Add Support for Ingress¶
We added the
INGRESS configuration parameter to the
[service] section of a stage's
config.ini file. This is a mandatory configuration parameter that must be set to either
False. If an NGINX ingress controller has been installed in the cluster and if
INGRESS=True, then a path from the external IP of the ingress controller, to the service, will be created automatically when the service is created. This enables services to receive requests from sources external to the cluster, via URLs of the form
Enable Remote Workflow Executions on an ad hoc Basis¶
Workflows can now be executed via remote workflow-controller jobs, on an ad hoc basis, using the
bodywork deployment create command. Prior to this release, the only way to trigger a workflow that is managed as a job on a cluster, was via the
bodywork cronjob create command - ad hoc workflows could could only be executed by having the workflow-controller run locally, using the
bodywork workflow command.
Other Improvements and Bug Fixes¶
- We improved exception handling around internal git commands, so that users now get informative logs when git encounters problems cloning project repositories.
- We changed the working directory in which stages are run, to the directory containing the executable Python module for the stage. This makes it much easier to reference other files (e.g. trained models and other such artefacts), that may have be included with the other mandatory stage files.