* Improve the MarQS priority system by moving future messages into the LRE worker and using a priority timestamp offset to define priority in messages
* Add a test to make sure priority offsets don't unfairly favor environments
* requeuing should clear concurrency sets
* Heartbeats should only reschedule existing heartbeat jobs
* Fix type error
* New MarQS method requeueMessage, an atomic version of replace message
* Remove redundant call to remove the message from the env queue in requeueMessage
* run engine v1: orgs are no longer considered for concurrency
* Add reserve concurrency concept to allow waiting to resume parent tasks to release concurrency at the env level for child tasks to use (or else there is a deadlock). WIP recursive tasks
* child tasks inherit the queue timestamp from their parent tasks to prioritize completing child tasks based on when their parent started
* handle reserve concurrency with recursive deadlocks
* Finish docs update for concurrency
* Some fixes from badge conflict resolution
* WIP priority queues
* Implement MarQS priority queues
* Fix the migrations
* add timeout support to sendWithAck
* coordinator will retry completion submission
* actually retry
* increase default retries
* something went wrong there, add this back in
* add changeset
* don't crash run on lazy attempt errors and count on retry
* remove abort controller after checkpoint in all cases
* duration checkpoints
* manual checkpoints
* add busybox to coordinator image
* add async toggle to process in background
* remove early returns
* add changeset for manual checkpoint schema
* WIP with webhook SDK function and types
* JSDocs added to the schema
* Webhooks are working
* Expanded the alert docs
* Remove duplicate export of waitUntil.js
* Use uncrypto
* Don’t rate limit webhooks
* Create slow-olives-fix.md
* remove unused imports
* tell run to exit before force requeue
* handle exit for case where we already retried after oom
* improve retry span and add machine props
* don't try to exit run in dev
* WIP fix for ResumeAttemptService selecting the wrong attempt (which has no error or output)
* Don’t create an attempt if the run is already in a final status
* Don’t get all the columns for the query. Improved the logging.
* Added a log to the batch example
* Filter out the undefined values
* Create new partitioned TaskEvent table, and switch to it gradually as new runs are created
* Add env var for partition window in seconds
* Make startCreatedAt required in task event store
* OOM retrying on larger machines
* Create forty-windows-shop.md
* Update forty-windows-shop.md
* Only retry again if the machine is different from the original
* If there’s a heartbeat error and no attempts we put it back in the queue to try again
* When nacking, return whether it was put back in the queue or not
* Try and nack, if it fails then fail the run
* Consolidated switch statement
* Fail executing/retrying runs
* Added isPrismaRetriableError()
* Retry completeBatchTaskRunItem if they fail because of a retriable Prisma error
* Retry using Redis worker
* Handle more retriable errors. Add special condition in for race condition error
* Added Postgres connection_timeout with default 20s
* Added a simple batchTriggerAndWait example
* Move the task run heartbeats to RedisWorker
* Move alerts to redis worker, improving redis worker
* Fix typecheck errors
* Use single threaded tests for redis worker
* Enable/disable the redis workers independently
* Remove preview release from PR checks
- remove calls to findUnique, especially when including/selecting relations
- add some missing indexes
- add spans to $transaction calls to help track down long running txs
* Automatically retry TriggerTaskService when hitting a unique constraint error on idempotency key
* improve batch completion system for run engine v1
* Rename batch stuff to v3 so it's not confusing
* Handle unique constraint error on BatchTaskRunItem creation and allow different limits for batchTrigger and batchTriggerAndWait
* Remove unnecessary disabled org check
* Add support for separate redis servers for caching, rate limiter, marqs, and the pub/sub stuff (they all default to the previously used env vars for redis)
* Remove the socket count from shared queue consumer cause it can take up to 5s to complete
* Add support for re-using queue snapshots across dequeues
* Only consider the top N orgs when dequeuing, to help mitigate large spikes in queues (like around the hour and half hour marks)
* Various fixes for run engine v1
- Make sure there are connected providers before sending a scheduled attempt message, nack and retry if there are not
- Fail runs that fail task heartbeats when pending and locked
- More and better logging around shared queue consumer
- Fix bug when failing a task run with no attempt
* Prevent findUnique from bringing down our database
Eric Allam