Hi everyone,
I would like to open a discussion on FLINK-39566 [1] / PR #28060 [2], which
adds
a checkpoint duration timeline ("Gantt") view to the Checkpoints tab in the
Web
UI. It came up during review that this is worth wider input before it goes
further.
Today, diagnosing a slow checkpoint means scanning the per-subtask table
row by
row. The new "Timeline" tab shows one stacked bar per subtask across the
four
phases (start_delay, alignment, sync, async), sorted by duration. A
screenshot is
in the PR.
It is frontend only: no backend, REST or public API changes, no new
dependencies,
and no new data -- it renders the same fields the subtask table already
shows.
Any thoughts would be welcome.
[1] https://issues.apache.org/jira/browse/FLINK-39566
[2] https://github.com/apache/flink/pull/28060
