v0.15.0

Added

• The River CLI will now respect the standard set of PG* environment variables like PGHOST, PGPORT, PGDATABASE, PGUSER, PGPASSWORD, and PGSSLMODE to configure a target database when the --database-url parameter is omitted. PR #702.
• Add missing doc for JobRow.UniqueStates + reveal rivertype.UniqueOptsByStateDefault() to provide access to the default set of unique job states. PR #707.

Changed

• Sleep durations are now logged as Go-like duration strings (e.g. "10s") in either text or JSON instead of duration strings in text and nanoseconds in JSON. PR #699.
• Altered the migration comments from river migrate-get to include the "line" of the migration being run (main, or for River Pro workflow and sequence) to make them more distinguishable. PR #703.
• Fewer slice allocations during unique insertions. PR #705.

Fixed

• Exponential backoffs at degenerately high job attempts (>= 310) no longer risk overflowing time.Duration. PR #698.

v0.14.3

Changed

• Dropped internal random generators in favor of math/rand/v2, which will have the effect of making code fully incompatible with Go 1.21 (go.mod has specified a minimum of 1.22 for some time already though). PR #691.

Fixed

• 006 migration now tolerates previous existence of a unique_states column in case it was added separately so that the new index could be raised with CONCURRENTLY. PR #690.

v0.14.2

Fixed

• Cancellation of running jobs relied on a channel that was only being received when in the job fetch routine, meaning that jobs which were cancelled would not be cancelled until the next scheduled fetch. This was fixed by also receiving from the job cancellation channel when in the main producer loop, even if no fetches are happening. PR #678.
• Job insert middleware were not being utilized for periodic jobs. This insertion path has been refactored to rely on the unified insertion path from the client. Fixes #675. PR #679.

v0.14.1

Fixed

• In PR #663 the client was changed to be more aggressive about re-fetching when it had previously fetched a full batch. Unfortunately a clause was missed, which resulted in the client being more aggressive any time even a single job was fetched on the previous attempt. This was corrected with a conditional to ensure it only happens when the last fetch was full. PR #668.

v0.14.0

Added

• Expose JobCancelError and JobSnoozeError types to more easily facilitate testing. PR #665.

Changed

• Tune the client to be more aggressive about fetching when it just fetched a full batch of jobs, or when it skipped its previous triggered fetch because it was already full. This should bring more consistent throughput to poll-only mode and in cases where there is a backlog of existing jobs but new ones aren't being actively inserted. This will result in increased fetch load on many installations, with the benefit of increased throughput. As before, FetchCooldown still limits how frequently these fetches can occur on each client and can be increased to reduce the amount of fetch querying. Thanks Chris Gaffney (@gaffneyc) for the idea, initial implementation, and benchmarks. PR #663.

Fixed

• riverpgxv5 driver: Hijack() the underlying listener connection as soon as it is acquired from the pgxpool.Pool in order to prevent the pool from automatically closing it after it reaches its max age. A max lifetime makes sense in the context of a pool with many conns, but a long-lived listener does not need a max lifetime as long as it can ensure the conn remains healthy. PR #661.

v0.13.0

⚠️ Version 0.13.0 removes the original advisory lock based unique jobs implementation that was deprecated in v0.12.0. See details in the note below or the v0.12.0 release notes.

Added

• A middleware system was added for job insertion and execution, providing the ability to extract shared functionality across workers. Both JobInsertMiddleware and WorkerMiddleware can be configured globally on the Client, and WorkerMiddleware can also be added on a per-worker basis using the new Middleware method on Worker[T]. Middleware can be useful for logging, telemetry, or for building higher level abstractions on top of base River functionality.

  Despite the interface expansion, users should not encounter any breakage if they're embedding the WorkerDefaults type in their workers as recommended. PR #632.

Changed

• Breaking change: The advisory lock unique jobs implementation which was deprecated in v0.12.0 has been removed. Users of that feature should first upgrade to v0.12.1 to ensure they don't see any warning logs about using the deprecated advisory lock uniqueness. The new, faster unique implementation will be used automatically as long as the UniqueOpts.ByState list hasn't been customized to remove required states (pending, scheduled, available, and running). As of this release, customizing ByState without these required states returns an error. PR #614.
• Single job inserts are now unified under the hood to use the InsertMany bulk insert query. This should not be noticeable to users, and the unified code path will make it easier to build new features going forward. PR #614.

Fixed

• Allow river.JobCancel to accept a nil error as input without panicking. PR #634.

v0.13.0-rc.1

⚠️ Version 0.13.0 removes the original advisory lock based unique jobs implementation that was deprecated in v0.12.0. See details in the note below or the v0.12.0 release notes.

Added

• A middleware system was added for job insertion and execution, providing the ability to extract shared functionality across workers. Both JobInsertMiddleware and WorkerMiddleware can be configured globally on the Client, and WorkerMiddleware can also be added on a per-worker basis using the new Middleware method on Worker[T]. Middleware can be useful for logging, telemetry, or for building higher level abstractions on top of base River functionality.

  Despite the interface expansion, users should not encounter any breakage if they're embedding the WorkerDefaults type in their workers as recommended. PR #632.

Changed

• Breaking change: The advisory lock unique jobs implementation which was deprecated in v0.12.0 has been removed. Users of that feature should first upgrade to v0.12.1 to ensure they don't see any warning logs about using the deprecated advisory lock uniqueness. The new, faster unique implementation will be used automatically as long as the UniqueOpts.ByState list hasn't been customized to remove required states (pending, scheduled, available, and running). As of this release, customizing ByState without these required states returns an error. PR #614.
• Single job inserts are now unified under the hood to use the InsertMany bulk insert query. This should not be noticeable to users, and the unified code path will make it easier to build new features going forward. PR #614.

Fixed

• Allow river.JobCancel to accept a nil error as input without panicking. PR #634.

v0.12.1

Changed

• The BatchCompleter that marks jobs as completed can now batch database updates for all states of jobs that have finished execution. Prior to this change, only completed jobs were batched into a single UPDATE call, while jobs moving to any other state used a single UPDATE per job. This change should significantly reduce database and pool contention on high volume system when jobs get retried, snoozed, cancelled, or discarded following execution. PR #617.

Fixed

• Unique job changes from v0.12.0 / PR #590 introduced a bug with scheduled or retryable unique jobs where they could be considered in conflict with themselves and moved to discarded by mistake. There was also a possibility of a broken job scheduler if duplicate retryable unique jobs were attempted to be scheduled at the same time. The job scheduling query was corrected to address these issues along with missing test coverage. PR #619.

v0.12.0

⚠️ Version 0.12.0 contains a new database migration, version 6. See documentation on running River migrations. If migrating with the CLI, make sure to update it to its latest version:

go install github.com/riverqueue/river/cmd/river@latest
river migrate-up --database-url "$DATABASE_URL"

If not using River's internal migration system, the raw SQL can alternatively be dumped with:

go install github.com/riverqueue/river/cmd/river@latest
river migrate-get --version 6 --up > river6.up.sql
river migrate-get --version 6 --down > river6.down.sql

The migration includes a new index. Users with a very large job table may want to consider raising the index separately using CONCURRENTLY (which must be run outside of a transaction), then run river migrate-up to finalize the process (it will tolerate an index that already exists):

ALTER TABLE river_job ADD COLUMN unique_states BIT(8);

CREATE UNIQUE INDEX CONCURRENTLY river_job_unique_idx ON river_job (unique_key)
    WHERE unique_key IS NOT NULL
      AND unique_states IS NOT NULL
      AND river_job_state_in_bitmask(unique_states, state);

go install github.com/riverqueue/river/cmd/river@latest
river migrate-up --database-url "$DATABASE_URL"

Added

• rivertest.WorkContext, a test function that can be used to initialize a context to test a JobArgs.Work implementation that will have a client set to context for use with river.ClientFromContext. PR #526.
• A new river migrate-list command is available which lists available migrations and which version a target database is migrated to. PR #534.
• river version or river --version now prints River version information. PR #537.
• Config.JobCleanerTimeout was added to allow configuration of the job cleaner query timeout. In some deployments with millions of stale jobs, the cleaner may not be able to complete its query within the default 30 seconds.

Changed

⚠️ Version 0.12.0 has two small breaking changes, one for InsertMany and one in rivermigrate. As before, we try never to make breaking changes, but these ones were deemed worth it because of minimal impact and to help avoid panics.

• Breaking change: Client.InsertMany / InsertManyTx now return the inserted rows rather than merely returning a count of the inserted rows. The new implementations no longer use Postgres' COPY FROM protocol in order to facilitate return values.

  Users who relied on the return count can merely wrap the returned rows in a len() to return to that behavior, or you can continue using the old APIs using their new names InsertManyFast and InsertManyFastTx. PR #589.

• Breaking change: rivermigrate.New now returns a possible error along with a migrator. An error may be returned, for example, when a migration line is configured that doesn't exist. PR #558.

  # before
  migrator := rivermigrate.New(riverpgxv5.New(dbPool), nil)
  
  # after
  migrator, err := rivermigrate.New(riverpgxv5.New(dbPool), nil)
  if err != nil {
      // handle error
  }

• Unique jobs have been improved to allow bulk insertion of unique jobs via InsertMany / InsertManyTx, and to allow customizing the ByState list to add or remove certain states. This enables users to expand the set of unique states to also include cancelled and discarded jobs, or to remove retryable from uniqueness consideration. This updated implementation maintains the speed advantage of the newer index-backed uniqueness system, while allowing some flexibility in which job states.

  Unique jobs utilizing ByArgs can now also opt to have a subset of the job's arguments considered for uniqueness. For example, you could choose to consider only the customer_id field while ignoring the trace_id field:

  type MyJobArgs {
    CustomerID string `json:"customer_id" river:"unique`
    TraceID string `json:"trace_id"`
  }

  Any fields considered in uniqueness are also sorted alphabetically in order to guarantee a consistent result, even if the encoded JSON isn't sorted consistently. For example encoding/json encodes struct fields in their defined order, so merely reordering struct fields would previously have been enough to cause a new job to not be considered identical to a pre-existing one with different JSON order.

  The UniqueOpts type also gains an ExcludeKind option for cases where uniqueness needs to be guaranteed across multiple job types.

  In-flight unique jobs using the previous designs will continue to be executed successfully with these changes, so there should be no need for downtime as part of the migration. However the v6 migration adds a new unique job index while also removing the old one, so users with in-flight unique jobs may also wish to avoid removing the old index until the new River release has been deployed in order to guarantee that jobs aren't duplicated by old River code once that index is removed.

  Deprecated: The original unique jobs implementation which relied on advisory locks has been deprecated, but not yet removed. The only way to trigger this old code path is with a single insert (Insert/InsertTx) and using UniqueOpts.ByState with a custom list of states that omits some of the now-required states for unique jobs. Specifically, pending, scheduled, available, and running can not be removed from the ByState list with the new implementation. These are included in the default list so only the places which customize this attribute need to be updated to opt into the new (much faster) unique jobs. The advisory lock unique implementation will be removed in an upcoming release, and until then emits warning level logs when it's used.

  PR #590.

• Deprecated: The MigrateTx method of rivermigrate has been deprecated. It turns out there are certain combinations of schema changes which cannot be run within a single transaction, and the migrator now prefers to run each migration in its own transaction, one-at-a-time. MigrateTx will be removed in future version.

• The migrator now produces a better error in case of a non-existent migration line including suggestions for known migration lines that are similar in name to the invalid one. PR #558.

Fixed

• Fixed a panic that'd occur if StopAndCancel was invoked before a client was started. PR #557.
• A PeriodicJobConstructor should be able to return nil JobArgs if it wishes to not have any job inserted. However, this was either never working or was broken at some point. It's now fixed. Thanks @semanser! PR #572.
• Fixed a nil pointer exception if Client.Subscribe was called when the client had no configured workers (it still, panics with a more instructive error message now). PR #599.

v0.12.0-rc.1

⚠️ Version 0.12.0 contains a new database migration, version 6. See documentation on running River migrations. If migrating with the CLI, make sure to update it to its latest version:

go install github.com/riverqueue/river/cmd/river@latest
river migrate-up --database-url "$DATABASE_URL"

If not using River's internal migration system, the raw SQL can alternatively be dumped with:

go install github.com/riverqueue/river/cmd/river@latest
river migrate-get --version 6 --up > river6.up.sql
river migrate-get --version 6 --down > river6.down.sql

The migration includes a new index. Users with a very large job table may want to consider raising the index separately using CONCURRENTLY (which must be run outside of a transaction), then run river migrate-up to finalize the process (it will tolerate an index that already exists):

ALTER TABLE river_job ADD COLUMN unique_states BIT(8);

CREATE UNIQUE INDEX CONCURRENTLY river_job_unique_idx ON river_job (unique_key)
    WHERE unique_key IS NOT NULL
      AND unique_states IS NOT NULL
      AND river_job_state_in_bitmask(unique_states, state);

go install github.com/riverqueue/river/cmd/river@latest
river migrate-up --database-url "$DATABASE_URL"

Added

• rivertest.WorkContext, a test function that can be used to initialize a context to test a JobArgs.Work implementation that will have a client set to context for use with river.ClientFromContext. PR #526.
• A new river migrate-list command is available which lists available migrations and which version a target database is migrated to. PR #534.
• river version or river --version now prints River version information. PR #537.
• Config.JobCleanerTimeout was added to allow configuration of the job cleaner query timeout. In some deployments with millions of stale jobs, the cleaner may not be able to complete its query within the default 30 seconds.

Changed

⚠️ Version 0.12.0 has two small breaking changes, one for InsertMany and one in rivermigrate. As before, we try never to make breaking changes, but these ones were deemed worth it because of minimal impact and to help avoid panics.

• Breaking change: Client.InsertMany / InsertManyTx now return the inserted rows rather than merely returning a count of the inserted rows. The new implementations no longer use Postgres' COPY FROM protocol in order to facilitate return values.

  Users who relied on the return count can merely wrap the returned rows in a len() to return to that behavior, or you can continue using the old APIs using their new names InsertManyFast and InsertManyFastTx. PR #589.

• Breaking change: rivermigrate.New now returns a possible error along with a migrator. An error may be returned, for example, when a migration line is configured that doesn't exist. PR #558.

  # before
  migrator := rivermigrate.New(riverpgxv5.New(dbPool), nil)
  
  # after
  migrator, err := rivermigrate.New(riverpgxv5.New(dbPool), nil)
  if err != nil {
      // handle error
  }

• Unique jobs have been improved to allow bulk insertion of unique jobs via InsertMany / InsertManyTx, and to allow customizing the ByState list to add or remove certain states. This enables users to expand the set of unique states to also include cancelled and discarded jobs, or to remove retryable from uniqueness consideration. This updated implementation maintains the speed advantage of the newer index-backed uniqueness system, while allowing some flexibility in which job states.

  Unique jobs utilizing ByArgs can now also opt to have a subset of the job's arguments considered for uniqueness. For example, you could choose to consider only the customer_id field while ignoring the trace_id field:

  type MyJobArgs {
    CustomerID string `json:"customer_id" river:"unique`
    TraceID string `json:"trace_id"`
  }

  Any fields considered in uniqueness are also sorted alphabetically in order to guarantee a consistent result, even if the encoded JSON isn't sorted consistently. For example encoding/json encodes struct fields in their defined order, so merely reordering struct fields would previously have been enough to cause a new job to not be considered identical to a pre-existing one with different JSON order.

  The UniqueOpts type also gains an ExcludeKind option for cases where uniqueness needs to be guaranteed across multiple job types.

  In-flight unique jobs using the previous designs will continue to be executed successfully with these changes, so there should be no need for downtime as part of the migration. However the v6 migration adds a new unique job index while also removing the old one, so users with in-flight unique jobs may also wish to avoid removing the old index until the new River release has been deployed in order to guarantee that jobs aren't duplicated by old River code once that index is removed.

  Deprecated: The original unique jobs implementation which relied on advisory locks has been deprecated, but not yet removed. The only way to trigger this old code path is with a single insert (Insert/InsertTx) and using UniqueOpts.ByState with a custom list of states that omits some of the now-required states for unique jobs. Specifically, pending, scheduled, available, and running can not be removed from the ByState list with the new implementation. These are included in the default list so only the places which customize this attribute need to be updated to opt into the new (much faster) unique jobs. The advisory lock unique implementation will be removed in an upcoming release.

  PR #590.

• Deprecated: The MigrateTx method of rivermigrate has been deprecated. It turns out there are certain combinations of schema changes which cannot be run within a single transaction, and the migrator now prefers to run each migration in its own transaction, one-at-a-time. MigrateTx will be removed in future version.

• The migrator now produces a better error in case of a non-existent migration line including suggestions for known migration lines that are similar in name to the invalid one. PR #558.

Fixed

• Fixed a panic that'd occur if StopAndCancel was invoked before a client was started. PR #557.
• A PeriodicJobConstructor should be able to return nil JobArgs if it wishes to not have any job inserted. However, this was either never working or was broken at some point. It's now fixed. Thanks @semanser! PR #572.
• Fixed a nil pointer exception if Client.Subscribe was called when the client had no configured workers (it still, panics with a more instructive error message now). PR #599.