As part of the work to make changefeeds handle quotas more gracefully (plus the work we did for admission control), changefeeds can quietly end up in a throttled state for many reasons. We should indicate when changefeeds are in a throttled state in the db console:
We track how much time we spend paused due to throttling because of kafka quotas, we should surface this in a graph
We also know per node/ per processor what byte rate are we targeting, we should surface that in a graph
As part of the work to make changefeeds handle quotas more gracefully (plus the work we did for admission control), changefeeds can quietly end up in a throttled state for many reasons. We should indicate when changefeeds are in a throttled state in the db console:
See https://github.com/cockroachdb/cockroach/pull/104446 Jira issue: CRDB-28927