Important
- This release includes large, internal changes for handling staging. As a result, some staging tasks may fail during deployment when the cloud controller jobs are updating.
- DEA is no longer supported as a backend runtime.
Highlights
- Introduce a
read
permission for service brokers to query when displaying dashboards for service instances. Read here for more. - Fixes some issues regarding tasks erroneously failing or being stuck in a
RUNNING
state. - Introduce Builds resource in the V3 API to represent the staging process.
CC API Version: 2.84.0 and 3.19.0
Service Broker API Version: 2.11
CAPI Release
Cloud Controller
- API client can discover the network-policy api location details
- API client can observe STAGING_START and STAGING_STOP usage events based on builds details
- API client can see that LANG environment variable is set for tasks details
- API client should NOT receive errors when creating/deleting buildpacks concurrently details
- API client should NOT see any fields related to builds on droplet details
- API client should create droplet by POST /v3/builds instead of POST /v3/packages/:guid/droplets details
- API client should see fields nested under result move to the top level on droplet details
- API client should see that builds that do not finish staging are eventually marked failed details
- As a platform, I can send a context parameter to brokers when creating a service instance details
- Bit service can update package status and checksums details
- Bump nokogiri to >= 1.7.2 details
- CF users should need
cloud_controller.read
scope rather thancloud_controller.write
to view isolation segment for org an space details - Cloud Controller double-counts memory (and disk?) against quota when updating a task details
- Operator should observe that an interrupted sync job does not result in waiting 4 hours until bulking continues details
- Sync job should continue to update LRPs, tasks, and bump freshness even if a LRP Update returns a
Error_InvalidRequest
details - API Client should NOT see 500 error when GET /v2/organizations/:guid/users has an error communicating with UAA details
- API client can discover from a build the user that created the resource details
- API client can observe build create and droplet create audit events at the correct points in the staging lifecycle details
- investigate route deduping issue for buildpack vs docker with legacy bridge details