This is an automated email from the ASF dual-hosted git repository.
mimaison pushed a commit to branch trunk
in repository https://gitbox.apache.org/repos/asf/kafka.git
The following commit(s) were added to refs/heads/trunk by this push:
new 7f9f34a7111 KAFKA-19649: Document ListOffsets API v11 and follower
fetch from tiered of… (#22038)
7f9f34a7111 is described below
commit 7f9f34a7111347803a8a2facf3201845e46bc104
Author: Abhijeet Kumar <[email protected]>
AuthorDate: Wed Apr 15 00:29:39 2026 +0530
KAFKA-19649: Document ListOffsets API v11 and follower fetch from tiered
of… (#22038)
Document features introduced via KIP-1023.
Reviewers: Mickael Maison <[email protected]>
---
docs/getting-started/upgrade.md | 2 ++
1 file changed, 2 insertions(+)
diff --git a/docs/getting-started/upgrade.md b/docs/getting-started/upgrade.md
index 15ee8b65dae..fb5ad37d83e 100644
--- a/docs/getting-started/upgrade.md
+++ b/docs/getting-started/upgrade.md
@@ -50,6 +50,8 @@ type: docs
* New group configs have been introduced: `share.delivery.count.limit`,
`share.partition.max.record.locks` and `share.renew.acknowledge.enable`, along
with equivalent broker configs for specifying minimum and maximum values. In
addition, the validation of group configs has been improved. For further
details, please refer to
[KIP-1240](https://cwiki.apache.org/confluence/x/tIHMFw).
* A new `group.coordinator.background.threads` config has been added to
control the size of the group coordinator's thread pool used to updating
regular expression subscriptions. Previously, the thread pool only had a single
thread. Now the thread pool has two threads by default and its size is
configurable. For further details, please refer to
[KIP-1263](https://cwiki.apache.org/confluence/x/DIE8G).
* New `group.consumer.assignment.interval.ms`,
`group.share.assignment.interval.ms` and `group.streams.assignment.interval.ms`
configs have been added to set the interval between assignment updates for
consumer, share and streams groups, along with broker configs for specifying
minimum and maximum values and group configs. These default to an interval of 1
second and previously had an effective value of 0. For further details, please
refer to [KIP-1263](https://cwiki.apache.org/conflue [...]
+ * A new dynamic broker configuration
`follower.fetch.last.tiered.offset.enable` (default: `false`) has been added.
When enabled on a cluster with tiered storage, a newly added follower replica
that has no local data will skip directly to the earliest pending upload offset
on the leader, avoiding re-fetching data that is already stored in remote
storage. This reduces bootstrap time significantly for large tiered-storage
topics. For further details, please refer to [KIP-1023](https://cwi [...]
+ * The `ListOffsets` API has been extended to version 11, adding support for
the `EARLIEST_PENDING_UPLOAD_TIMESTAMP` (-6) timestamp type. This allows
clients to query the earliest offset on the leader that has not yet been
uploaded to tiered storage. For further details, please refer to
[KIP-1023](https://cwiki.apache.org/confluence/x/8op3EQ).
## Upgrading to 4.2.0
