Our kafka broker logs show episodes where several partitions have
"Shrinking ISR ..." messages followed, usually less than 2 seconds
later, by corresponding "Expanding ISR ..." messages that restore the
original set of brokers for all the partitions. Does anyone have any
suggestions on how to inve
es, gc and memory usage on Kafka brokers?
> When pause times are high kafka broker disconnects from zk. goes out of
> cluster and rejoins later makes shrink and expand on isr.
>
>
> > On May 2, 2019, at 5:39 PM, Steven Taschuk
> wrote:
> >
> > Our kafka broke
