Retries; exponential backoff & max backoff duration about specification

[tsurdilo]

@jorgenj
Re "exponential backoff" - I think that is a valid use case. Could you provide an example json/yaml def the way you would like to see it and we can discuss details? Would just make it easier to see it as an example that can be used also for examples and docs.

Re: "max backoff duration" - again would be possible to see an example of how you would like this to be? Current impl can handle similar cases with the event-based switch state and the callback states for example, where the amount of time waiting for event can be expressed via a timeout value.

[jorgenj]

For exponential backoff, I'm partial to the idea of renaming the current 'multiplier' field which accepts a duration to 'increment' instead, and adding a new 'multiplicatier' field that accepts a numeric value. For example:

{ "expression": ...,
"interval": "PT2S",
"increment": "PT2S",
"multiplier": 1.5
}

This example would result in the following series of backoff durations between retries:

• 2s
• 6s ((2s + 2s) * 1.5)
• 12s ((6s + 2s) * 1.5)
• ...

For max backoff, something like the following (using multiplier redefinition from above):

{ "expression": ...,
"interval": "PT2S",
"maxDuration": "PT15S",
"multiplier": 2
}

This example would result in the following series of backoff durations between retries:

• 2s ( min(2s, 15s) )
• 4s ( min(2s * 2, 15s) )
• 8s ( min(4s * 2, 15s) )
• 15s ( min(8s * 2, 15s) )
• 15s ( min(15s * 2, 15s) )
• ...

[tsurdilo]

depends on pr #175

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.