Document the value of shutdown-after-unsuccessful

[ignasi35]

The value of shutdown-after-unsuccessful... is set to 60s but that value should be aligned with readiness and liveness settings in the YAML deployment files

https://github.com/lagom/shopping-cart-scala/blob/e30b41723922ed635a8c072ea138262147a69cea/shopping-cart-impl/src/main/resources/prod-application.conf#L27-L29

This settings should include extra documentation wrt it's value and how it interacts with other settings on the sample app.

[TimMoore]

Do these settings actually interact? The liveness check will pass even if the cluster hasn't formed, and the readiness check won't result in a shutdown. What did you have in mind here, @ignasi35?

[ignasi35]

What did you have in mind here, @ignasi35?

interaction isthe wrong word. My point was that there are 2 settings that may result into a node/pod being removed before it has the chance to join the cluster:

• shutdown-after-unsuccessful
• liveness and readiness use failureThreshold

Operators only familiar with the YAML settings may face a situation were they use very generous retrials and threasholds but shutdown-after-unsuccessful may still be too strict. IIUC, only making both settings ample enough, slow clusters will be able to form.

[TimMoore]

Got it. But maybe the best place to document it is in the Lagom documentation rather than the example application?

[ignasi35]

Got it. But maybe the best place to document it is in the Lagom documentation rather than the example application?

Sure! I generally like comments cross-referencing from settings to code (or settings to other settings) when there are interactions between things. But a deeper explanation or just a simple introduction to the problem and the related settings/knobs in the documentation will also be necessary/useful.

[ignasi35]

raised lagom/lagom#2337