Skip to content

Latest commit

 

History

History
217 lines (134 loc) · 7.99 KB

CHANGELOG.md

File metadata and controls

217 lines (134 loc) · 7.99 KB
Raw

2.0.0 (August 20, 2024)

Breaking

  • mssql dependency has been upgraded from 9.x to 11.x.
  • The minimum Node.js version is now 18.x.

1.0.0 (October 12, 2022)

Breaking

  • mssql dependency has been upgraded from 6.x to 9.x. See the changelog for that package for a list of changes that may be required to your configuration options.
  • Node >=14 is now required

Additionally, the package is now published as @godaddy/mssql-pool-party

0.5.2 (January 10, 2020)

Features

  • defaultConnectionPoolFactory is now exported at the root of the module, making it easier to create a custom connection pool factory that extends the default behavior.

Security Fixes

  • Cleaned up some npm audit errors (all on devDependencies).

0.5.1 (November 4, 2019)

Bug Fixes

  • Added missing @babel/runtime dependency. @babel/plugin-transform-runtime was added in 0.5.0, which expects @babel/runtime as a runtime dependency.

0.5.0 (November 4, 2019)

No known breaking changes, but we're being cautious with a major version bump since so many packages were updated.

Maintenance

  • Updated all packages to latest
  • Linting fixes

Bug Fixes

  • Fixed error reporting on streams (not all errors were being emitted)

0.4.0 (March 11, 2019)

Features

  • New healthy property in pool stats which indicates whether or not a pool was capable of serving requests since the last request was served. This means that the secondary pool could show as healthy but actually be unhealthy because it hasn't served any requests recently. If the primary pool shows as healthy, it's likely in a healthy state if it's receiving any semblance of traffic.
  • New acquireTimeoutMillis and createTimeoutMillis properties added to pool stats. These are the timeouts used by the tarn pool inside the mssql package.

Bug Fixes

  • No more stuck requests, resource churn, and broken failover when a database goes down. 0.3.* suffered from a bug in the mssql package due to some unexpected/broken behavior in generic-pool. mssql was bumped to 6.0.0-alpha.2, which replaced generic-pool with tarn.
  • The lastPromotionAt and lastHealAt stats should propagate to new pools after a heal.

Breaking Changes

  • Bumped mssql from ^4.3.0 to 6.0.0-alpha.2. We do not believe there are any breaking changes in mssql-pool-party for this upgrade, but the mssql bump brings with it bumps in the tedious driver and other internal changes which could be impacting. That, coupled with the alpha state of mssql means we're doing a major bump to be safe.

Other

  • Improved integration tests to handle failover, healing, and promotion scenarios by introducing a second mssql container.

0.3.2 (January 28, 2019)

Bug Fixes

  • Properly close unhealthy connections (PR #7, thanks @DullReferenceException!)

0.3.1 (January 8, 2019)

Features

  • Upgrade node-mssql to ^4.3.0.

0.3.0 (September 17, 2018)

Features

  • Upgrade node-mssql to ^4.2.1. This paves the way for supporting Availability Groups (more testing needed on this front).
  • Better DEBUG logging
  • Added the following properties to the stats output: config.appName, config.encrypt.

Bug Fixes

  • Added workaround for tediousjs/node-mssql#705 until it's fixed upstream. This isn't a bug in a previous version, but in node-mssql v4.

Breaking Changes

  • Since mssql-pool-party is a wrapper around node-mssql that aims to minimize interface changes, all breaking changes in node-mssql v4 are relevant to this new version, see here for those changes.
  • All responses from Request methods (query, execute, etc), for each style (promise, cb, stream), have been unified to match the changes made in v4. They all return a single object (and an error, if applicable). The result object will look something like the following (first example applies to callback/promise/async-await, second example applies to stream). See README.md for richer documentation.
{
  output: {},
  recordset: [{}],
  recordsets: [
    [{}],
  ],
  returnValue: 0,
  rowsAffected: [],
}

{
  columns: {
    ColumnName: {},
  },
  output: {},
  returnValue: 0,
  rows: [{}],
  rowsAffected: [],
}
  • Removed the following properties from the stats output: config.driver, config.tdsVersion, timeouts.cancel,
  • Enable the encrypt driver options by default. Disabled by default has already been deprecated in tedious and will be removed in a newer version, we're just jumping ahead a little to get rid of an annoying console message. If you don't want encryption, use this config:
connectionPoolConfig: {
  options: {
    encrypt: false
  }
}

Other

  • Upgrade jest and babel-jest from ^19.02 to ^23.6.0
  • Removed jest-environment-node-debug and jest-unit from devDependencies
  • Enabled test coverage, dropped requirement to 75% (goal of 100% in future)
  • sqlcmd is now on the PATH inside the integration test Docker container
  • Fixed some bugs involved with running tests in parallel
  • Added missing stream tests
  • Added package-lock.json

0.2.4 (November 13th, 2017)

Bug Fixes

  • Fixed a bug in the default raceWarmup strategy that was incorrectly resolving the warmup if a pool failed to connect properly. This should make initial connections a little more reliable.

0.2.3 (November 10th, 2017)

Features

  • Added priority to stats output.

0.2.2 (November 9th, 2017)

Bug Fixes

  • Fixed a bug in the reconnect process that was uncovered by the bug fix in 0.2.1. In 0.2.1, we started detecting pool healing failures correctly, which started triggering a condition in the reconnect process that prevented a reconnect if no pools were healed, which turned out not to be desirable. Now, if no pools were healed during a reconnect, it jumps right to another heal attempt.

0.2.1 (November 9th, 2017)

Features

  • Added optional pool prioritization. This feature is intended to provide a means to failback to a preferred DSN after previously promoting a lower priority DSN. See here for configuration options.

Bug Fixes

  • Fixed a bug in the heal process that was preventing heal failures from being detected properly in most cases. This will prevent unnecessary retries against unhealthy pools.

0.2.0 (August 14th, 2017)

Breaking Change

  • The pool property has been removed from the PoolError class due to concerns about exposing references to internal connections and DSNs. There is now a dsn property which contains information about the DSN owned by the pool involved in the error.

0.1.6 (August 3rd, 2017)

Other

  • Version bump with no change in functionality. This was done to avoid conflits with a previous version hosted on our internal registry.

0.1.5 (July 21th, 2017)

Other

  • Improved debug logs
  • All debug logs go to stdout
  • No changes in production functionality

0.1.4 (June 27th, 2017)

Bug Fixes

  • Fixed error when running connection.request().query('...') commands that did not return a recordset (e.g. TRUNCATE TABLE ...)

Other

  • Added a high-write integration test
  • Added unit test
  • Added a bit more debug logging

0.1.3 (April 5th, 2017)

Bug Fixes

  • Removed pool.dsn.password from PoolError to avoid passwords showing up in error logging

0.1.2 (April 3rd, 2017)

Bug Fixes

  • Fixed returnValue not being return when calling execute
  • Fixed broken execute callback tests

0.1.1 (April 3rd, 2017)

Features

  • Added connectionPoolConfig option to the ConnectionPoolParty constructor. This new option lets you set configuration options that will be passed to the mssql ConnectionPool constructor. This provides a way to set things like timeouts without having to create a custom connectionPoolFactory or dsnProvider. Note: any options set here will override options created by the dsnProvider

0.1.0 (April 3rd, 2017)

First Release!