Currently only the Docker run parameters cpus and memory (from Configuration) are converted into parameters for a JobStep/Kubernetes job. Any other parameters in either Processor or Configuration are silently ignored.
As for the cpus parameter: This information shall be taken as the sum over ...getConfiguredProcessor().getProcessor().getTasks() of Task::getNumberOfCpus() (which may be null).
To be discussed whether:
The current data model is to be kept, which would mean that more/other parameters would be configurable, and that a logic would have to be defined to translate Docker run parameters into Kubernetes job parameters (which is not straightforward),
The attribute dockerRunParameters is to be dropped from Processor and Configuration, and specialized attributes are introduced, e. g. memoryRequested,
A new model is to be found to configure Kubernetes job parameters directly,
Any other solution not mentioned here shall be adopted.
Currently only the Docker run parameters
cpus
andmemory
(from Configuration) are converted into parameters for a JobStep/Kubernetes job. Any other parameters in either Processor or Configuration are silently ignored.As for the
cpus
parameter: This information shall be taken as the sum over...getConfiguredProcessor().getProcessor().getTasks()
ofTask::getNumberOfCpus()
(which may benull
).To be discussed whether:
dockerRunParameters
is to be dropped from Processor and Configuration, and specialized attributes are introduced, e. g.memoryRequested
,