Antalya 26:6 Fix alter operations for iceberg by subkanthi · Pull Request #2157 · Altinity/ClickHouse

subkanthi · 2026-08-04T01:00:24Z

continuation of work from #1841

Changelog category (leave one):

Bug Fix (user-visible misbehavior in an official stable release)

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

...

Documentation entry for user-facing changes

...

CI/CD Options

Exclude tests:

Regression jobs to run:

github-actions · 2026-08-04T01:01:58Z

Workflow [PR], commit [62ba6a0]

subkanthi · 2026-08-06T21:11:05Z

ADD COLUMN

Ubuntu-2404-noble-amd64-base :) ALTER table ice.`default.dest5` add column new_column Nullable(UInt64);

ALTER TABLE ice.`default.dest5`
    (ADD COLUMN `new_column` Nullable(UInt64))

Query id: 825e8155-ec00-45f0-a54e-63291f46ef10

Ok.

0 rows in set. Elapsed: 0.314 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`default.dest5`;

SHOW CREATE TABLE ice.`default.dest5`

Query id: 0c79176a-55dd-4aa4-a4c4-2125f5cde5fe

   ┌─statement────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`default.dest5`                        ↴│
   │↳(                                                       ↴│
   │↳    `event_month` Int32,                                ↴│
   │↳    `id` Int64,                                         ↴│
   │↳    `event_time` DateTime64(6),                         ↴│
   │↳    `user_id` Int32,                                    ↴│
   │↳    `category_name` String,                             ↴│
   │↳    `value` Float64,                                    ↴│
   │↳    `payload` String,                                   ↴│
   │↳    `col2` Nullable(Int64),                             ↴│
   │↳    `new_column` Nullable(Int64)                        ↴│
   │↳)                                                       ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/dest6/') │
   └──────────────────────────────────────────────────────────┘

DROP COLUMN

 alter table ice.`default.dest5` drop column col2;

ALTER TABLE ice.`default.dest5`
    (DROP COLUMN col2)

Query id: a7905249-670b-4ac2-a626-f95535c02730

Connecting to database ice at localhost:9007 as user default.
Connected to ClickHouse server version 26.6.2.

Ok.

0 rows in set. Elapsed: 0.122 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`default.dest5`;

SHOW CREATE TABLE ice.`default.dest5`

Query id: 9d3b156a-ca6d-4997-917a-8bf36d4d7058

   ┌─statement────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`default.dest5`                        ↴│
   │↳(                                                       ↴│
   │↳    `event_month` Int32,                                ↴│
   │↳    `id` Int64,                                         ↴│
   │↳    `event_time` DateTime64(6),                         ↴│
   │↳    `user_id` Int32,                                    ↴│
   │↳    `category_name` String,                             ↴│
   │↳    `value` Float64,                                    ↴│
   │↳    `payload` String,                                   ↴│
   │↳    `new_column` Nullable(Int64)                        ↴│
   │↳)                                                       ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/dest6/') │
   └──────────────────────────────────────────────────────────┘

1 row in set. Elapsed: 0.016 sec.

RENAME COLUMN

 alter table ice.`default.dest5` rename column new_column to new_column_2;

ALTER TABLE ice.`default.dest5`
    (RENAME COLUMN new_column TO new_column_2)

Query id: e464f77e-0e3d-4ffc-883e-70db1260327f

Ok.

0 rows in set. Elapsed: 0.169 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`default.dest5`;

SHOW CREATE TABLE ice.`default.dest5`

Query id: 0b4b4e06-7d58-4583-821a-95e27f856ba9

   ┌─statement────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`default.dest5`                        ↴│
   │↳(                                                       ↴│
   │↳    `event_month` Int32,                                ↴│
   │↳    `id` Int64,                                         ↴│
   │↳    `event_time` DateTime64(6),                         ↴│
   │↳    `user_id` Int32,                                    ↴│
   │↳    `category_name` String,                             ↴│
   │↳    `value` Float64,                                    ↴│
   │↳    `payload` String,                                   ↴│
   │↳    `new_column_2` Nullable(Int64)                      ↴│
   │↳)                                                       ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/dest6/') │
   └──────────────────────────────────────────────────────────┘

1 row in set. Elapsed: 0.017 sec.

subkanthi · 2026-08-06T23:41:17Z

 {
    switch (type->getTypeId())
    {
+        case TypeIndex::UInt8:


Added support for bool

subkanthi · 2026-08-06T23:41:31Z

            return {"string", true};
        case TypeIndex::UUID:
            return {"uuid", true};
+        case TypeIndex::Decimal32:


Added support for iceberg decimal types.

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

subkanthi · 2026-08-07T20:19:46Z

Decimal support


ALTER TABLE ice.`flowers.sample`
ADD COLUMN dec_col Nullable(Decimal(10,2));

ALTER TABLE ice.`flowers.sample`
    (ADD COLUMN `dec_col` Nullable(Decimal(10, 2)))

Query id: 83a0ae12-2b00-40de-85ba-654daa0d7dc0

Ok.

0 rows in set. Elapsed: 0.203 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`flowers.sample`;

SHOW CREATE TABLE ice.`flowers.sample`

Query id: bb3b8676-9b7e-4000-9b08-2fcb30b65079

   ┌─statement─────────────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`flowers.sample`                                ↴│
   │↳(                                                                ↴│
   │↳    `value` Nullable(Int64),                                     ↴│
   │↳    `boolean_col` Nullable(Bool),                                ↴│
   │↳    `dec_col` Nullable(Decimal(10, 2))                           ↴│
   │↳)                                                                ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/flowers/sample/') │
   └───────────────────────────────────────────────────────────────────┘

1 row in set. Elapsed: 0.009 sec.

mkmkme · 2026-08-11T10:26:11Z

I asked Claude to analyze this PR and it came with some review. Please have a look: https://gist.github.com/mkmkme/b28a41f136a62d5a6d41a73dee0132bd

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

… fix medium defects

subkanthi · 2026-08-11T23:06:30Z


Ubuntu-2404-noble-amd64-base :) alter table ice.`flowers.sample` add column dec_col2 Nullable(Decimal(10,2));

ALTER TABLE ice.`flowers.sample`
    (ADD COLUMN `dec_col2` Nullable(Decimal(10, 2)))

Query id: 37668c59-de09-4b90-84ec-a542e702b05a

Ok.

0 rows in set. Elapsed: 0.249 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`flowers.sample`;

SHOW CREATE TABLE ice.`flowers.sample`

Query id: 54a5faf6-7016-41ab-865d-93f6102f6808

   ┌─statement─────────────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`flowers.sample`                                ↴│
   │↳(                                                                ↴│
   │↳    `value` Nullable(Int64),                                     ↴│
   │↳    `boolean_col` Nullable(Bool),                                ↴│
   │↳    `dec_col` Nullable(Decimal(10, 2)),                          ↴│
   │↳    `dec_col2` Nullable(Decimal(10, 2))                          ↴│
   │↳)                                                                ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/flowers/sample/') │
   └───────────────────────────────────────────────────────────────────┘

1 row in set. Elapsed: 0.012 sec. 

Ubuntu-2404-noble-amd64-base :) alter table ice.`flowers.sample` drop column dec_col2;

ALTER TABLE ice.`flowers.sample`
    (DROP COLUMN dec_col2)

Query id: 49447f08-f164-47a1-aff3-d4852e4736cb

Ok.

0 rows in set. Elapsed: 0.147 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`flowers.sample`;

SHOW CREATE TABLE ice.`flowers.sample`

Query id: 8cd4a422-13d5-472a-9cf2-5208c79e0a99

   ┌─statement─────────────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`flowers.sample`                                ↴│
   │↳(                                                                ↴│
   │↳    `value` Nullable(Int64),                                     ↴│
   │↳    `boolean_col` Nullable(Bool),                                ↴│
   │↳    `dec_col` Nullable(Decimal(10, 2))                           ↴│
   │↳)                                                                ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/flowers/sample/') │
   └───────────────────────────────────────────────────────────────────┘

1 row in set. Elapsed: 0.006 sec. 

Ubuntu-2404-noble-amd64-base :) alter table ice.`flowers.sample` rename column dec_col to dec_col_2;

ALTER TABLE ice.`flowers.sample`
    (RENAME COLUMN dec_col TO dec_col_2)

Query id: 816db79c-7370-4ab2-8a6c-d8535337c2aa

Ok.

0 rows in set. Elapsed: 0.119 sec. 

Ubuntu-2404-noble-amd64-base :) show create table ice.`flowers.sample`;

SHOW CREATE TABLE ice.`flowers.sample`

Query id: b752811b-2a78-4097-ba40-5411e1f2a140

   ┌─statement─────────────────────────────────────────────────────────┐
1. │ CREATE TABLE ice.`flowers.sample`                                ↴│
   │↳(                                                                ↴│
   │↳    `value` Nullable(Int64),                                     ↴│
   │↳    `boolean_col` Nullable(Bool),                                ↴│
   │↳    `dec_col_2` Nullable(Decimal(10, 2))                         ↴│
   │↳)                                                                ↴│
   │↳ENGINE = Iceberg('http://localhost:9000/bucket1/flowers/sample/') │
   └─────────────────────────────────────────────────────────────────

subkanthi · 2026-08-13T03:50:30Z

I asked Claude to analyze this PR and it came with some review. Please have a look: https://gist.github.com/mkmkme/b28a41f136a62d5a6d41a73dee0132bd

@mkmkme can u check again please

mkmkme · 2026-08-13T13:57:34Z

I asked Claude to analyze this PR and it came with some review. Please have a look: https://gist.github.com/mkmkme/b28a41f136a62d5a6d41a73dee0132bd

@mkmkme can u check again please

@subkanthi the new one is quite cleaner, but there're still some issues worth fixing: https://gist.github.com/mkmkme/98ffbceed6d86e14339537481a5f162c

DimensionWieldr · 2026-08-13T15:08:34Z

+
 }

 MetadataGenerator::MetadataGenerator(Poco::JSON::Object::Ptr metadata_object_)


Just noting that dropping a column used in partitioning or ordering has had its behaviour changed, especially that of an ordering column which is now completely rejected.

This is an iceberg spec:

(Section "Schema Evolution") explicitly forbids this: a column that is part of the current sort order or partition spec cannot be removed without first changing the sort order/partition spec to no longer reference it.

subkanthi · 2026-08-13T15:48:46Z

I asked Claude to analyze this PR and it came with some review. Please have a look: https://gist.github.com/mkmkme/b28a41f136a62d5a6d41a73dee0132bd

@mkmkme can u check again please

@subkanthi the new one is quite cleaner, but there're still some issues worth fixing: https://gist.github.com/mkmkme/98ffbceed6d86e14339537481a5f162c

Thanks @mkmkme , i will take a shot at fixing 1, 2, for #5 Glue catalog we have separate issues which I thought is better to go in a separate PR.
5. A failed schema commit on a non-transactional catalog permanently bricks ALTER on that table — and the new error message actively misleads. For Glue/Unity/Hive, Mutations.cpp writes v<N+1>.metadata.json itself, then calls catalog->updateSchema. If that returns false, the loop retries; but the retry re-reads metadata from the catalog's location (Mutations.cpp:832-859), which still points at N. So it regenerates, targets N+1 again, and writeMetadataFileAndVersionHint returns false because the object already exists (Utils.cpp:300-301). 100 iterations, then:

ALTER TABLE commit kept losing to concurrent modifications after 100 retries

There was no concurrent modification. And the orphan v<N+1>.metadata.json makes every subsequent ALTER on that table repeat this until someone deletes it by hand. The loop structure is pre-existing; the misleading message is new, and the unused cleanup failpoint suggests the real fix was intended.

subkanthi · 2026-08-13T19:42:16Z

I asked Claude to analyze this PR and it came with some review. Please have a look: https://gist.github.com/mkmkme/b28a41f136a62d5a6d41a73dee0132bd

@mkmkme can u check again please

@subkanthi the new one is quite cleaner, but there're still some issues worth fixing: https://gist.github.com/mkmkme/98ffbceed6d86e14339537481a5f162c

Thanks @mkmkme , i will take a shot at fixing 1, 2, for #5 Glue catalog we have separate issues which I thought is better to go in a separate PR. 5. A failed schema commit on a non-transactional catalog permanently bricks ALTER on that table — and the new error message actively misleads. For Glue/Unity/Hive, Mutations.cpp writes v<N+1>.metadata.json itself, then calls catalog->updateSchema. If that returns false, the loop retries; but the retry re-reads metadata from the catalog's location (Mutations.cpp:832-859), which still points at N. So it regenerates, targets N+1 again, and writeMetadataFileAndVersionHint returns false because the object already exists (Utils.cpp:300-301). 100 iterations, then:

ALTER TABLE commit kept losing to concurrent modifications after 100 retries

There was no concurrent modification. And the orphan v<N+1>.metadata.json makes every subsequent ALTER on that table repeat this until someone deletes it by hand. The loop structure is pre-existing; the misleading message is new, and the unused cleanup failpoint suggests the real fix was intended.

Scope
Review of Iceberg ALTER TABLE schema evolution logic: MetadataGenerator, RestCatalog, Mutations, DataLakeConfiguration, FailPoint registration, and associated integration/unit tests.

Confirmed Defects (All Fixed)
High
1. Null-pointer dereference in icebergTypesEqual for complex types

Impact: Server exception on any ALTER TABLE ... MODIFY COLUMN with Array/Map/Tuple types.
File: src/Storages/ObjectStorage/DataLakes/Iceberg/MetadataGenerator.cpp
Fix: Commit a73db85 -- added null guard before extracting Poco::JSON::Object::Ptr.
Medium
2. isAddColumnApplied false negative for complex types (Array/Map/Tuple)

Impact: Retry after "commit state unknown" would re-apply an already-committed ADD COLUMN, causing a catalog-level conflict or schema corruption for non-primitive columns.
Root cause: icebergTypesEqual performed exact JSON comparison including embedded field IDs. A previous commit allocated IDs from a lower last-column-id, so the same structural type produces a different JSON descriptor on retry.
File: src/Storages/ObjectStorage/DataLakes/Iceberg/MetadataGenerator.cpp:152-175, 240-261
Fix: Commit 7dc6cfb -- introduced stripNestedFieldIds + icebergTypesEqualIgnoringIds that deep-clones and removes id/element-id/key-id/value-id before comparison. isAddColumnApplied now uses this relaxed comparison.
Test: gtest_iceberg_metadata_generator.cpp -- AddColumnAppliedDetectsCommittedComplexColumn, AddColumnAppliedRejectsComplexTypeMismatch.
3. Schema deduplication in buildUpdateSchemaRequestBody never fires

Impact: Every schema update creates a new schema entry in the schemas array even when the schema is structurally identical to an existing one. This bloats metadata, violates the Iceberg spec's intent for deduplication, and can trigger "too many schemas" limits in some catalogs.
Root cause: schemasEquivalentIgnoringId stripped schema-id but not identifier-field-ids. Schemas committed through the REST catalog carry "identifier-field-ids": [] while freshly generated schemas omit the field entirely. The mismatch caused icebergJsonObjectEquals to always return false.
File: src/Databases/DataLake/RestCatalog.cpp:161, 227-239
Fix: Commit 7dc6cfb -- schemasEquivalentIgnoringId now also strips identifier-field-ids when it is null or an empty array, treating absent-vs-empty as semantically identical.
Test: gtest_rest_catalog_update_metadata.cpp -- EquivalentSchemaDeduplicatesAcrossIdentifierFieldIds, NonEmptyIdentifierFieldIdsPreventDeduplication.
4. MODIFY COLUMN on geometry types throws a misleading error

Impact: ALTER TABLE ... MODIFY COLUMN g Geometry on an existing geometry column throws "Using geometry/geography types is not allowed without allow_experimental_geo_types_in_iceberg flag" even when the column already has that type (should be a no-op).
Root cause: MetadataGenerator hard-coded allow_geo_parser = false in calls to getSimpleType/getIcebergType, so geometry type lookup always failed regardless of the user's setting.
File: src/Storages/ObjectStorage/DataLakes/Iceberg/MetadataGenerator.h:20, Mutations.cpp:875
Fix: Commit 5968163 -- MetadataGenerator constructor now accepts bool allow_geo_parser_ and threads it to type-resolution calls. Mutations.cpp passes context->getSettingsRef()[Setting::allow_experimental_geo_types_in_iceberg].
Low
5. Wrong doc comment on buildUpdateSchemaRequestBody

Impact: Misleading for developers; no runtime effect.
File: src/Databases/DataLake/RestCatalog.h:248-258
Fix: Commit 846e827 -- replaced with an accurate description of schema-update body construction.
6. Dead code: buildUpdateMetadataRequestBody null-snapshot check

Impact: Unreachable branch; misleading for maintainers.
File: src/Databases/DataLake/RestCatalog.cpp (caller throws NOT_IMPLEMENTED before the builder is reached).
Fix: Commit 5968163 -- removed the dead if (!new_snapshot) return nullptr; branch.
7. Dead failpoint: iceberg_alter_orphan_metadata_cleanup_fail

Impact: Registered in FailPoint.cpp with no injection site anywhere -- leftover from unimplemented cleanup logic.
Fix: Commit 5968163 -- removed registration entirely.
8. Misnamed failpoint: iceberg_alter_catalog_update_metadata_fail

Impact: Injected in updateSchema, not updateMetadata; no test exercised it.
Fix: Commit 5968163 -- renamed to iceberg_alter_catalog_update_schema_fail.
9. Unreachable failpoint datalake_iceberg_metadata_create_fail with weakened invariant

Impact: The failpoint fabricated an impossible state (null metadata after create), and assertInitialized was downgraded from LOGICAL_ERROR to NOT_INITIALIZED solely to let the test assert on it. This lost CI/fuzzer signal.
Fix: Commit 5968163 -- removed the failpoint, removed the test test_writes_when_metadata_is_not_initialized, restored LOGICAL_ERROR in assertInitialized.
10. Stale test assertion in test_writes_modify_column.py

Impact: The overly broad assertion ("bad cast" in el or "can not convert" in el or ...) would pass even if the error message regressed to an unrelated exception.
Fix: Commit 5968163 -- tightened to assert "doesn't allow schema evolution" in el.
Residual Risks / Not Tested
Signed overflow in getNextSchemaId -- Returns max_id + 1 where max_id is Int32. Overflow at INT32_MAX schemas is theoretically possible but requires ~2 billion schema evolutions -- not a realistic production scenario.
Retry logic for MODIFY COLUMN -- No isModifyColumnApplied predicate exists. A "commit state unknown" during MODIFY cannot be safely retried without it. Low risk: MODIFY is rare and type widening is idempotent in Iceberg.
deepCopy via stringify/parse -- Used in stripNestedFieldIds path. Allocates and parses JSON on every isAddColumnApplied call. Acceptable for ALTER frequency but would be problematic if ever called in a hot path.
Summary
10 findings total: 1 High, 4 Medium, 5 Low.
All confirmed defects are fixed across commits a73db85, 846e827, 5968163, c72983e, 7dc6cfb.
New unit tests cover the two medium-severity fixes (complex-type retry detection, schema deduplication).
Integration test assertions tightened for regression coverage.
Confidence: High -- all fixes verified by compilation and unit test execution; logic validated by code inspection against the Iceberg spec.```

mkmkme · 2026-08-14T10:41:44Z

Could you fix the conflicts please?

DimensionWieldr · 2026-08-14T13:33:30Z

Agent found a potential issue:

Nested ADD COLUMN + commit-unknown recovery

iceberg_alter_catalog_commit_reported_as_failed makes RestCatalog::updateSchema apply the catalog commit and then return false. The ALTER retry path is supposed to notice the column is already there via MetadataGenerator::isAddColumnApplied and treat the op as succeeded.

That works for a primitive ADD COLUMN z Nullable(String) (test_writes_alter_when_commit_is_reported_as_failed). It does not work for nested types.

SET allow_insert_into_iceberg = 1, enable_nullable_tuple_type = 1;
ALTER TABLE … ADD COLUMN tuple_col Nullable(Tuple(Int32, String));

Code: 36. DB::Exception: Column tuple_col already exists. (BAD_ARGUMENTS)

The catalog already committed the column. Recovery missed it, so the retry called generateAddColumnMetadata again.

Cause

isAddColumnApplied (MetadataGenerator.cpp ~181–199) rebuilds the expected Iceberg type with getIcebergType(type, last_column_id) using the current last-column-id (already advanced by the successful commit).

For primitives that is a string ("string", "long", …) and the compare succeeds.

For Array / Map / Tuple, getIcebergType allocates new nested field ids from that high watermark. icebergTypesEqual stringifies the JSON, so element-id / key-id / value-id / struct field ids do not match the committed type → returns false → retry → Column already exists.

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

DimensionWieldr · 2026-08-14T22:53:54Z

PR CI Triage Summary

Iceberg ALTER coverage passed (iceberg_1). None of these are caused by this PR. Same jobs are already red on MasterCI antalya-26.6
(https://github.com/Altinity/ClickHouse/actions/runs/31713806632).

iceberg_2 — EXPORT PARTITION tests against a 26.6 binary that does not have that feature. Unrelated to Iceberg ALTER.

s3_export_part — same export-feature mismatch, then cascade after the server died. Sanity cases passed.

settings — missing snapshots for new 26.6 settings. Already failing on MasterCI.

Stateless amd_debug + distributed plan + s3 — three 600s timeouts. Passed on the other stateless jobs in this run. Known flakes.

swarms — historically flaky node-restart test. Already failing on MasterCI.

tiered_storage_minio — disk-name flake. Passed on the other tiered-storage jobs in this run and on MasterCI.

Fix alter operations for iceberg

e93ce1c

subkanthi added 2 commits August 4, 2026 10:18

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

fa311a9

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

818f446

subkanthi changed the title ~~Fix alter operations for iceberg~~ Antalya 26:6 Fix alter operations for iceberg Aug 6, 2026

subkanthi added 3 commits August 5, 2026 22:16

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

d62b9ac

fix medium defects with retries

0eb31fc

Added support for boolean and decimal

f76fd1d

subkanthi marked this pull request as ready for review August 6, 2026 04:26

subkanthi added 2 commits August 6, 2026 16:46

Add missing include to gtest_iceberg_type_mapping

bdb0d15

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

d3b65e3

subkanthi closed this Aug 6, 2026

subkanthi reopened this Aug 6, 2026

subkanthi added the antalya-26.6 label Aug 6, 2026

subkanthi closed this Aug 6, 2026

subkanthi reopened this Aug 6, 2026

subkanthi closed this Aug 6, 2026

subkanthi reopened this Aug 6, 2026

subkanthi closed this Aug 6, 2026

subkanthi reopened this Aug 6, 2026

subkanthi commented Aug 6, 2026

View reviewed changes

subkanthi added 2 commits August 7, 2026 03:13

Restored removed tests

04c009e

Merge branch 'antalya_26_6_fix_alter_table_iceberg' of https://github…

a4e6f39

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

This was referenced Aug 7, 2026

Iceberg DataLakeCatalog: ALTER DELETE fails with Metadata is not initialized (LOGICAL_ERROR) #2136

Open

Iceberg ALTER ADD COLUMN does not support Bool/Decimal #2089

Open

subkanthi added 2 commits August 7, 2026 22:34

Fix to support decimal datatypes in iceberg

6e6f1f7

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

ad9e5c7

subkanthi added 9 commits August 11, 2026 18:29

Fix nullptr exception when there are no snapshots

8e9c6f7

Merge branch 'antalya_26_6_fix_alter_table_iceberg' of https://github…

3da03ba

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

Fix compiler error

bedcad9

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

304fc24

Fix compiler error

12df291

Merge branch 'antalya_26_6_fix_alter_table_iceberg' of https://github…

7f35e6c

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

Revert back change to settings regarding iceberg manifest compaction,…

d4c5df2

… fix medium defects

Reverted back settings changes(removal)

422f409

Reverted back settings changes(removal)

8bbea38

subkanthi added 2 commits August 12, 2026 01:07

Address medium defects

c72983e

Fix high defect with nullptr in failsafe

a73db85

DimensionWieldr reviewed Aug 13, 2026

View reviewed changes

subkanthi added 3 commits August 13, 2026 18:58

Fix wrong doc comment

846e827

Remove dead code, fix/remove failpoint

5968163

Address medium defects

7dc6cfb

subkanthi requested review from DimensionWieldr and mkmkme August 13, 2026 20:14

subkanthi added 2 commits August 14, 2026 09:15

Merge branch 'antalya-26.6' into antalya_26_6_fix_alter_table_iceberg

a983d97

Merge branch 'antalya_26_6_fix_alter_table_iceberg' of https://github…

3a09c32

….com/Altinity/ClickHouse into antalya_26_6_fix_alter_table_iceberg

DimensionWieldr mentioned this pull request Aug 14, 2026

Iceberg ADD COLUMN of nested types fails commit-unknown recovery with Column already exists #2215

Open

Replace AssertInitialized with AssertInitializedDL

62ba6a0


		}

		MetadataGenerator::MetadataGenerator(Poco::JSON::Object::Ptr metadata_object_)

Uh oh!

Conversation

subkanthi commented Aug 4, 2026 • edited 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 to CHANGELOG.md):

Documentation entry for user-facing changes

CI/CD Options

Exclude tests:

Regression jobs to run:

Uh oh!

github-actions Bot commented Aug 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

subkanthi commented Aug 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

subkanthi Aug 6, 2026

Choose a reason for hiding this comment

Uh oh!

subkanthi Aug 6, 2026

Choose a reason for hiding this comment

Uh oh!

subkanthi commented Aug 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mkmkme commented Aug 11, 2026

Uh oh!

subkanthi commented Aug 11, 2026

Uh oh!

subkanthi commented Aug 13, 2026

Uh oh!

mkmkme commented Aug 13, 2026

Uh oh!

DimensionWieldr Aug 13, 2026

Choose a reason for hiding this comment

Uh oh!

DimensionWieldr Aug 13, 2026

Choose a reason for hiding this comment

Uh oh!

subkanthi commented Aug 13, 2026

Uh oh!

subkanthi commented Aug 13, 2026

Uh oh!

mkmkme commented Aug 14, 2026

Uh oh!

DimensionWieldr commented Aug 14, 2026

Nested ADD COLUMN + commit-unknown recovery

Cause

Uh oh!

DimensionWieldr commented Aug 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR CI Triage Summary

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

subkanthi commented Aug 4, 2026 •

edited

Loading

github-actions Bot commented Aug 4, 2026 •

edited

Loading

subkanthi commented Aug 6, 2026 •

edited

Loading

subkanthi commented Aug 7, 2026 •

edited

Loading

DimensionWieldr commented Aug 14, 2026 •

edited

Loading