* WIP
* Allow marqsv2 and v2 graphile to run in parallel
* Fix missing GraphileLogger import
* Fixed heartbeat after rebase
* Replace postgres based run counters with redis ones with a backfill
* Add back in the graphile logger
* Remove duplicate visibility timeout calls
* Clamp simple weighted strategy to max of 5
* Created a v3-catalog test script for queues
* SDK: Fix for calling trigger and passing a custom queue
* Support custom queue in TriggerTaskService
* Improved the script in the catalog so it’s clearer what’s going on
* Remove the concurrencyLimit from a queue if the limit is null
* Fix for the test code… stupid
* Improved the existing runs API
* WIP next runs API
* Improve the returned ApiPromise to add ability to return response
* More WIP
* WI{
* Added offset/limit pagination stuff like the cursor one, and converted all API methods to use ApiPromise
* More run API stuff
- Adding schedule output from the retrieveRun endpoint
- Ability to filter by schedule and isTest
* Remove env from retrieve run in openAPI
* prefer duplication over merge
* WIP docs
* Use spread to DRY up some run API schemas
* Finish the overview docs
* Adding changeset
* Fixed typecheck errors
* Typo fix
* Re-export zodfetch from core so the v3 CLI can use it
* Fixed type errors
---------
Co-authored-by: Matt Aitken <matt@mattaitken.com>
* WIP worker TaskRunAttempt creation
* Handling failing task runs that cannot create an attempt for whatever reason
* Move the visibility queue stuff into a graphile job
* Fixed task runs with unsanitized queue names
* “Borrow” the code from alerts PR to get self hosted deployments working
* Add an admin API endpoint to get info about the shared marqs queue
* Allow admins to view any project metrics
* start adding lazy attempts to prod
* lazy attempt creation for prod workers
* resurrect prod stack traces
* add exception event to failed run spans
* simplify dependency resumes
* fix typecheck
* fix merge
* fresh process for all attempts
* always try sigterm first
* stop heartbeat timeout on non-inplace replace message
* add missing ack on checkpoint creation service failure
* bypass dequeue for retries with running worker
* respect retry delays
* crash runs with invalid run status for execution
* remove debug logs
* fix nack message
* fix version locking
* fresh attempt processes in dev and prod
* improve handling of ipc timeouts
* consider checkpoint failures on cancellation
* add basic chaos monkey to checkpointer
* changeset
* control forced checkpoint simulation via env var
* fix merge
* kill old attempt processes before checkpointing
* detailed perf logging for checkpointing
* add coordinator otlp endpoint example
* improve prod run cancellation
* rename supports lazy attempts migration
* fix graceful exit
* fix retry mechanics
* clear paused state before retry
* remove checkpoint image after push
* crash worker on unrecoverable errors
* refactor unrecoverable error emit
* switch to do hosted busybox image
* increase wait for duration ipc timeout
* add changeset for misc fixes
* fix merge
* fix retry delay span runId
* fix dev retries
* improve prod worker logging
* log checkpoint sizes
* add lazy attempts catalog entries
* Fixed merge issue: use zodFetch, not wrapZodFetch
* Revert "Fixed merge issue: use zodFetch, not wrapZodFetch"
This reverts commit d137e4e1fe.
* importEnvVars uses wrapZodFetch now
* add backwards compat for retries without checkpoints
* handle more cases of unrecoverable runs
* don't kill the child process if it shouldn't be killed
---------
Co-authored-by: nicktrn <55853254+nicktrn@users.noreply.github.com>
Co-authored-by: Matt Aitken <matt@mattaitken.com>
* WIP env var management API
* Add import env var API endpoint
* Adding docs and support for using both API keys and PATs when interacting with the env var endpoints
* WIP envvar SDK
* Uploading env vars in a variety of formats now works
* Finish env var endpoints and add resolveEnvVars hook
* Add changeset
* Fix: API rate limit error has the correct seconds until reset
* When a v2 run hits the rate limit, reschedule using the reset timestamp
* Still throw AutoYieldRateLimitErrors
* Reschedule runs from the rate limit
* The stress test timeout should be inside the task
* If the rate limit error is thrown, don’t retry the API request
* WIP on multi-select
* WIP on simple checkbox
* CheckboxWIthLabel and Checkbox
* Multi-selection of runs across pages is working
* Fix for selection on seconds page
* Focus the run filter on page load
* Don’t focus the checkbox
* BulkActionBar now shows/hides and has buttons
* Some state to stop escape clearing the selection when the modals are open
* Delete unused formData util
* Improvements to the page
* Created the replay resource action. It doesn’t do anything useful yet.
* Database schema created for BulkActionGroup/BulkActionItem
* The BulkActionService is creating the right data, now we need to process it
* WIP on bulk processing
* Added failed state and made the sourceRun required
* Bulk replaying is working
* WIP on bulk action filtering
* Fixed bulk filters displaying
* Filtering by batch is working
* Some fixes for the bulk id filtering
* Style tweaks
* Load the extra info in parallel
* Bulk canceling working
* Get the most recent 20 bulk actions to display in the filter menu
* Even if the run isn’t cancelable add it to the final list
* Maximum of 250 runs can be bulk actioned
* Don’t let them select more than the maximum (250 currently)
* Separate each bulk item action into it’s own separate graphile job to increase resiliency
---------
Co-authored-by: Eric Allam <eallam@icloud.com>
* Alerts v1
* Encrypt alert webhook secrets and allow them to be generated by the server
* Alert v1 UI
* Remove unnecessary emails
* Move to using `@react-email/components`
* WIP slack alerts
* More slack alerts WIP
* Update pnpm lock after rebase
* Finish implementing Slack alerts
* Use a more error like emoji
* New secondary variant for the segmented control
* Added a simple checkbox style variant to storybook
* Style tweak to the segmented control
* UI improvements to the alert modal
* Use searchable Select for alerts. Changed default variant for SegmentedControl
* Secondary button now using secondary colour
* segmented control style tweak
* Improved the channel column in the alerts table
* Updated logo-mono.png
* Updated email styles
* Email templates updated to new styles
* Don’t log the decrypted secret
* await enqueing the deployment alert when an index fails
* await enqueing the timeout alert
---------
Co-authored-by: James Ritchie <james@jamesritchie.co.uk>
Co-authored-by: Matt Aitken <matt@mattaitken.com>
* fix spacing for delete hints
* don't try to resume deps on wait resume
* sending duration wait resumes is not an error anymore
* set correct status with new wait resume flow
* cancel checkpoint schema v2
* don't mix messages and schemas
* prevent unintended case fallthrough in tree view
* completely switch to platform-led duration wait resumes
* prevent infinite restores
* some entries for the catalog
* add pg to additional packages
* add checkpoint safe timeout
* prevent duplicate spans after restore
* wait for post start
* add sdk version to deploy tab
* fail on impossible checkpoint scenarios
* remove debug logs
* lotsOfLogs task now outputs much larger logs
* Removed tree view collapse/expand animation
* JSDocs for useDebounce
* WIP moving filtering into the state
* Reworked the reducer to do the filtering
* Adds support for `emitDecoratorMetadata: true` and `experimentalDecorators: true` in your tsconfig
* Implement task.onSuccess/onFailure and config.onSuccess/onFailure
* Added onStart and more docs for lifecycle functions
* Use onStart instead of init for TypeORM
* Fixes an issue that was treating v2 trigger directories as v3
* Make msw a normal dependency (for now) to fix Module Not Found error in Next.js.
* Extract out all the zod* stuff from core so the SDK does not import it
* Add a changeset
* Fixing typecheck errors in the webapp
* Export the Task and TaskOptions types
* Extract additional exports from core/v3 that aren’t used in the SDK
* Move to our global system from AsyncLocalStorage for the current task context storage
* Update the esbuild core bundling plugin for the new core v3 exports
* Fix v3 CLI telemetry
* Add support for tasks located in subdirectories inside trigger dirs
* Remove the env var check during deploy (too many false negatives)
* Fixes various issues with triggerAndWait and batchTriggerAndWait
When using idempotency keys, triggerAndWait and batchTriggerAndWait will still work even if the existing runs have already been completed (or even partially completed, in the case of batchTriggerAndWait)
- TaskRunExecutionResult.id is now the run friendlyId, not the attempt friendlyId
- A single TaskRun can now have many batchItems, in the case of batchTriggerAndWait while using idempotency keys
- A run’s idempotencyKey is now added to the ctx as well as the TaskEvent and displayed in the span view
- When resolving batchTriggerAndWait, the runtimes no longer reject promises, leading to an error in the parent task
* Remove the default queue concurrency limit as we now have env and org concurrency limits
* Use the run friendlyId in the completion result id
* Added some error logging
* graceful exit with timeout
* handle and display graceful timeout errors
* fix for very long waits
* changeset
* increase termination grace period to an hour
* fix dev dependecy resumes
* add catalog entry with multiple trigger waits
* update docker provider lifecycle command
* switch to x for clearing run filters
* Revert "fix dev dependecy resumes"
This reverts commit b4061f2ae71f21889adcf061928e2e77c436f0ed.
* fix dependency resumes, properly this time
* add catalog entry for dependency waits in loops
* advice in docs re parallel dependency waits
* fix link from v3 to v2 docs
* move lifecycle command logging to debug only
* Removed batchOptions from the trigger options and the docs
---------
Co-authored-by: Matt Aitken <matt@mattaitken.com>
* Improve the SDK function types and expose a new APIError instead of the APIResult type
* Skip triggering scheduled tasks if the task isn’t found in the current deployment
Also fixes an issue when editing the environments of a schedule
* Database schema and migrations for schedules
* Added schedules to the side menu
* The pagination can optionally hide the page numbers for a compactive mode
* Filters for the schedule page
* Added triggerSource (“STANDARD”, “SCHEDULED”) to BackgroundWorkerTask
* Added the ability to disabled a LinkButton
* Started work on the schedule page
* Environment buttons
* The new schedule form styles
* Added a cxouple of extra fields
* Allow a checkbox to have a rich label
* Added cronstrue package to the webapp
* WIP creating tasks using the form
* Improved the form styling
* Creating schedules is working in the UI
* Minor improvements
* Basic schedule table is displaying
* Creating tasks with triggerSource = scheduled. Refactored how task metadata is stored and accessed to be cleaner
* Resource route
* Fixes in the form
* Added a gap between the environments
* WIP on OpenAI generating CRON expressions
* AI generated CRON expressions is working
* Fix for the CRON field being uneditable after an AI generation
* Improvements
* Table padding
* useThrottle now behaves correctly
* Added filtering to the schedules list
* Improved the layout and fixed CRON search
* Fixed pagination for the schedules list. Just use a regular Prisma query
* Page size of 20
* Added links to the schedule rows
* Get rid of the Last run column for now
* Implement triggered scheduled tasks
Also implemented superjson payloads and dev runtime environment “presence” with RuntimeEnvironmentSession
* Move CronPattern and CreateSchedule into a common client-accessible file
* Latest UI changes
* Fix for creating a task schedule with a blank dedup key
* Refactor the human to cron stuff into a separate file and use json_object OpenAI response format
* Fix for trying to use a hook on the server-side…
* A couple of fixes to the new schedule form
* WIP on viewing a scheduled run
* Make the filters all optional
* Use the RunListPresenter from the schedule presenter
* Display a table of runs… the wrong runs but still
* Runs from the schedule
* Deleting schedules from the UI
* Tidied imports and fixed name of options object
* Disabling a schedule
* Editing schedules
* Tidied imports
* Added icons to the task list, needs some design love
* Added a tooltip for CRON pattersn
* Show the last run in the schedules table
* Some tweaks
* Added a placeholder to the CRON AI field
* Improved the trigger source icon
* Dim out disabled schedules
* Scheduled tasks have the correct icon in a run
* Added the task source icon to the test task list
* Added the date field component to storybook
* Style improvements to the date field
* Implement Task Schedule API
* added a medium sized variant to the date field
* Fixed replay run for superjson payload types
Also now linking from the replaying run to the original run (using span links)
Also added a project metrics prometheus endpoint to detect the state of the queues
* WIP on allowing different forms for testing
* If you pass a string to prettyPrintPacket which is json or superjson, safeParse it first
* Test page, deal json and superjson. Started splitting UI for schedules
* WIP on schedule form
* Removed the Label from the DateField
* WIP on test schedule form
* Fix for the runs page showing the wrong message when there are no runs from time filtering
* Removed labels from the DateField
* Fixes for the form
* Test runs are working for schedules
* Fix for nextScheduledTimestamps in triggerScheduledTask
* Implement idempotency key support and fix issue with cancelled runs causing concurrency usage
* Added API documentation for the schedule API
* Remove log
* Removed console log from runs page
* Transform the recent runs test data on the server
* Fix for hydration mismatch
* Current date as the default for the test form
* Recent payloads working
* Delete schedule modal
* Deal with empty strings from the form
* Set the initial value for the scheduled test form
* Add option to print console logs in the dev CLI locally (issue #1014)
* Export queue from the SDK
* Fix for schedules list when you have no schedule tasks
* Blank states improved
* Make task schedules more generic, to support additional schedule generators in the future
* Removed log from maqrs
* Removed “v3/schedules” export from the SDK
---------
Co-authored-by: Eric Allam <eallam@icloud.com>
Co-authored-by: James Ritchie <james@jamesritchie.co.uk>
* WIP on replaying a task from the run page
* Don’t pass the existing runs idempotency key, it will cause the replay to always return the original run
* Replay from the run list
* Don’t use fetchers in the replay/cancel dialogs
* API endpoint for replaying a run
* REST API docs (mostly coming soon) but added replay run
* replayRun function added to the SDK
* Cancel run added to the SDK
* v3-catalog file to test canceling and replaying
* Changed the SDK to be runs.replay and runs.cancel
* Removed comment
* Latest lockfile
* WIP fixing pnpm nested install strat
* Add changeset
* Pass (and log) cwd during background worker initialization
* Log out the env
* manually add the .pnpm/node_modules to NODE_PATH if in a pnpm repo
* Remove supports color dep
* Upgrades to MarQS
- It’s “queue choosing” algorithm is now MUCH better (pretty much didn’t work before) and should be performant even when we have a bunch of prod queues
- There are now concurrency limits at the environment and organization level, as well as the task/queue level. So if any of them are at capacity the message won’t be dequeued. This means we can have an org wide concurrency limit, as well as different limits for dev/prod/staging
- I’ve added an admin API that can be used to update the org/env concurrency limits
* More MarQS upgrades
- Extract the queue priortity choosing strategy into an interface
- Implement a much better weighted average strategy
- “Slide the window” of queue candidates if the parent queue sends all at-capacity queues (so we won’t get stuck attempting to choose the same 12 full queues)
- Added some unit tests for the priority stuff
- Added some ideas for expanding the priority choosing strategy with more dynamic features
* Fixes an issue with the shared queue consumer relying on the queue name to get the env id
* v3: implement configurable log levels via config file and TRIGGER_LOG_LEVEL
Also, test runs automatically set the TRIGGER_LOG_LEVEL to debug
* Fix type error and changeset
* Added a Node.js runtime check for the CLI dev command
Eric Allam