Released on September 7, 2022.
WARNING: This release of the Helm chart includes a breaking change with the upgrade from PostgreSQL 9 to 11. This is required because bitnami no longer hosts the old version of Postgres. If updating from a prior Helm chart release, please take care to back up your database.
- Allow users to specify a Kubernetes config map to be mounted as
jobTemplateFilein the agent's container - #379 - Improve init container template for graphql - #370
- Create agent's env vars only if input variables are not empty - #380
- Update PostgreSQL from ~9 to ~11 — #375
- Allow specification of resources for the tenant creation job — #371
Released on April 14, 2022.
- Fix issue where the Helm chart did not correctly set UI/Core image version tags - #362
Released on March 29, 2022.
- Reduce maximum flow runs rescheduled by Lazarus at once - #350
- Add
includeChartNameInComponentstoggle to prefix component names with the chart name for improved UX as a subchart - #355 - Fix missing server configs on towel deployment - #346
Released on January 25, 2022.
- Upgrade version of Hasura GraphQL engine to v2.1.1 - #337
Released on January 12, 2022.
- Upgrade Hasura to
v2.0.9- #328
Released on December 4, 2021.
- Allow users to disable/enable apollo playground - #326
Released on November 30, 2021.
- Allow configuration of the maximum artifact payload - #320
- Fix agent environment block construction - #312
- Allow annotations to be provided for service accounts - #314
- Allow custom flow run job template path to be provided to agents - #322
Released on November 9, 2021.
- Preserves structural representation of serialized flows - #306
- Upgrade
apollo-serverto address GraphQL playground vulnerability. See the advisory for details.
Released on October 21, 2021.
- Add ability to set keep alive timeout for graphql server - #290
- Add handling a keepalive value to Apollo via env variable APOLLO_KEEPALIVE_TIMEOUT - #293
- Add a submitted state lock to prevent the same flow from being run by multiple agents - #291
- Add default flow parameters to flow runs to have consistent flow run parameter history - #298
- Fix setting
imagePullSecretsin k8s jobs with Prefect flows in Prefect Server Helm chart - #289
Released on September 2, 2021.
- Make maximum number of scheduled runs per flow configurable - #280
- Increase Apollo payload size limit to 5mb - #273
- Separate out task and edge registration to better handle large flow registrations - #277
- Update Helm chart RoleBinding API version - #276
Released on July 6, 2021.
- Add traceback info to some unexpected logs for debugging - #259
- Add the option to defer setting a flow run state to scheduled when creating a FlowRun - #261
Released on May 25, 2021.
- Introduce **kwargs to
create_logsfor Cloud compatibility - #247
Released on April 27, 2021.
- Simplify
LoopService.loop_secondshandling and add debug mode to speedup tests - #242
- Add unique constraint for flow run idempotency - #219
Released on April 6, 2021.
- Helm: Add ingress support for
uiandapollocomponents - #229
- Log
set_schedule_activefailure while creating a flow - #236
Released on March 20, 2021.
- Add more permissions to agent role in helm chart - #221
- Set uvicorn server log level and access logs - #217
- Ensure the scheduler queries with deterministic order - #215
- Ensure that errors in scheduling do not interrupt other scheduling operations - #224
- Improve run detail triggers when version is the same or missing - #218
Released on March 10, 2021.
- Release the Server Helm Chart to a Helm repository - #209
Released on March 6, 2021.
- Add
APIErrorto replace 'database query error' and consolidate Cloud/Server exception types - #204 - Improve
archive_flowperformance with auto-scheduled flow runs - #206
- Allow
create_agentto be called without passing settings - #207
Released on March 4, 2021.
- Upgrade hasura version to 1.3.3 - #188
- Upgrade apollo base image to debian-buster - #198
- Install
tinifrom apt instead of GitHub in apollo image - #199
- Fixed the execution of the
create-tenant-jobin the Helm charts to allow for custom Apollo ports - #201 - Improve
delete_flow_runperformance by avoiding slow deletion triggers - #202
- Server instances with managed databases require manual SQL to upgrade - #188
Released on February 22, 2021.
- Add
tiniinit process to GraphQL and Apollo containers to reap zombies - #191
Released on February 18, 2021.
- Capture all 'connect' errors and uniformize into DB query errors - #189
Released on February 12, 2021.
- Helm: Optionally create the default tenant before running Agent - #183
- Cancel ad-hoc created flow runs instead of deleting them when archiving a flow - #185
- Upgrade pydantic to 1.7.x - #187
- Fix asyncpg errors on Apple M1 / arm64 - #186
Released on February 5, 2021.
- Install curl in server docker image for healthchecks - #182
- Improve
LoopServiceinterval handling - #43
- Move to
YYYY.MM.DDversioning scheme away fromYYYY-MM-DD- #178
Released on January 25, 2021.
- Add 'description' field for Flow Groups along with routes for updating - #177
Released on January 19, 2021.
- Add
sumaggregate to ORM - #174
Released on January 5, 2021.
- Support per-flow-run and per-flow-group
run_configoverrides - #166
- Upgrade Apollo NodeJS to
14.15.1- #153 - Helm: configure strategy for deployment - #165
- Allow for timezone specification on flow group schedules - #169
- Add
flow_group.run_configandflow_run.run_config- #166
Released on December 16, 2020.
- Add the
eventsRBAC permission to agents created by the helm chart - #157
Released on December 14, 2020.
- Add documentation on how to connect the CLI to a remote deployed server - #159
- Refactor the Zombie Killer to rely entirely on flow run heartbeats - #158
- Add support for custom Hasura root fields - #161
- Add jobs/status to role for agent - #154
Released on December 4, 2020.
- Add flag for shutting down loop services - #144
- Support configurable primary keys in ORM - #145
- Add support for expanded Hasura update operations - #146
- Helm chart: Always pull prefecthq docker images by default - #151
- Listen on all agent interfaces for healthchecks - #148
- Add task run name index for text search - #149
Released on November 29, 2020.
- Add KubernetesAgent to the Helm chart - #136
- Add two new GraphQL routes for Core functionality - #143
- Fix handling for managed postgres usernames with
@signs - #139
Released on November 13, 2020.
- Add the Prefect Core version to the api query - #126
- Remove global 24 hour expiration on flow run idempotency keys - #134
- prefect-server dev infrastructure would sometimes fail to start with 'connection closed' - #130
- Fix helm chart when using existing postgres instance - #132
- Fix scheduling duplicate runs due to expired idempotency keys - #134
Released on November 11, 2020.
- Add experimental Helm chart for deploying in K8s - #123
Released on November 10, 2020.
- Add API for persisting and retrieving task run artifacts - #121
Released on October 29, 2020.
- Add idempotency keys to
flows.create_flow- #116
- Expose database upgrade errors during infrastructure start - #117
Released on October 27, 2020.
- Check for existence of agent before updating flow run agent - #114
Released on October 22, 2020.
- Update flow settings graphql logic to use supported API routes - #113
Released on October 14, 2020.
- Allow for scheduling different parameters and different run labels at the exact same time - #111
Released on October 13, 2020.
- Allow for scheduling changing labels on a per-flow run basis - #109
- Fixes issue with Agent Config schema - #107
Released on September 24, 2020.
- Registering agents will now retrieve IDs from matching entries in order to prevent duplication - #95
- Add routes for setting run names - #77
- Fix
namefield indelete_agentGraphQL resolver - #95 - Replace all uses of
database.modelswithprefect.models- #94
Released on September 15, 2020.
- Hasura metadata archives use the most recent alembic revision ID - #78
- Fix and consolidate behavior for active schedules - #86
Released on September 11, 2020.
- Add database structure and routes for persisting Agents - #58
- Add route for querying mapped children - #73
- Fix init.py for deprecated API module - #71
- Remove
run_countanddurationcolumns - #72
- Increase performance of state update triggers - #72
Released on September 1, 2020.
- Automatically upgrade local database - #69
- Fix flow run creation not using default parameters from flow group - #64
- Ensure Lazarus ignores flow runs with active tasks - #67
- Deprecate flow settings that are actually flow group settings - #63
Released on August 26, 2020.
- Refactor Lazarus and Zombie Killer for modularity - #46
- Add API mode to queryable reference info - #59
Released on August 21, 2020.
- Upgrade Hasura to 1.3.0 - #18
- Set schedule to inactive if neither Flow nor Flow Group have a schedule - #37
- Dedupe flow group labels when set - #40
- Set Apollo's
post-start.shscript to wait for the GraphQL service's healthcheck - #41 - Upgrade Hasura to 1.3.1 - #52
- Use Hasura default cache size - #52
- Adds RELEASE_TIMESTAMP for UI to display - #54
- Remove unused UI from services CLI - #45
- Cloud hooks require explicit states - #51
- Cloud hooks do not match state parents - #51
- Remove unused code and config settings - #53
- Remove unused
state_idcolumn - #35
- Release
- None
- None
- Fixes bug where updating a Flow Group schedule didn't clear old scheduled runs - #9
- None
- None