Github user jianbzhou commented on the issue:
https://github.com/apache/storm/pull/1131
@hmcl, just fyi, we found one new issue:
when rebalance happened, in method
onPartitionsAssigned->initialize(Collection<TopicPartition> partitions)
we can see below two lines:
**final OffsetAndMetadata committedOffset = kafkaConsumer.committed(tp);
final long fetchOffset = doSeek(tp, committedOffset);**
If the committedOffset is out of range(say kafka log file removed), the
when the poll() method is called, the offset will reset as the property
auto.offset.reset. This will cause the newly polled message has bigger offset,
so there is a break between committed offset and the acked offset, no
continuous offset will be found.
We will apply a quick fix for this.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
