TimoutExceptions during commitSync and commited

Hi,

We have a Consumer that occasionally catches a TimeoutException when
trying to commit an offset after polling. Since it's a
ReatriableException the Consumer tries to roll back and read from the
last committed offset. However when trying to fetch the last committed
offset with committed(), it throws another TimeoutException.

[Logs from committed()]:

org.apache.kafka.common.errors.TimeoutException: Timeout of 60000ms
expired before the last committed offsett for partition <partition-name>
could be determined.

[Logs from commitSync()]:

Offset commit failed on partition <partition-name> at offset <number>:
The coordinator is loading and hence can't process requests.

Offset commit failed on partition <partition-name> at offset <number>:
The coordinator is loading and hence can't process requests.

Offset commit failed on partition <partition-name> at offset <number>:
The coordinator is loading and hence can't process requests.

...

I've tried changing default.api.timeout.ms to two minutes with no luck,
still getting the same behavior after a total of four minutes.

1. Does this indicate that the Broker reloading its cache?
2. What could cause the Coordinator to be in a loading state after a poll?
3. Is there anything  in our Consumer that could cause this behavior,
i.e polling intervals?
4. Is there anything in our Broker that could cause this behavior, i.e
re-balancing?
5. If it is reloading it's cache, shouldn't this be avoided by the
heartbeats sent from the Consumer?

Kind regards,

Sergio