SEZ9 commented on PR #11382:
URL: https://github.com/apache/seatunnel/pull/11382#issuecomment-5651006594

   Thanks for the detailed walkthrough of the wait/notify shape on F1 against 
`1ee22aced026`. The description (both methods on the same monitor, the `while` 
loop around `wait(remainingMillis)`, and a single up-front `deadline`) speaks 
to exactly the concerns behind F1. I'll keep F1 open until I've confirmed that 
shape in the diff itself rather than resolving it from the description; if you 
can point at the relevant hunk in `SchemaChangeDrainGuard`, that will speed 
things up.
   
   The comment above appears to be cut off at F2, and the earlier per-finding 
answers it refers to aren't visible to me here, so rather than guess, could you 
restate the current status of each of the following against `1ee22aced026` in 
one place?
   
   - **F2 (HIGH)** — re-delivery of the typed completion notification after 
failover: the intended recovery path when restoring from a checkpoint taken 
before the schema change, or whether a fix is still pending.
   - **F3 / F8** — `CheckpointFinishedOperation` wire-format change: version 
guard, and null `checkpointType` / unknown type name handling on 
(de)serialization.
   - **F4** — drain-ready state not being part of the sink checkpoint snapshot.
   - **F5** — whether any coordinator-side call site sends a typed abort, so 
the guard's typed-abort cleanup is reachable.
   - **F6** — single-slot tracking under overlapping schema changes and the 
type-based abort reset.
   - **F7** — tests beyond the `SchemaChangeDrainGuard` unit tests, covering 
the wiring, `CheckpointFinishedOperation` serialization, and a 
schema-evolution-with-failover scenario.
   
   Once I have that summary I can verify the addressed ones against the diff 
and we can focus on what's genuinely left, with F2 being the main blocker as it 
stands.
   
   <!-- streview-comment:1018 -->


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to