/containerd/containerd` from 1.7.18 to 1.7.27
- [Release notes](https://github.com/containerd/containerd/releases)
- [Changelog](https://github.com/containerd/containerd/blob/main/RELEASES.md)
- [Commits](containerd/containerd@v1.7.18...v1.7.27)

---
updated-dependencies:
- dependency-name: github.com/containerd/containerd
  dependency-type: indirect
  dependency-group: go_modules
...

Signed-off-by: dependabot[bot] <support@github.com>

* Remove toolchain directive from go.mod

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: meiji163 <meiji163@github.com>

* use docker for ci tests

* set charset server

* remove unusued cibuild-gh-ost-replica-tests

* add doc for how to run localtests

Bumps the go_modules group with 1 update in the / directory: [golang.org/x/net](https://github.com/golang/net).


Updates `golang.org/x/net` from 0.36.0 to 0.38.0
- [Commits](golang/net@v0.36.0...v0.38.0)

---
updated-dependencies:
- dependency-name: golang.org/x/net
  dependency-version: 0.38.0
  dependency-type: direct:production
  dependency-group: go_modules
...

Signed-off-by: dependabot[bot] <support@github.com>

…-bbb8b02913

Bump golang.org/x/net from 0.36.0 to 0.38.0 in the go_modules group across 1 directory

…roblems with --panic-on-warnings (#1557)

* Fix CalculateNextIterationRangeEndValues order by.

The split between Temptable and Offset query builders was originally introduced in #471. Then, both query builders have been changed to calculate actual chunk sizes in #1500.

The updated implementation of the Offset query introduced a bug, where a missing `order by` clause in `select_osc_chunk` can result in end values potentially surpassing the chunk_size. This wasn't detected during our initial testing where the query just returned rows in their original order, but may happen in real-world scenarios in case the db returns data in an undefined order.

An obvious fix would be to just add an `order by` to the Offset builder subquery, however since both builders use temptables now, it makes more sense to simplify and use only one of them.

Alternatively, the builder could only use the less performant count query variants when `--panic-on-warnings` is enabled and otherwise use the simpler ones from pull/471. We decided to not follow this path for now, hoping `--panic-on-warnings` becomes an updated and safer default in the future.

Co-authored-by: Bastian Bartmann <bastian.bartmann@shopify.com>

* Fix builder's chunk order and drop panic-on-warnings row count check

- Removed count subqueries from range builders to restore the more performant approach from PR #471
- Modified panic-on-warnings logic to trigger errors based solely on SQL warnings, not on row count mismatches
- This addresses potential race conditions where row count comparisons could produce false positives due to concurrent table modifications

---------

Co-authored-by: Bastian Bartmann <bastian.bartmann@shopify.com>

When migrating a table with a `binary` key in `-verbose` mode, different
sequences of bytes may prevent execution output from being logged, or
correctly logged, to the console.

This happens because the `MigrationRange*Values` are printed to the
screen without any type of encoding. One particularly problematic
sequence is `0x27`, `Escape`, which causes the terminal to stop logging
for the duration of the migration:

```
2025-05-23 11:53:27 INFO Listening on unix socket file: /tmp/gh-ost.test.binfoo.sock
2025-05-23 11:53:27 INFO Intercepted changelog state ReadMigrationRangeValues
2025-05-23 11:53:27 INFO Handled changelog state ReadMigrationRangeValues
2025-05-23 11:53:27 INFO Migration min values: [
```

This commit changes the to-string rendering for `binary` keys, rendering
the values as a hex string rather than an unescaped series of bytes.

Co-authored-by: meiji163 <meiji163@github.com>