Releases: secretflow/kuscia
v0.8.0b0
Added
- Kuscia supports arm architecture.
- Kuscia Job supports retrying, pausing, and canceling capability.
- Added documentation "How to use the Kuscia API to run a SecretFlow Serving".
- Added documentation "Introduction to Kuscia listening ports".
- Added documentation for DataMesh API.
- Added kuscia.sh and init_example_data.sh scripts.
Changed
- Modified the default task state in the job query results from Kuscia API.
- Optimized processing flows for KusciaJob and KusciaTask.
- Added isTLS field to the kuscia API for creating DomainRoute interface.
- Kuscia API has added a Suspended state enumeration to kusciaJob.
- Revised container launch deployment steps in point-to-point and centralized deployment documentation.
- Modified the pre-validation logic of the master address when initiating lite mode.
Breaking Changes
- Deprecated start_standalone.sh and deploy.sh. Recommended using kuscia.sh.
Fixed
- Fixed an occasional concurrent read-write map issue causing panics in the DomainRoute module.
- Fixed an issue with image repository names in the Agent module.
- Strengthened integration tests for improved stability.
- Fixed an issue where changes to the Configmap in Runk mode did not take effect for Serving.
- Fixed an error when creating DomainDataSource with Kusica API.
- Fixed a potential issue with abnormal startup when the protocol is set to TLS in the Kuscia configuration file.
v0.9.0.dev240508
Added
- [Documentation] SCQL on Kuscia Tutorial.
- [Feature] Support for ARM architecture.
- [Feature] Support for reverse tunneling multiple replicas.
- [Feature] New CGROUP restrictions under RunP and RunC for Kuscia nodes.
Changed
- [Security Hardening] Change the node registration request's MD5 hash to Sha256.
- [Security Hardening] Node certificate issuance requires the CN field to be a Domain name.
- [Script] kuscia.sh upgrade retains k3s resources, memory restrictions.
- [KusciaAPI] KusciaJob's TaskStatus new field alias.
- [Integration Testing] Improved integration testing stability.
- [Optimization] Modified the Access-Domain information that allows access to the Service.
- [Upgrade] Upgraded the dependent Envoy version to 1.29.4.
Fixed
- Fixed the issue where the kusciaTask status was incorrect when stopping a KusciaJob.
- Fixed the issue of nodes not scheduling applications upon deletion/upgrade.
- Fixed the issue of not being able to delete task sub-resources.
- Fixed the issue where Service name was too long, causing task creation to fail.
v0.7.0b0
Added
- Add serving docs.
- Add kuscia ports description.
Changed
- Optimize kuscia agent config
- Update ss metrics
- Update standalone.sh path pre_check
- Update scheduler log
- Update check domain exists
- Reduce additional metrics of ssexporter
- Update kuscia config doc
Fixed
- Fix kusciaTask default state
- Fix domainRoute rolling update and refactor code.
- Fix datamesh probe
- Fix invalid repository name
- Fix the command of getting runtime
- Fix typos in kusciajob_cn.md and kuscia-monitor.md
- Fix datamesh bazel file
v0.8.0.dev240409
Fixed:
- fixed envoy 404
v0.8.0.dev240331
Added
[Alpha] Kuscia Job now has new interface capabilities for retrying, pausing, and canceling jobs.
[Alpha] One-way Networking: Support for collaboration where only one party needs to expose a public port, eliminating the need for the other party to also do so. Documentation to follow.
[Documentation] How to use the Kuscia API to run a SecretFlow Serving.
[Documentation] Introduction to Kuscia listening ports.
[Documentation] Added documentation explaining DataMesh interface.
[Scripts] Added kuscia.sh and init_example_data.sh scripts.
Changed
Modified the default task state in the job query results from Kuscia API.
Optimized processing flows for KusciaJob and KusciaTask.
Added isTLS field to the kuscia API for creating DomainRoute interface.
Kuscia API has added a Suspended state enumeration to kusciaJob.
Revised container launch deployment steps in point-to-point and centralized deployment documentation.
Modified the pre-validation logic of the master address when initiating lite mode.
Breaking Changes
Deprecated:start_standalone.sh,deploy.sh. Recommended use kuscia.sh
Fixed
Fixed an occasional concurrent read-write map issue causing panics in the DomainRoute module.
Fixed an issue with image repository names in the Agent module.
Strengthened integration tests for improved stability.
Fixed an issue where changes to the Configmap in Runk mode did not take effect for Serving.
Fixed an error when creating DomainDataSource with Kusica API.
Fixed a potential issue with abnormal startup when the protocol is set to TLS in the Kuscia configuration file.
v0.6.0b0
Added
Upgrade interconnection protocol from kuscia-alpha to kuscia-beta to support interconnection between Kuscia-Master and Kuscia-Autonomy.
Kuscia monitor, Kuscia exposes a set of metric data, which can be used as data sources for collection by external monitoring tools (such as Prometheus).
The Kuscia API added a job approve interface ,allowing participants to review jobs .
Add some pre-check before kuscia running, such as health check of the connection of database.
Add parameter validation to the kuscia api.
The create job interface of kuscia API added the attribute 'customed-fields' .
Support configuring the application's image ID in AppImage to prevent domain's application image from being tampered with.
Added the curl command example for requesting the kuscia API.
polish the agent runtime docs.
Changed
Changed some kuscia-crds (KusciaJob,KusciaTask,KusciaDeployment)from cluster to namespace (cross-domain).
Fixed
Correct some inaccurate descriptions in the document.
v0.7.0.dev240229
[0.7.0.dev240229] - 2023-02-29
Added
- add the documents of datasource api
- add kusciaapi errorcode doc
- add kuscia init command generating kuscia config
Changed
- update domain register and handshake error
- report events when Pod failed to start in RunK mode
Breaking Changed
- Change the mounting directory for logs and data of Kuscia deploying with Docker
- {{ROOT}}/kuscia-{{DEPLOY_MODE}}-{{DOMAIN_ID}}-data -> {{ROOT}}/{{DOMAIN_CONTAINER_NAME}}/data.
- {{ROOT}}/kuscia-{{DEPLOY_MODE}}-{{DOMAIN_ID}}-logs -> {{ROOT}}/{{DOMAIN_CONTAINER_NAME}}/logs.
- {{ROOT}}/kuscia-{{DEPLOY_MODE}}-{{DOMAIN_ID}}-certs was deleted.
Fixed
- fix some unit test case
- fix the issue of inconsistent states among multiple parties in KusciaDeployment
- fix the issue of ClusterDefine only having unilateral information under KusciaDeployment in P2P scenarios
- fix kusciaapi grpc role check
- Upgrade certain dependency packages to fix security vulnerabilities in pkg.
v0.6.0.dev240131
Added
- Upgrade interconnection protocol from kuscia-alpha to kuscia-beta to support interconnection between Kuscia-Master and Kuscia-Autonomy.
- Kuscia monitor, Kuscia exposes a set of metric data, which can be used as data sources for collection by external monitoring tools (such as Prometheus).
- The Kuscia API added a job approve interface ,allowing participants to review jobs .
- Add some pre-check before kuscia running, such as health check of the connection of database.
- Add parameter validation to the kuscia api.
- The create job interface of kuscia API added the attribute 'customed-fields' .
- Support configuring the application's image ID in AppImage to prevent domain's application image from being tampered with.
- Added the curl command example for requesting the kuscia API.
- polish the agent runtime docs.
Changed
- Changed some kuscia-crds (KusciaJob,KusciaTask,KusciaDeployment)from cluster to namespace (cross-domain).
Fixed
- Correct some inaccurate descriptions in the document.
v0.6.0.dev240115
Added
- Add network error troubleshooting document.
- Add steps for pre creating data tables in the process of deploying kusica on K8s.
Changed
- The token from lite to master supports rotation.
Fixed
- When deploying using deploy.sh, no kuscia API client certificate was generated.
v0.5.0b0
Added
- Support deploying kuscia on K8s.
- Support running algorithm images based on runp and runk modes.
- Support configuring Path prefix in domain public URL addresses.
Changed
- Optimize deployment configuration and add configuration documentation.
- Optimize error information of task and error logs of kuscia.
Fixed
- When there is a duplicate node error, the node will not exit but will try again.
- Change ClusterDomainRoute status to be unready when dest domain is unreachable.