core/oc-schedulerd
5
0
Fork 0
Code Issues 6 Pull Requests Packages Projects Releases Wiki Activity

updated docs on yaml generation

Browse Source
This commit is contained in:
pb 2024-05-16 14:18:43 +02:00
parent 73fd329551
commit 4074b7d88c
3 changed files with 49 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
docs/argo_fomatting_rules.md
View File

@ -1,6 +1,5 @@
This document aims at centralizing all the informations about the yaml file that will be submitted to argo
- names can only contain alphanumeric characters or '-'
- Lists can either be flow `[value1, value2,...]` or

4
docs/todo.md
View File

@ -2,4 +2,6 @@
- currently arguments are stored in the computing object as a list of string
- argo_builder.getComputingCommands() takes the command as a parameter to treat `sh -c`'s argument differently, but this is not generic enough
- refactor on oc-catalog : we could store it as a string and parse it differently depending on the command
[ ] Are dependencies handled well :
- Branch/dependencies seem to be generated inconsistently, producing different branches with the same graph

65
docs/uml/generate_argo_struct.puml
View File

@ -25,34 +25,59 @@ note top : a node is a component in a link object,\n represented by its ID
argo -> argo
note bottom : We want to prepare the list of branches by removing\n nodes that appears several times\ne.g : two branches :\n1 - 2 2 - 3 3 - 4\n1 - 2 2 - 3 3 - 5\nWe will iterate over each branch to create the component's YAML\n and don't want to create components 1 2 et 3 twice or more
loop branch in branches
group if Source first link is not in list node written
loop while link Source is shared
argo -> argo : create an object representing the component
argo -> argo : add component to list of created nodes
loop computing in graph.computings
argo -> argo : create new container object
argo -> argoGraph : retrieve image name
argo -> argoGraph : add command from the computing component
argo -> argoGraph : add arguments from the computing component
argo -> argoGraph : retrieve environment variables' name from the computing component
argo -> argo : create a new list of Parameter
loop env variable in retrieved env variables
argo -> argo : add a Parameter object to the list with only the name attribute
end
argo -> argo : create a Template object from the name of the component and the Container object
argo -> argo : add attached volume (currently we share the same mounted volume)
argo -> argo : append the new Template
end
argo -> argo : create new Dag object
loop computing in graph.computings
argo -> argo : create a new Task
argo -> argoGraph : add a unique name from the component name and ID in the graph
argo -> argoGraph : add the name of the associated Template
argo -> argoGraph : add the list of environment variables from the component
loop branch in branches
group if component is in the branch
argo -> argo : add the previous component name to the list of dependencies
end
end
argo -> argo : add the task/step to the list of Tasks
end
loop branch in branches
loop link in branch
group if link.Destination not in list of nodes already written
argo -> argoGraph : get component
argoGraph --> argo : component
argo -> argo : generate YAML depending on type (attributes)
argo -> argo : add ID to list of node already written
end
end
argo -> argo : add new Template named DAG to the Templates
loop storage in graph.storages
argo -> argo : create a new VolumeClaimTemplate object
argo -> argoGraph : add name from the storage component
argo -> argoGraph : add access mode from the storage component
argo -> argoGraph : add storage capacity from the storage component
end
loop for each struct created
argo -> argo : Marshal into yaml
argo -> argo : write the marshalled string to a file
loop data in graph.data
argo -> argo :
end
argo -> argo : marshal the Workflow to yaml
argo -> argo : write the marshaled object to a .yml file
@enduml