Commit Graph

3586 Commits

Author SHA1 Message Date
Matt Aitken d2ee232fae Revert "Complete the original attempt span if retrying due to an OOM"
This reverts commit 5f652c6212.
2025-02-11 11:33:07 +00:00
Matt Aitken 5f652c6212 Complete the original attempt span if retrying due to an OOM 2025-02-11 11:01:21 +00:00
Matt Aitken 036a5061e1 Kubernetes OOMs appear as non-zero sigkills, adding support for treating these as OOMs 2025-02-11 10:32:01 +00:00
Matt Aitken bb65b2614a OOM retrying on larger machines (#1691)
* 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
2025-02-10 17:43:59 +00:00
Matt Aitken 4dd42cd6d4 Retry heartbeat timeouts by putting back in the queue (#1689)
* 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
2025-02-10 13:41:32 +00:00
Matt Aitken bc7d44582d If there’s no queue concurrency limit, default to the env concurrency limit (not 1m) (#1687) 2025-02-09 21:34:40 +00:00
James Ritchie 1f1a6b0405 Fixed runs page filter bug (#1681) 2025-02-07 17:32:30 +00:00
Eric Allam 9f843d0a0f Remove v2 event list (#1680) 2025-02-07 13:51:34 +00:00
Eric Allam 9dfd6a5a0a Rate limit alerts by channel for task run alerts using generic cell rate algo (#1679) 2025-02-07 13:36:37 +00:00
Eric Allam 5210d3a8bb Add internal spans to event repo (#1678) 2025-02-07 13:36:15 +00:00
Eric Allam b5867627be Use idempotency key index when finding existing runs by idempotency key in batch (#1677) 2025-02-07 13:36:03 +00:00
Eric Allam 0b555faf01 Force upgrade v1 batches to v3 (#1676)
* Force upgrade v1 batches to v3

* Actually lets process v1 batches async
2025-02-06 22:37:26 +00:00
Eric Allam dcf1ab6f38 Remove schedule constraints from TaskRun to prevent db load issues when a schedule or instance is deleted (#1674) 2025-02-06 19:16:48 +00:00
Matt Aitken c49af774ba Retry batch item completion (#1675)
* 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
2025-02-06 19:16:30 +00:00
Eric Allam f00ed9b13e Add the sha256 digest to the deployment image reference (#1673) 2025-02-06 19:14:34 +00:00
Eric Allam 3ec6983e45 Move run ttl and delays from graphile to redis worker (#1672) 2025-02-06 14:25:37 +00:00
Eric Allam a2c2d920b7 Use redis worker for run heartbeats and alerts (#1669)
* 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
2025-02-06 10:21:57 +00:00
nicktrn b946b9f38e Add optional env var and label to disable image pre-pulls (#1667)
* add env var to disable pre-pulls

* add label to disable pre-pulls on specific nodes
2025-02-05 16:49:57 +00:00
Eric Allam e61573600b Reduce contention on batchTaskRun when setting expected count (#1662) 2025-02-04 13:43:50 +00:00
Eric Allam 53e6d478ed Fix deploy timeout issues (#1661)
* deploy v2 streaming WIP

* finalize deployment now SSE and won't timeout

* handle zodfetchSSE connection errors

* Create tender-cycles-melt.md
2025-02-04 12:44:14 +00:00
nicktrn 7fa45ad337 Fix pre-pulling of deployment images onto worker nodes (#1660)
* lint

* get pre-pull data from finalized deployment
2025-02-04 10:07:07 +00:00
Eric Allam 6f52b00560 Set taskRunAttemptId on batch task run items on completion (#1658) 2025-02-03 15:25:12 +00:00
Eric Allam b377b577ca db efficiency improvements (#1657)
- 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
2025-02-03 13:59:58 +00:00
Eric Allam b67ea9a2f5 improve batch completion system for run engine v1 (#1656)
* 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
2025-02-03 11:29:51 +00:00
Eric Allam 3641647add Add the keyPrefix option when redis is not in cluster mode (#1655) 2025-01-31 10:50:52 +00:00
Eric Allam 813d73d1a3 Fix issue with heavy contention on TaskQueue updating concurrency limit (#1653)
* Fix issue with heavy contention on TaskQueue updating concurrency limit

* Remove concurrency limit when creating background worker tasks
2025-01-30 14:01:11 +00:00
Eric Allam 3eede3c339 Better handle redis error events (#1652) 2025-01-29 17:27:52 +00:00
Eric Allam b3afbcf558 Correctly configure redis cluster for aws elasticache (#1651) 2025-01-29 16:56:51 +00:00
Eric Allam f43de6ab74 Support redis/valkey cluster mode (#1650) 2025-01-29 16:28:51 +00:00
Eric Allam ef0fc6b0b4 Support for logically separated redis instances (#1647)
* 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)
2025-01-29 13:53:58 +00:00
Eric Allam a24b2a7732 More v1 run engine fixes (#1644)
* 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)
2025-01-28 16:16:37 +00:00
Eric Allam 6da5e7a4cf Various fixes for run engine v1 (#1643)
* 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
2025-01-27 17:15:17 +00:00
Eric Allam 246eb4361e If a deployment is already deployed, just return the deployment so the request succeeds (#1639) 2025-01-24 18:47:26 +00:00
Eric Allam f6eeaa1001 Fix broken cloud deploys by using depot ephemeral registry, skip the registry proxy (#1637) 2025-01-24 17:22:35 +00:00
Eric Allam eaf46bac58 engine v1 improved dequeue selection algorithm (#1632)
* engine v1 fairness improvements (env based queue selection)

* Add ability to bias env selection by concurrency limit and capacity

* Add some randomization to the list of queues within each evn

* Configure biases using env vars

* Typo

* revert v3 catalog project ref change

* Fix v2 marqs
2025-01-23 15:01:42 +00:00
James Ritchie 16cf357380 Focus visible style refinements (#1577)
* reduces the thickness of the focus-visible

* More subtle Input style

* Updates focus-visible styles for the SimpleSelect

* Tables tab through rows and action cells only

* Refined focus visible style

* Adds the custom focus visible to the TextLink

* Moves the onClick handling to the table row rather than the cell

* Makes table row heights consistent

* Adds gap between task search bar and button

* Prevents long tag values from wrapping

* Removes unnecessary rows from table header

* Added gap between search and filters

* Fixes the schedules table pagination staying fixed to the bottom of the page

* Removed more unnecessary header table rows to prevent them being selectable

* Removed isSelected styles (not working)

* Added <tr> back to the main Table compontent

* Table row handles modifier keys

* Adds to={path} to the TableRow only

* Revert "Adds to={path} to the TableRow only"

This reverts commit 8a814d434ff85b8ab9c2305d336074f659d86094.

* Revert "Table row handles modifier keys"

This reverts commit b2394741fa2ed6b908e00673b1a395ae89ca7146.

* Table reverted to use linked cells rather than rows

* Set the tab index of a cell and style the table row when tabbed

* Tabbed row style applied to the sticky cells

* Adds isTabbableCell to each table

* Improves the spcificity of the row highlighting

* Reduces the height of the task rows to match the other tables

* Adds tab styles to fill in row dividers top and bottom

* Removed old row onClick and to props

* Removed duplicate table header row

* Creates table style variants so tables look good in the inspector panels

* Full width tables on the schedules inspector panel

* Reduce padding on Alerts page table rows

* Updates the Deploy page inspector table to the new style

* Removes the duplicate Table Row from the table headers in v2

* Adds TableRow to the EventsTable

* Adds TableRow to more tables

* Adds TableRow to more tables

* Adds TableRow to more tables

* Adds TableRow to more tables

* Fix for showing correct cursor for linked table rows

* Removed variants from some of the table components
2025-01-21 21:11:59 +00:00
Eric Allam f96bf7210b Engine v1 improvements (#1627)
* Efficiency improvements to SharedQueueConsumer and MarQS for run engine v1

* Reliabily resolve task queue for a run and ack runs where we can't find the queue

* engine v1: stop nacking after reaching the max nack count, ack to remove the message

* Handle division by 0 possibility in distributeQueues
2025-01-21 14:24:02 +00:00
James Ritchie c855e4a96c v3 pages have page titles (#1582)
* Adds tasks page title

* Adds page titles to the <head>
2025-01-21 12:07:32 +00:00
Eric Allam 5750f35add Shared queue consumer telemetry improvements and perf changes (#1619)
* Shared queue consumer telemetry improvements

* findUnique -> findFirst to stop the prisma dataloader issue on api key auth

* Added action stats, audited all actions and improved reasons

* Better debug logging filtering
2025-01-17 09:15:21 +00:00
Matt Aitken 5eb30d026d Improve worker deployment queries (#1616)
* workerDeployment change findUnique to findFirst

* Changed the worker deployment queries to only select what they need, some of the columns are huge

* Only get the bare minimum data for tasks too
2025-01-15 20:34:28 +00:00
nicktrn 719ae83057 Add support for specifying machine preset at trigger time (#1608)
* add machine preset trigger option

* add changeset

* add examples to changeset

* machinePreset -> machine

* task options now also accept machine as simple string

* update openapi spec
2025-01-15 13:47:13 +00:00
Eric Allam 7eda22851e Prevent v2 batches from being resumed before all items have been created (#1614) 2025-01-14 14:37:04 +00:00
Eric Allam 14ce599218 Add reason to MarQS acknowledgeMessage to track why it's called (#1607) 2025-01-13 17:29:09 +00:00
Eric Allam 0a18e1d919 Additional dev queue consumer logging (#1606) 2025-01-13 16:37:19 +00:00
Eric Allam ca51f65394 Truncate ingested log messages to 4096, and truncate message to 256 in the summary tree view (#1604) 2025-01-13 11:35:39 +00:00
nicktrn 472e5f93e1 Fix: Prevent failed resumes for very large batches (#1598)
* findUnique -> findFirst

* remove tx

* make query more legible

* cut resume payload queries in half

* remove the additional status checks

* tweak the catalog batch parent task

* fix error message
2025-01-10 18:33:25 +00:00
Eric Allam ab69b60348 New metadata deployed task fix (#1588)
* Add more logging around new metadata system

* Make the authenticated env optional when updating run metadata
2025-01-08 22:42:34 +00:00
Eric Allam 3ca0c15161 Add ability to disable metadata flushing and logging (#1587) 2025-01-08 12:37:43 +00:00
James Ritchie 0aa597104c Shortcut improvements (#1573)
* Added shortcuts (with tooltip) to the pagination

* WIP fixing the disabled hover state

* Styled the tooltip

* Added a shortcuts panel to the help menu

* Adds new shortcut to list

* Changes “meta” for “mod”

* Adds more shortcuts to the list

* Adding shortcut to open the shortcuts panel

* tweak gap between shortcut letters

* button component now has icon spacing adjustment (for lucide icons)

* Fixed some ilegal markup

* Pagination uses disabled prop rather than a disabled wrapper

* Improved the Switch styles

* Makes the shortcut modifier optional

* Added new icon based shortcut keys for mac and win

* Updated PC modifier shortcuts

* Adds a new windows key icon

* Allows variants and react nodes to be used as the modifier key

* Adds more shortcuts to the storybook

* Adds missing focus-visible styles to the pagination

* Removed test modifier keys

* number style is tabular

* Update apps/webapp/app/components/primitives/ShortcutKey.tsx

Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>

* Tooltip now just 1 prop on the button component

---------

Co-authored-by: Matt Aitken <matt@mattaitken.com>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
2025-01-08 11:58:18 +00:00
Eric Allam 34f8bd588e Add ability to update parent and root run metadata from children (#1563)
* Increase the number of active streams from 2 to 5

* WIP parent metadata updates

* Fix noop metadata manager

* Implement run metadata updates from ancestor tasks

* Add changeset

* Add ability to stream into parent and root task runs

* Don't duplicate exporting run metadata types

* Add ability to fetch streams through metadata

* Couple of fixes from CodeRabbit

* Fix metadata tests

* Make sure streams are subscribed in the "background"

* Move the stream subscription stuff to the API client, expose it through `runs.fetchStream`

* Fixed run stream tests
2025-01-08 11:29:24 +00:00