Flow state & Reinstall Mechanism on Next

[brunovg]

The current situation:

1. Frontend installs the flow in the backend.
2. Frontend gets report back, and stores this state in MongoDB
3. The report service stores the flow state every minute to Elastic DB.

Pages:

• Flow Manager gets the data from Elastic, with a fall back of MongoDB (1 minute old data)
• Flow Manager details (stats) get data directly from the backend (live data)
• Flow Designer gets data from mongodb (expected flow state). Not sure about this one.

The new situation:

In the new situation we want to remove the Rake script, as well the flow state in mongodb. New situation:

To achieve this we need:

• Remove the Rake script.
• Add environment variable ASSIMBLY_FLOWCACHE to the backend test/production containers
• Set all flows to stopped status
• Remove the flow state from MongoDB
• Remove fall back to get the flow state from MongoDB for the flow manager
• Enhance data in the DIL format so that all needed data is in the backend **

Pages:

• Flow Manager gets the data from Elastic (1 minute old data)
• Flow Manager details (stats) get data directly from the backend (live data)
• Flow Designer gets data directly from the backend.

** It maybe needed to extend Assimbly API. If I know the new fields in the dil format. The flow section in the DIL format can have the following fields:

id
name
type
version
steps
properties

You can add any field in the properties section, but maybe installed_by and installed_at can stay purely on the frontend.

[Hooghof]

Ideas:

• Store the last stop timestamp, so that you know by filtering which flows you can restart
• Use stopping all flows in the update process