I am observing the same with our .10.2 cluster where consumers are hanged for those partitions where the current offset's data gets deleted due to retention. It looks like a bug to me.
Thanks! On Jul 6, 2017 9:23 AM, "Ghosh, Achintya (Contractor)" < achintya_gh...@comcast.com> wrote: > Hi, > > If we have some high response in backend( let's say database high > response), the few of the partitions messages are piled up. Once the > backend gets normal still those partitions are not processing any message > and it creates a huge lag. > > Any idea why it happened? We thought once the backend gets back to normal > those partitions will be processing but its not happening. Somehow some > threads are stuck. > > Thanks > >