[
https://issues.apache.org/jira/browse/FLINK-39566?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Purushottam Sinha updated FLINK-39566:
--------------------------------------
Description:
Problem
Today the Checkpoints tab is a wall of numbers. To find why a checkpoint was
slow, you scroll through subtask rows and eyeball columns. It's hard to spot
stragglers or compare checkpoints over time.
Proposal
Add a Gantt-style view to the Checkpoints tab:
- Recent strip — last 60(or max possible if less) checkpoints as colored bars
(width = duration, color = status: completed / savepoint / in-progress /
failed). Auto-refreshes.
- Per-checkpoint Gantt — one row per subtask, grouped by operator, with stacked
segments for the four checkpoint phases. Sorted by total duration so stragglers
float to the top. Outliers and aborts highlighted.
Interactions
- Click a bar in the strip → pins the Gantt to that checkpoint
- "Follow newest" → resume live tracking
- "Export PNG" → snapshot for incident reports
Scope
Frontend-only. Uses existing REST endpoints, renders via @antv/g2. No backend
changes.
was:
Problem
Today the Checkpoints tab is a wall of numbers. To find why a checkpoint was
slow, you scroll through subtask rows and eyeball columns. It's hard to spot
stragglers or compare checkpoints over time.
Proposal
Add a Gantt-style view to the Checkpoints tab:
- Recent strip — last 60 checkpoints as colored bars (width = duration, color =
status: completed / savepoint / in-progress / failed). Auto-refreshes.
- Per-checkpoint Gantt — one row per subtask, grouped by operator, with stacked
segments for the four checkpoint phases. Sorted by total duration so stragglers
float to the top. Outliers and aborts
highlighted.
Interactions
- Click a bar in the strip → pins the Gantt to that checkpoint
- "Follow newest" → resume live tracking
- "Export PNG" → snapshot for incident reports
Scope
Frontend-only. Uses existing REST endpoints, renders via @antv/g2. No backend
changes.
> [runtime-web] Add checkpoint duration Gantt view
> ------------------------------------------------
>
> Key: FLINK-39566
> URL: https://issues.apache.org/jira/browse/FLINK-39566
> Project: Flink
> Issue Type: New Feature
> Components: Runtime / Web Frontend
> Reporter: Purushottam Sinha
> Priority: Minor
> Labels: pull-request-available
>
> Problem
> Today the Checkpoints tab is a wall of numbers. To find why a checkpoint was
> slow, you scroll through subtask rows and eyeball columns. It's hard to spot
> stragglers or compare checkpoints over time.
>
> Proposal
>
>
> Add a Gantt-style view to the Checkpoints tab:
> - Recent strip — last 60(or max possible if less) checkpoints as colored bars
> (width = duration, color = status: completed / savepoint / in-progress /
> failed). Auto-refreshes.
>
> - Per-checkpoint Gantt — one row per subtask, grouped by operator, with
> stacked segments for the four checkpoint phases. Sorted by total duration so
> stragglers float to the top. Outliers and aborts highlighted.
>
>
>
>
> Interactions
>
>
> - Click a bar in the strip → pins the Gantt to that checkpoint
> - "Follow newest" → resume live tracking
> - "Export PNG" → snapshot for incident reports
>
>
>
> Scope
> Frontend-only. Uses existing REST endpoints, renders via @antv/g2. No backend
> changes.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
