gabotorresruiz commented on PR #44009:
URL: https://github.com/apache/superset/pull/44009#issuecomment-5688718237

   Delta-verified the three commits since my last pass, empirically rather than 
by reading:
   
   - Re-ran my injected-delay probe at `e0aafeae3a`: 200ms added to the 
before-flush capture and 150ms after the finalize timer start now both surface, 
as a 210.6ms `capture_initial_states.latency` sample and a 156.7ms 
`finalize.latency` sample under their own stages. On the earlier head the 200ms 
was invisible (0.0ms), so the two-stage split genuinely closes the kill-switch 
blind spot.
   - One entity over 4 flushes: 1 pre-state read, 1 sample. An all-failed batch 
still emits its sample. Three entities with the middle capture raising: all 
attempted, 1 and 3 retained, and the commit persisted the user's rows.
   - A stats backend whose `timing()` raises, hit 3 times: 3 single-line 
warnings, zero tracebacks.
   - Every new test fails on its parent commit (3 at `0e767ded`, 2 at 
`53dcb5b8`, 3 at `d5817838`), so they pin the fixes, not just the invariants. 
Full suite at head: 167 passed.
   
   On the red CI: all eight failing legs die in `flask_migrate` with `Multiple 
head revisions`, and this PR touches no migrations. Master's heads were broken 
until #44288 merged at 20:50 UTC, after your 17:14 UTC run started. A rebase or 
rerun should clear them.
   
   Holding my approval only for @aminghadersohi's three open comments at this 
head (I independently reproduced the `_uncaptured_identity` one, see that 
thread) and a green rerun. This has become a really solid piece of 
instrumentation, thanks for the thorough turnaround on every round, Mike.


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to