8.11 Release notes
Release notes for new features included in the 8.11 minor release, including alpha feature releases.
8.11.0-alpha1
| Release date | Changelog(s) |
|---|---|
| 10 November 2026 |
APIs & tools
Read the engine clock and add time without pinning it
With two new Orchestration Cluster REST API endpoints, you can read the engine clock and move it without stopping it.
| Endpoint | What it does | Required permission | Java client |
|---|---|---|---|
GET /v2/clock | Returns the current time of the engine clock. If the clock is pinned, it returns the pinned time. | READ on SYSTEM | newClockGetRequest() |
POST /v2/clock/add | Moves the clock forward or backward by an ISO 8601 duration, such as PT1H or -PT1H, and returns the new time. | UPDATE on SYSTEM | newAddClockCommand().offset(Duration) |
Both endpoints return the time as an ISO 8601 date-time, for example { "timestamp": "2025-01-01T12:00:00.000Z" }.
Unlike pinning, adding time keeps the clock running from the new time, so job timeouts and retry backoffs still run out in tests. If the clock is pinned, it stays pinned and moves to the pinned time plus the offset. To remove the added time, reset the clock with POST /v2/clock/reset.
Orchestration Cluster
Evenly distributed record indices in Elasticsearch and OpenSearch
The Elasticsearch and OpenSearch exporters now spread Zeebe records more evenly across index shards. The zeebe-record-job and zeebe-record-process-instance indices default to 5 shards, up from 3. The zeebe-record-user-task index defaults to 1 shard, down from 3. With a three-partition cluster, the records of each partition land on a separate shard instead of two partitions sharing one shard.
Only newly created dated indices use the new defaults. An existing index keeps the shard count it was created with. If you set number-of-shards, it still overrides the template defaults for all indices.