morningman commented on PR #68101:
URL: https://github.com/apache/doris/pull/68101#issuecomment-5708739244
<!-- doris-repo-review:v1:begin -->
### Local pipeline review — ✅ PASS
```yaml
schema: doris-repo-review/v1
status: PASS
pr: apache/doris#68101
commit: 0437c740040e233db395a6780c5321f33f391730
base: dea1b99e528f57b96d9b3bd9fa68f4dc3e726d32
reviewed_at: 2026-09-17T12:47+08:00
reviewer: morningman
model: claude-opus-5
effort: max
findings: {blocker: 0, major: 0, minor: 1, nit: 11}
rounds: 1
converged: true
```
**Notes for maintainers**
- Second review: the `ed7692b` Major (default sub-quota `-1` lets
never-closed Flight sessions fill `qe_max_connection`
and refuse MySQL logins; token cache 1024; sub-quota refusal unreachable
while tokens <= sub-quota) is resolved by
declaration: the `arrow_flight_max_connections` description and the
release note now state the consequence, the
remedy (set the sub-quota explicitly on an FE serving both protocols), the
token-cache sizing, the eviction symptom
and the `-1` downgrade note; an explicit value above `qe_max_connection`
is capped with a startup WARN. The default
itself stays `-1` by the maintainer's decision (plan item D-3) — a
declared design choice this receipt takes no
position on.
- Residual Minor (wording): the new "a Flight session ends only with
CloseSession, KILL or wait_timeout" omits
bearer-token expiry (`arrow_flight_token_alive_time_second`, ends an
active session 24 h after the handshake) and
the per-user token LRU (`max_user_connections / 2`), and
`arrow_flight_token_cache_size`'s "what bounds their
sessions in practice" points at the 4096 knob while the binding number is
the 1024 sub-quota.
- Verified from the pinned jars, read-only: `flight-sql-jdbc-driver` 19.0.0
sends `CloseSession` from `close()` only
when `catalog` is set (the release-note sentence); arrow-memory-core
19.0.0 `BaseAllocator.close()` is idempotent
and reports outstanding bytes, so the new release-session test is
falsifiable. No build, unit test or cluster run by
the reviewer; the author's FE UT and `arrow_flight_sql_p0` results are
taken from the handoff.
- Eleven Nits remain (test gates for the two-argument index removal, the
MySQL-unregister/Flight-counter invariant,
the `(current: N)` fixtures and the channel-close `catch`; suite hygiene;
two wording over-statements; the MySQL
client refused by a Flight-full pool gets no hint; the doc PR to
doris-website is still to be opened).
<sub>Reviewed locally with the `doris-repo-review` pipeline. Repository
policy may accept this receipt for the matching commit; it is not a human
Apache approval.</sub>
<!-- doris-repo-review:v1:end -->
--
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]
