Description
I think that "enqueuing" temporarily gets kind-of-stuck, especially visible during the same experiment (deleting two similar-sized large buckets at once), for one bucket state is switched to "monitoring" very fast, but for another it gets stuck at "enqueuing" for hour or so, even though the queue growth stops for both buckets at about the same time. Only when both buckets are nearly deleted on storage nodes, "enqueuing" changes to "monitoring" for second bucket as well. Both buckets switch to "finished" state at around the same time, it's only "enqueuing->monitoring" switch which seem to happen much later than you'd think it should. But I cannot say if this really is a bug or just behaves this way for some reason, either way, it doesn't lead to consistency problems.
Activity