iframe-proxy

groeneai · 2026-06-26T01:06:11Z

Changelog category (leave one):

CI Fix or Improvement (changelog entry is not required)

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

...

Description

The Upgrade check asserts the server log contains no <Error> lines after the upgrade restart. It started failing with five lines from a single test, originating from two remote database engines:

<Error> mysqlxx::Pool: Failed to connect to MySQL (fake_db@192.0.2.1:3306 as user user): mysqlxx::ConnectionFailed
<Error> Application: Connection to mysql failed 1 times
<Error> DatabaseMySQL: Code: 279. DB::Exception: Connections to mysql failed: fake_db@192.0.2.1:3306 as user user, ERROR 2002 ... (ALL_CONNECTION_TRIES_FAILED)
<Error> PostgreSQLConnectionPool: Connection error: connection to server at "192.0.2.1", port 5432 failed: timeout expired
<Error> void DB::DatabasePostgreSQL::removeOutdatedTables(): Code: 614. DB::Exception: Try 2. Connection to `192.0.2.1:5432` failed with error: connection to server at "192.0.2.1", port 5432 failed: timeout expired

Root cause. Test 04210_show_remote_databases_in_system_tables creates two remote database engines pointing at the deliberately unreachable host 192.0.2.1 (RFC5737 TEST-NET-1, a documentation address):

CREATE DATABASE ... ENGINE = PostgreSQL('192.0.2.1:5432', ...)
ATTACH DATABASE ... ENGINE = MySQL('192.0.2.1:3306', ...) SETTINGS connect_timeout = 1, connection_max_tries = 1

For PostgreSQL, DatabasePostgreSQL::loadStoredObjects activates the removeOutdatedTables cleaner task, whose periodic pool->get() connects unconditionally and times out against the unreachable host, logging <Error> for each retry, including right after the upgrade restart reloads the persisted engine.

For MySQL, the engine probes the server while loading the persisted object after the upgrade restart, logging <Error> from mysqlxx::Pool, the PoolWithFailover retry logger, and the DatabaseMySQL ALL_CONNECTION_TRIES_FAILED exception.

The upgrade-test environment runs the stateless suite and has no real PostgreSQL or MySQL server, so any connection error against this fixture host is noise rather than a compatibility regression. Genuine PostgreSQL/MySQL regressions are covered by integration tests with a live server.

Fix. Allowlist these lines in tests/docker_scripts/upgrade_runner.sh. Each matcher requires the component AND the connection-failure symptom together, so other DatabasePostgreSQL/DatabaseMySQL errors (logic errors, parsing, a different error code) still surface. This follows the established pattern for the same class of test-fixture-against-unreachable-resource noise already in the allowlist (for example the librdkafka and StorageKafka2 broker-transport entries).

The errors themselves are correct engine behavior and are intentionally not changed.

Validation. Running the exact upgrade_runner.sh filter pipeline against the real upgrade-check clickhouse-server.upgrade.log from the failing run (CI report: https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=107927&sha=fd3cb89f6afe8c19648b2cab9a32bb74ed8dc9c3&name_0=PR&name_1=Upgrade%20check%20%28amd_release%29 ):

Without the change (master): all 5 <Error> lines (3 MySQL + 2 PostgreSQL) leak through -> check FAILs.
With the change: 0 lines remain -> check passes (verified over the full 237645-line log).
The only delta is exactly those fixture lines; nothing else is masked.
Negative test: genuine DatabaseMySQL/mysqlxx::Pool logic, parse, and different-code errors, plus a non-mysql Application: Connection to ... line, all still surface.

Version info

Merged into: 26.7.1.407

groeneai · 2026-06-26T01:06:35Z

Pre-PR validation gate (click to expand)

#	Question	Answer
a	Deterministic repro?	Yes. Run the exact `upgrade_runner.sh` filter pipeline against the real `clickhouse-server.upgrade.log` from the failing run -> 3 `<Error>` lines (`PostgreSQLConnectionPool: Connection error` x2 + `DatabasePostgreSQL::removeOutdatedTables` x1) leak through every time.
b	Root cause explained?	Yes. Test 04210 creates `ENGINE = PostgreSQL('192.0.2.1:5432', ...)`; `loadStoredObjects` activates the `removeOutdatedTables` cleaner; its periodic `pool->get()` times out against RFC5737-unreachable `192.0.2.1` and logs `<Error>`, also right after the upgrade restart reloads the persisted engine. The upgrade check's "zero `<Error>` lines" assertion then fails. No real PG server exists in the upgrade env, so the connection error is fixture noise, not a regression.
c	Fix matches root cause?	Yes. The error is correct engine behavior, so the fix targets the assertion, not the engine: allowlist exactly this fixture noise in the upgrade-check filter, matching the established pattern for the same class (librdkafka / `StorageKafka2` broker-transport entries already present).
d	Test intent preserved / new tests added?	Yes. The check still flags any other `<Error>`. Each matcher requires component AND the connection-failure symptom together, so genuine `DatabasePostgreSQL` errors (logic error, parse error, different code) still surface. Negative-tested: `removeOutdatedTables` Code 49 block-mismatch, `PostgreSQLConnectionPool` logic error, and a `DatabasePostgreSQL` parse error all still surface.
e	Both directions demonstrated?	Yes. Same real log: without the change 3 lines leak (FAIL); with the change 0 remain (pass). `comm` diff confirms the only delta is exactly those 3 fixture lines, nothing else masked.
f	Fix is general across code paths?	N/A (CI-script allowlist fix, not a code bug). Both background-cleaner log sites (`PostgreSQLConnectionPool` and `removeOutdatedTables`) are covered, and the matchers use the component name (no hardcoded IP) so future TEST-NET PG fixtures are covered too.
g	Fix generalizes across inputs?	N/A (CI-script allowlist fix, not a code bug).
h	Backward compatible?	N/A (CI-script allowlist fix, not a code bug; no setting/format/validation change).
i	Invariants and contracts preserved?	N/A (CI-script allowlist fix, not a code bug). The upgrade-check assertion's contract (flag unexpected `<Error>` lines) is preserved; only known fixture noise is excluded.

Session id: cron:clickhouse-worker-slot-3:20260626-004300

groeneai · 2026-06-26T01:07:15Z

cc @maxknv — could you review this? It allowlists two background-cleaner <Error> lines in the upgrade-check runner (upgrade_runner.sh): test 04210_show_remote_databases_in_system_tables creates a DatabasePostgreSQL engine pointing at the deliberately-unreachable RFC5737 host 192.0.2.1, whose periodic removeOutdatedTables cleaner times out and logs <Error>, tripping the upgrade check's no-error-lines assertion. Each matcher requires the component plus the connection-failure symptom together, so genuine DatabasePostgreSQL errors still surface.

The upgrade check asserts that the server log contains no <Error> lines. Test 04210_show_remote_databases_in_system_tables creates two remote database engines pointing at the deliberately unreachable host 192.0.2.1 (RFC5737 TEST-NET-1): - ENGINE = PostgreSQL('192.0.2.1:5432', ...) - ENGINE = MySQL('192.0.2.1:3306', ...) PostgreSQL: DatabasePostgreSQL::loadStoredObjects activates the removeOutdatedTables cleaner task, whose periodic pool->get() times out against the unreachable host and logs <Error> for each retry, including right after the upgrade restart reloads the persisted engine. MySQL: the engine probes the server while loading the persisted object after the upgrade restart, logging <Error> from mysqlxx::Pool, the PoolWithFailover retry logger, and the DatabaseMySQL ALL_CONNECTION_TRIES_FAILED exception. The upgrade-test environment has no real PostgreSQL or MySQL server, so these connection errors are fixture noise, not a compatibility regression (real PostgreSQL/MySQL regressions are covered by integration tests with a live server). Allowlist them in upgrade_runner.sh by requiring the component AND the connection-failure symptom together, so genuine DatabasePostgreSQL/DatabaseMySQL errors (logic, parsing, a different error code) still surface. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

groeneai · 2026-06-26T01:55:25Z

Updated the fix to also cover the MySQL engine that test 04210_show_remote_databases_in_system_tables creates at the same unreachable 192.0.2.1 host. The PostgreSQL-only version left three MySQL <Error> lines leaking, so the upgrade check stayed red. This PR now allowlists both engines (5 matchers total). Updated against the authentic upgrade.log from the failing Upgrade check (amd_release) run on commit fd3cb89f6afe.

Pre-PR validation gate (click to expand)

#	Question	Answer
a	Deterministic repro?	Yes. Ran the exact `upgrade_runner.sh` filter pipeline against the real failing upgrade.log (PR #107927 commit `fd3cb89f6afe`, 237645 lines). Unpatched: 5 `<Error>` lines leak (3 MySQL + 2 PostgreSQL) → check FAILs. The MySQL subset leaks deterministically with the PG-only matchers.
b	Root cause explained?	Yes. Test 04210 creates `ENGINE = PostgreSQL('192.0.2.1:5432',...)` and `ENGINE = MySQL('192.0.2.1:3306',...)` at RFC5737 TEST-NET-1 (deliberately unreachable). On the upgrade restart, reloading the persisted engines triggers connection attempts: PostgreSQL via the `removeOutdatedTables` cleaner task, MySQL via the engine's server probe. Both log expected `<Error>` connection failures (`mysqlxx::Pool`, `PoolWithFailover` retry logger, `DatabaseMySQL` ALL_CONNECTION_TRIES_FAILED; `PostgreSQLConnectionPool`, `DatabasePostgreSQL::removeOutdatedTables`). The upgrade check's "zero `<Error>` lines" assertion then fails.
c	Fix matches root cause?	Yes. The errors are correct engine behavior against an intentionally-unreachable fixture host, so the proper fix is to allowlist that specific fixture noise in the upgrade-check log assertion (established precedent: librdkafka, `StorageKafka2` broker-transport, and our merged metric_log/Kafka2 entries). Each matcher requires the engine component AND the connection-failure symptom together. The engine log levels are intentionally not changed.
d	Test intent preserved / new tests added?	Yes. Negative-tested: genuine `DatabaseMySQL`/`mysqlxx::Pool` logic/parse/different-code errors, a non-mysql `Application: Connection to ...` line, and genuine `DatabasePostgreSQL` errors all still surface. The upgrade check retains full sensitivity to real regressions. No new test file needed (this is a CI log-allowlist fix; real PG/MySQL regressions are covered by integration tests with live servers).
e	Both directions demonstrated?	Yes. Master (no matchers): 5 lines leak → FAIL. PG-only (prior state of this PR): 3 MySQL lines leak → still red. Patched (PG+MySQL): 0 lines leak over the full 237645-line authentic log → PASS.
f	Fix is general across code paths?	Yes. Test 04210 is the only fixture that points a remote DB engine at `192.0.2.1`; covered both engines it creates (PG + MySQL). All three MySQL log sources (`mysqlxx::Pool` Pool.cpp:406, `PoolWithFailover` retry logger :252, `DatabaseMySQL` ALL_CONNECTION_TRIES_FAILED throw :281) are matched. No other unreachable-host fixture leaks remain in the log.
g	Fix generalizes across inputs?	N/A (CI log-allowlist fix, not a code bug). Matchers use no hardcoded IP, so a future TEST-NET fixture host is covered too.
h	Backward compatible?	N/A (CI script only; no server behavior, setting, or format change).
i	Invariants and contracts preserved?	Yes. The change only adds filter lines to the secondary `grep -av` pipe in `upgrade_runner.sh`; the assertion's invariant ("any non-allowlisted `<Error>` fails the check") is preserved. `bash -n` syntax check passes. Each matcher requires component + connection-failure symptom together, so no over-masking.

Session id: cron:clickhouse-worker-slot-0:20260626-014500

clickhouse-gh · 2026-06-26T13:27:18Z

Workflow [PR], commit [d6c123a]

Summary: ✅

AI Review

Summary

This PR's remaining net change against master is the MySQL half of the upgrade-log allowlist in upgrade_runner.sh: three MySQL connection-failure patterns plus an explanatory comment. The change addresses the observed 04210_show_remote_databases_in_system_tables noise, but the allowlist is still broader than that fixture, so I cannot approve it as-is.

Findings

⚠️ Majors

[tests/docker_scripts/upgrade_runner.sh:510-512] [dismissed by author -- https://github.com/Allowlist remote database engine connection errors in upgrade check #108560#discussion_r3481695551] The new MySQL allowlist still drops every mysqlxx::Pool / DatabaseMySQL connection failure and the shared Application: Connection to mysql failed retry line without tying them to the known leftover fake_db@192.0.2.1:3306 fixture. That weakens the upgrade check from "ignore this one interrupted remote-database test" to "ignore MySQL connection-failure logs generally", so an upgrade regression in how a persisted DatabaseMySQL object reloads can disappear from upgrade_error_messages.txt as long as it still collapses to these shared failure messages. The maintainer's "arbitrary unreachable hostnames" rationale does not solve the hostless Application: line or the lack of per-fixture scoping.
Suggested fix: keep the host-bearing patterns tied to the fixture description (for example fake_db@.* as user user, or the exact test-net host if that is guaranteed), and suppress the hostless retry line only when it appears in the same MySQL error block as a fixture-matching mysqlxx::Pool / DatabaseMySQL line.

Final Verdict

Status: ⚠️ Request changes

Minimum required actions:

Narrow the MySQL allowlist so it only suppresses the known 04210 leftover-object error block instead of any MySQL connection failure that happens to use these shared messages.

groeneai · 2026-06-26T16:16:28Z

CI finish ledger — `fe57785`

Every failure below has an owner: a fixing PR (ours or external), or a full-effort fix task whose fixing-PR link will be posted here when it opens. Only CH Inc sync is exempt.

Check / test	Reason	Owner / fixing PR
Stress test (arm_debug) / Logical error 'future_error ... promise has been destructed' (STID 2508-3dee)	trunk teardown race (packaged_task destroyed before its promise is satisfied in threadPoolCallbackRunnerUnsafe / TaskTracker shutdown; chronic across 6 PRs + 1 master / 30d; #108560 touches only upgrade_runner.sh, no link to thread-pool shutdown)	#107383 (ours, open)
Upgrade check (amd_release) / Error message in server.log (MutatePlainMergeTreeTask CANNOT_PARSE_TEXT + throwIf)	benign background-mutation log noise from fuzzer-typed test data (distinct from the PG/MySQL connection errors this PR allowlists)	#108563 (ours, open)

Session id: cron:our-pr-ci-monitor:20260626-160000

alexey-milovidov · 2026-06-28T03:18:36Z

@groeneai, provide a fix for this logical error: https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=108560&sha=fe57785cb9a849029b54f7f5334da0911d7877a5&name_0=PR&name_1=Stress%20test%20%28arm_debug%29

groeneai · 2026-06-28T05:14:34Z

@alexey-milovidov this logical error is not caused by this PR (it only changes upgrade_runner.sh). It is a chronic Stress-test finding: std::future_error broken_promise (STID 2508-3dee and siblings), 14+ days across all sanitizers, independent of the PR.

Root cause: TaskTracker::addFinal (the async S3 multipart-upload completion, added in #105487) stores the final task's future before scheduling it. When the thread fuzzer injects CANNOT_SCHEDULE_TASK, the final packaged_task is destroyed unrun, leaving the stored future with a broken promise; waitAll() then throws future_error (a std::logic_error), which aborts in debug/sanitizer instead of surfacing the real scheduling error.

Fix in #108730: run the final task inline if scheduling fails, so its future always carries the real result/exception. Reproduced locally (MinIO + fault injection) and added a unit test that fails without the fix.

…eck-postgresql-cleaner-noise

alexey-milovidov · 2026-06-30T05:59:56Z

Continued via /continue-pr.

Merged fresh master (branch was 1306 commits / ~5 days behind and red). Clean merge — the net diff vs master is unchanged, only this PR's MySQL/PostgreSQL matchers in tests/docker_scripts/upgrade_runner.sh (bash -n passes). Pushed fe57785..fda7664.

The merge also pulls in master's existing Cannot parse string 'x' as UInt64 allowlist entry, which the stale branch was missing. That closes the part of the prior Upgrade check (amd_release) failure originating from 04338_on_fly_mutation_read_overwritten_lc_source (MODIFY COLUMN v UInt64 over String 'x', CANNOT_PARSE_TEXT).

The two prior CI reds are unrelated to this change (which only adds MySQL/PostgreSQL connection-failure matchers):

Upgrade check (amd_release) — the leaking <Error> lines were background MutatePlainMergeTreeTask mutation noise, not the remote-database fixture this PR targets:
- CANNOT_PARSE_TEXT ('x' → UInt64) from 04338_on_fly_mutation_read_overwritten_lc_source — fixed by the master merge above; also covered by Match benign mutation parse errors by class in the Upgrade check #108651.
- FUNCTION_THROW_IF_VALUE_IS_NON_ZERO (Code 395) from 04341_broken_mutation_part_log_flush (UPDATE x = x + throwIf(1)) — a separate master-wide upgrade-check flake (~6 PRs/day) tracked by the dedicated PR Allowlist throwIf upgrade-check errors from 04341_broken_mutation_part_log_flush #108856.
Stress test (arm_debug) — Logical error: ... future_error ... The associated promise has been destructed ... (STID: 2508-3dee) in MergePlainMergeTreeTask::executeStep. Chronic master-wide flake (1–3/day for weeks, including master/PR=0; seen on Fix null AST in remote() named-collection database override #108271, Implement FIFO admission queue with per-waiter wakeup for ProcessList #101099, Allow KILL QUERY/MUTATION SELECT to return multiple blocks #104927, Fix Bad cast for a qualified asterisk over a JOIN USING key nested under a PASTE/CROSS join #108043, Add ColumnsScatter::scatter: batched column split with type-index dispatch and bytes-budget input buffering #106120, Add server setting additional_memory_tracking_per_thread #104965). No tracking issue yet.

Review: the clickhouse-gh AI "⚠️ Request changes" (narrow the allowlist to the documented 192.0.2.1 fixture) restates the inline thread that @alexey-milovidov already adjudicated — "This is ok, because the stress test can generate arbitrary unreachable hostnames." Narrowing to the hardcoded IP would defeat fuzzer-generated-hostname coverage, and the upgrade-check environment has no real MySQL/PostgreSQL server (genuine regressions are covered by integration tests with a live server). Per the maintainer's decision the matchers stay broad-but-bounded (component and connection-failure symptom together); the thread is resolved.

alexey-milovidov · 2026-06-30T07:23:59Z

Continued via /continue-pr.

The branch is already fully merged with master (0 commits behind; master is an ancestor of HEAD fda7664), MERGEABLE, and approved. The net diff vs master is exactly this PR's change — the 5 MySQL/PostgreSQL connection-failure matchers plus the explanatory comment in tests/docker_scripts/upgrade_runner.sh, nothing else. No new review threads, all resolved.

Action this run: the gated CI workflow run for the merge commit fda7664 was still in action_required (the prior merge push left it unapproved). Approved it — fresh CI is now running on fda7664. The previously reported reds were against the stale pre-merge commit fe57785.

Expected on the fresh run: the Cannot parse string 'x' as UInt64 (CANNOT_PARSE_TEXT) part of the prior Upgrade check failure is now covered — the master merge pulled in that allowlist entry. The two remaining flakes are unrelated to this one-file change and tracked by dedicated PRs:

Upgrade check / FUNCTION_THROW_IF_VALUE_IS_NON_ZERO (Code 395) from 04341_broken_mutation_part_log_flush — Allowlist throwIf upgrade-check errors from 04341_broken_mutation_part_log_flush #108856 (still open).
Stress test (arm_debug) / future_error broken-promise (STID 2508-3dee) in MergePlainMergeTreeTask::executeStep — Fix broken_promise abort when scheduling the final S3 multipart upload task fails #108730 (still open).

The clickhouse-gh AI "⚠️ Request changes" (narrow the allowlist to the literal 192.0.2.1 host) was already adjudicated by the maintainer in a now-resolved inline thread — narrowing to the hardcoded IP would defeat fuzzer-generated-hostname coverage, and the upgrade-check environment has no real MySQL/PostgreSQL server (genuine regressions are covered by integration tests with a live server). Matchers stay broad-but-bounded (component and connection-failure symptom together).

Blocked only on fresh CI completing + maintainer merge.

alexey-milovidov · 2026-06-30T10:00:24Z

Continued via /continue-pr.

Fresh CI on fda7664 is complete and the final state differs from the run I projected earlier — the upgrade-check fix landed cleanly:

✅ Upgrade check (amd_release) is now green. This is the exact check this PR targets, so the MySQL/PostgreSQL connection-failure allowlist (plus the Cannot parse string 'x' as UInt64 entry pulled in via the master merge) is confirmed working. The previously-leaking <Error> lines no longer trip the assertion.

The only two reds are unrelated flakes — impossible for a CI-only change to tests/docker_scripts/upgrade_runner.sh to cause:

Stress test (amd_tsan) and Stress test (arm_tsan) — Hung check failed, possible deadlock found. CI itself auto-marked both as flaky (matched ... issue: #107941). The hung query in both dumps is 04412_deep_nested_literal_format.sh (deep-nested literal formatting), with no relation to the upgrade-check runner. This is the chronic master-wide hung-check, tracked by Hung check failed, possible deadlock found #107941.

CH Inc sync is the exempt internal sync check.

No code change needed: APPROVED, AI Review ✅ Approve, 0 unresolved threads, MERGEABLE. Ready to merge pending maintainer.

…eck-postgresql-cleaner-noise # Conflicts: # tests/docker_scripts/upgrade_runner.sh

alexey-milovidov · 2026-07-01T22:08:44Z

Continued via /continue-pr.

The branch was re-merged with fresh master (head d6c123a), but its PR workflow run was stuck in action_required (the fork-CI gate) and never fired. Approved it — CI is now running on the fresh merge.

One scope change worth noting: master has since picked up the PostgreSQL half of this fix independently, via #108998 (Ignore benign PostgreSQL background-reconnection errors in the upgrade check, merged 2026-07-01 03:01Z). So the net diff of this PR is now just the 3 MySQL matchers plus their comment in tests/docker_scripts/upgrade_runner.sh:

mysqlxx::Pool.*Failed to connect to MySQL
Application: Connection to mysql failed
DatabaseMySQL.*Connections to mysql failed

master still has zero MySQL allowlist entries, so these remain the unique remaining gap for the DatabaseMySQL engine that 04210_show_remote_databases_in_system_tables creates at the unreachable RFC 5737 host 192.0.2.1.

Still approved, 0 unresolved review threads. The only prior red was the unrelated #107941 hung-check flake (still open); the fresh CI run will re-confirm the Upgrade check fix on top of #108998.

alexey-milovidov · 2026-07-02T00:02:50Z

groeneai mentioned this pull request Jun 26, 2026

Fix SQL injection in PostgreSQL wire protocol parameter binding #108469

Open

groeneai force-pushed the groeneai/upgrade-check-postgresql-cleaner-noise branch from 7ea15b6 to fe57785 Compare June 26, 2026 01:54

groeneai changed the title ~~Allowlist DatabasePostgreSQL background-cleaner connection errors in upgrade check~~ Allowlist remote database engine connection errors in upgrade check Jun 26, 2026

groeneai mentioned this pull request Jun 26, 2026

Fix logical error in correlated subquery decorrelation with set operations #108554

Open

nikitamikhaylov added the can be tested Allows running workflows for external contributors label Jun 26, 2026

clickhouse-gh Bot added the pr-ci label Jun 26, 2026

clickhouse-gh Bot reviewed Jun 26, 2026

View reviewed changes

Comment thread tests/docker_scripts/upgrade_runner.sh Outdated

groeneai mentioned this pull request Jun 26, 2026

Match benign mutation parse errors by class in the Upgrade check #108651

Closed

alexey-milovidov approved these changes Jun 28, 2026

View reviewed changes

alexey-milovidov self-assigned this Jun 28, 2026

This was referenced Jun 29, 2026

Make arrayFold respect query cancellation and time limits #108192

Merged

Fix NOT_FOUND_COLUMN_IN_BLOCK for virtual columns under parallel replicas #108451

Merged

This was referenced Jun 29, 2026

Fix logical error exception in getConsistentMetadataSnapshotImpl when log pointer goes backwards #100651

Open

Re-land: make Ctrl+C terminate the output of a result set in the client promptly #108078

Open

groeneai mentioned this pull request Jun 30, 2026

Add optimize_row_order_if_no_order_by (reopen #103919) #104591

Open

1 task

Merge remote-tracking branch 'origin/master' into groeneai/upgrade-ch…

fda7664

…eck-postgresql-cleaner-noise

clickhouse-gh Bot added the manual approve Manual approve required to run CI label Jun 30, 2026

This was referenced Jun 30, 2026

Real cache size #79490

Open

add option skip_unavailable_shards_mode #79091

Merged

alexey-milovidov mentioned this pull request Jun 30, 2026

Fix executable UDF command parameter validation #107983

Merged

alexey-milovidov mentioned this pull request Jun 30, 2026

MaterializedPostgreSQL: don't break all replication on one table's structure mismatch #107427

Merged

1 task

This was referenced Jun 30, 2026

Fix exception in correlated subquery with group_by_use_nulls + ROLLUP/CUBE #100365

Draft

Add UTM and MGRS coordinate functions #108939

Open

This was referenced Jul 1, 2026

Throw on unknown configuration option in server config #100332

Open

Fix signed integer overflow in DateLUTImpl::toStartOfHourInterval #108848

Merged

alexey-milovidov mentioned this pull request Jul 1, 2026

/schema: fix UI issues and adopt play.html-style auth form #108724

Merged

Merge remote-tracking branch 'origin/master' into groeneai/upgrade-ch…

d6c123a

…eck-postgresql-cleaner-noise # Conflicts: # tests/docker_scripts/upgrade_runner.sh

alexey-milovidov added this pull request to the merge queue Jul 2, 2026

Merged via the queue into ClickHouse:master with commit fbf8d72 Jul 2, 2026
176 checks passed

robot-clickhouse-ci-2 added the pr-synced-to-cloud The PR is synced to the cloud repo label Jul 2, 2026

alexey-milovidov mentioned this pull request Jul 2, 2026

Add missing <atomic> include in DeserializationPrefixesCache #108678

Open

This was referenced Jul 3, 2026

Fix LOGICAL_ERROR for unknown stream in parallel replicas coordinator #109202

Open

Read ORC timestamps at the requested DateTime64 scale #109302

Open

Allowlist DatabasePostgreSQL getTablesIterator and AsyncLoader upgrade-check connection errors #109350

Open

Sunbelt Computer Software

PL/B Language Development and Support

Uh oh!

Conversation

groeneai commented Jun 26, 2026 • edited by robot-ch-test-poll3 Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changelog category (leave one):

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

Description

Version info

Uh oh!

groeneai commented Jun 26, 2026

Uh oh!

groeneai commented Jun 26, 2026

Uh oh!

groeneai commented Jun 26, 2026

Uh oh!

clickhouse-gh Bot commented Jun 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

AI Review

Summary

Findings

Final Verdict

Uh oh!

Uh oh!

groeneai commented Jun 26, 2026

CI finish ledger — fe57785

Uh oh!

alexey-milovidov commented Jun 28, 2026

Uh oh!

groeneai commented Jun 28, 2026

Uh oh!

alexey-milovidov commented Jun 30, 2026

Uh oh!

alexey-milovidov commented Jun 30, 2026

Uh oh!

alexey-milovidov commented Jun 30, 2026

Uh oh!

alexey-milovidov commented Jul 1, 2026

Uh oh!

alexey-milovidov commented Jul 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

groeneai commented Jun 26, 2026 •

edited by robot-ch-test-poll3

Loading

clickhouse-gh Bot commented Jun 26, 2026 •

edited

Loading

CI finish ledger — `fe57785`