We're seeing an issue with some __consumer_offsets partitions that host
consumer groups being used in EOS-style read-process-write cycles. Although
the brokers' log cleaners are running, we are finding more and more segment
files for these partitions, for example:
```
$ ls -lh data/__consumer_offs
Weichu,
while I don't have a solution we are seeing the same thing in our world. I
put out a query to the mailing list a week or two ago (no responses
unfortunately):
https://lists.apache.org/thread.html/04273f5cfe4f6c6ed9ab370399f208a5cd780576880650aae839de25@%3Cusers.kafka.apache.org%3E
.
We're
Claudia,
This sounds like an issue I and other users have reported recently:
https://issues.apache.org/jira/browse/KAFKA-8335. Looks like a fix was
merged yesterday.
Michael
On Tue, May 14, 2019 at 4:44 AM Claudia Wegmann wrote:
> Dear kafka users,
>
> I run a kafka cluster (version 2.1.1) wit
