Skip to content

version-3.6.1.md

Latest commit

 

History

History
36 lines (30 loc) · 1.78 KB

version-3.6.1.md

File metadata and controls

36 lines (30 loc) · 1.78 KB

Version 3.6.1

Bugs Fixed

  • When a TrainingPortal spec lists the same Workshop resource multiple times with different alias values (e.g., to offer the same workshop with different environment variable overrides), the training portal correctly identifies each WorkshopEnvironment by its alias internally. However, the alias is never persisted on the WorkshopEnvironment Kubernetes resource - only the original Workshop resource name is stored in spec.workshop.name. This means the lookup service lacked the information which identified the workshop was to be requested via its alias and such workshops could not be requested as it always tried to use the name on the Workshop. An extra label is now added to WorkshopEnvironment listing the name of the workshop to be used when it is requested, ie., the alias if defined or otherwise the name of the Workshop. This is now used by the lookup service to properly expose the workshop allowing to be requested.

  • When a ingress CA was being supplied this was not being injected into the training portal, so if analytics webhook URL tried to report events to a service in the same cluster where CA was required, it would fail as it would not trust service certificate.

  • When analytics were being reported by the training portal and the service receiving the event accepted the connection but returned a non 200 response, it would be silently ignored and nothing was logged. If a non 200 response is returned details will now be logged.

  • Fixed ACME HTTP-01 solver image not being properly set in the cert-manager Deployment, which caused HTTP-01 challenge resolution to fail for Certificate Issuers.

  • The page_number property in analytics events for pages was not being passed through.