featzhang created FLINK-39076:
---------------------------------
Summary: [web ui] Display slot sharing group in operator nodes in
Flink Web UI
Key: FLINK-39076
URL: https://issues.apache.org/jira/browse/FLINK-39076
Project: Flink
Issue Type: New Feature
Components: Runtime / Web Frontend
Reporter: featzhang
**Description:**
Currently, the Flink Web UI displays various information about operators in the
job DAG (Directed Acyclic Graph), including parallelism, back pressure status,
and metrics. However, it does not show the slot sharing group ID that each
operator belongs to.
The slot sharing group ID is an important piece of information for users to
understand how operators are sharing slots in the TaskManager. This is
particularly useful for:
- Understanding resource allocation and utilization
- Debugging slot-related issues
- Optimizing job parallelism and resource configuration
- Understanding the logical grouping of operators
**Proposed Change:**
Display the slot sharing group ID in the operator node information within the
Flink Web UI. The display should:
1. Show the slot sharing group ID alongside other operator information
(parallelism, back pressure, etc.)
2. Only display when a slot sharing group ID is available (i.e., when the
operator is part of a slot sharing group)
3. Be visible in both the SVG node label and the detailed node popup
**Current Behavior:**
The slot sharing group ID is not displayed in the Web UI, even though it is
available from the backend REST API (vertices contain `slotSharingGroupId`
field).
**Desired Behavior:**
Users should be able to see the slot sharing group ID for each operator in the
job DAG, making it easier to understand slot sharing relationships and resource
allocation.
**Acceptance Criteria:**
1. Slot sharing group ID is displayed in the operator node SVG label
2. Slot sharing group ID is displayed in the operator node detail popup
3. Display is conditional - only shown when slotSharingGroupId is available
4. Display follows the existing UI style and formatting conventions
5. The feature works correctly for both streaming and batch jobs
6. No performance degradation
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
