Consider a Unified Approach For Pausing / Resuming Behavior

[tomazfernandes]

Hi,

I see we have some issues that might be convergent in some points:

• Consider An Option To Pause the Container Immediately #2280
• Add BackOffHandler to FailedRecordTracker #2262
• Revise RetryTopicConfigurationSupport TaskExecutor Strategy #2239

They all refer to pausing / resuming containers / partitions, while maybe keeping records in memory so we don't have to fetch again later. So I thought maybe we could come up with a unified approach for these?

I haven't really given much thought on this yet, but a few things come to mind. I'm guessing maybe @garyrussell already has something in mind in this direction?

So, I just wanted to bring this up - if you already have something in mind for this we can close this discussion, or I can also give it some further thought If that might help.

Thanks!

[tomazfernandes]

If appropriate, we could also take advantage of this to review the KafkaConsumerBackOffManager design / implementation.

[tomazfernandes]

Solved by #2328

Thanks!