Stavros Kontopoulos

@vagababov, as you probably noticed we (cc @julz @markusthoemmes) had a long discussion on slack (autoscaling channel), so absolutely that is one option that we discussed, but still within the...

@vagababov @julz should we move on with a workqueue approach, have a first version and take it from there?

@markusthoemmes from my POV, my definition of success would be to add configuration capabilities keeping the same performance as in the unbounded case and avoid corner cases as described in...

The way I see this is that the workqueue imposes a rate limiting strategy in order to optimize revision scraping throughput and latency. The current approach from operations perspective implicitly...

@vagababov reading these [benchmarks](https://github.com/panjf2000/ants#-performance-summary) from the `ants` lib shows that most likely unbounded goroutines (independently from scraping) are not the best choice beyond a concurrency number eg. 100K tasks in...

@julz describes an issue that occurs in large envs. So we need to clarify what is the number of pods being scraped when issues occur. Btw in theory if we...

@dprotaso need any help on this?

@dprotaso are you still working on this one?

/remove-lifecycle stale

/remove-lifecycle stale