feature (iceberg): Implement TRUNCATE TABLE for Iceberg Engine (REST …

[il9ue]

Changelog category (leave one):

• New Feature

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

Frontport for Antalya 26.3

• 26.1 Antalya port - Implement TRUNCATE TABLE for Iceberg Engine (REST catalog support) Feature: Support TRUNCATE TABLE for Iceberg engine #1529, It's a frontport from 26.1, contains:
  • feat(iceberg): Implement TRUNCATE TABLE for Iceberg Engine (REST catalog support) — Core implementation: metadata-only truncation generating a new overwrite snapshot with empty manifest list, committed atomically via REST catalog
  • fix(iceberg): pass new_snapshot to updateMetadata in IcebergStorageSink — Fixed silent breakage of all INSERTs on REST catalog tables (wrong JSON object passed to catalog->updateMetadata)
  • fix(iceberg): restore return false in RestCatalog::updateMetadata — Preserve retry contract; add LOG_WARNING for diagnostics
  • fix(iceberg): revert Mutations.cpp updateMetadata to pass new_snapshot — Same fix as IcebergStorageSink, applied to ALTER TABLE DELETE/UPDATE path
  • refactor(iceberg): add comment explaining Avro zigzag encoding — Reviewer-requested documentation for manual Avro OCF serialization
  • refactor(iceberg): address code review feedback on TRUNCATE implementation — Named zero arguments, helper functions, restart integration test

Adaptation notes for antalya-26.3

The following API changes between antalya-26.1 and antalya-26.3 required adaptation:

• RestCatalog constructor signature updated to match antalya-26.3 calling convention in DataLakeConfiguration.h
• DatabaseDataLakeSetting member names updated
• Cargo.lock restored to antalya-26.3 version (our changes do not touch any Rust code)

All feature logic is identical to the merged antalya-26.1 implementation.

CI/CD Options

Exclude tests:

• Fast test
• Integration Tests
• Stateless tests
• Stateful tests
• Performance tests
• All with ASAN
• All with TSAN
• All with MSAN
• All with UBSAN
• All with Coverage
• All with Aarch64
• All Regression
• Disable CI Cache

Regression jobs to run:

• Fast suites (mostly <1h)
• Aggregate Functions (2h)
• Alter (1.5h)
• Benchmark (30m)
• ClickHouse Keeper (1h)
• Iceberg (2h)
• LDAP (1h)
• Parquet (1.5h)
• RBAC (1.5h)
• SSL Server (1h)
• S3 (2h)
• S3 Export (2h)
• Swarms (30m)
• Tiered Storage (2h)

[github-actions]

Workflow [PR], commit [cc204b8]

[il9ue]

On CI Failures & Asking a review

Flagging the CI failures for review. All observed failures are pre-existing and unrelated to the Iceberg TRUNCATE changes in this PR.

03706_statistics_preserve_checksums_on_mutations
MergeTree checksum mismatch in Statistics/mutations code path. The full stack trace contains no Iceberg frames. Pre-existing flaky test unrelated to this diff.

Docker CVEs (CVE-2026-28388, CVE-2026-28389, CVE-2026-28390, CVE-2026-31790, CVE-2026-40200)
Grype scan failures against the Alpine base image:

• openssl 3.3.6-r0 — fixed in 3.3.7-r0 (CVEs 28388/28389/28390/31790)
• musl 1.2.5-r9 — fixed in 1.2.5-r11 (CVE-2026-40200)

These require a base image update by the Docker/infra team and are unrelated to any C++ code changes in this PR. The same CVEs appear on the base antalya-26.3 branch independently of this PR.

This PR is a forward-port of the Iceberg TRUNCATE TABLE feature from antalya-26.1 (merged as PR #1529), adapted for antalya-26.3 API changes. All 19 integration tests pass including test_iceberg_truncate and test_iceberg_truncate_restart.

[arthurpassos]

Have you submitted this to upstream?

[arthurpassos]

This will change once #1785 is merged. Not sure about the ordering. Looking at the "process correctness perspective", this frontport should be merged first. At the same time, that PR will certainly introduce conflicts that are best resolved by the author

[CarlosFelipeOR]

QA Verification — ✅ Approved

PR #1655 (Iceberg engine TRUNCATE TABLE — forward-port to antalya-26.3) is safe and did not introduce any regression. This is a forward-port of PR #1529 (merged into antalya-26.1 on 2026-03-31, ~6 weeks of production exposure) — feature logic is identical; only the API-adaptation changes documented in the PR body (and the setting alias change handled at the base-branch level — see Key evidence). The CI run 25840598909 exercises the feature across stateless, integration, and Iceberg regression suites with no failures attributable to this PR. The remaining failures are pre-existing on antalya-26.3, features not yet forward-ported, or regression-suite infrastructure being addressed separately by the QA team.

Feature tests ✅

Test	Type	Result
03376_iceberg_truncate	Stateless	OK on 9 build types (Fast test, amd_asan parallel 2/4, amd_debug s3 storage parallel, amd_debug parallel, amd_msan WasmEdge 2/4, amd_tsan parallel 2/2, amd_tsan s3 storage 2/2, amd_ubsan parallel, arm_asan azure parallel 2/4, arm_binary parallel)
test_iceberg_truncate.py::test_iceberg_truncate_restart	Integration	OK on 5 jobs (amd_asan db disk 5/6, amd_binary 5/5, amd_msan 5/6, amd_tsan 5/6, arm_binary distributed plan 4/4)
iceberg_1 regression suite	Regression	776 OKs / 0 fails on both architectures
iceberg_2 regression suite	Regression	1614 OKs / 0 fails on both architectures

Summary of remaining failures

Category	Tests / Jobs	Status
Features not yet forward-ported to antalya-26.3	swarms (release + aarch64), parquet (release + aarch64), s3_export_part (release + aarch64)	✅ Expected, not caused by PR
Pre-existing 26.3 fails (also failing on unrelated PRs in same time window)	version (release + aarch64), aggregate_functions_3 (aarch64)	✅ Not caused by PR
Regression-test infrastructure issues being addressed by QA team	settings (aarch64) — missing snapshot for new version, fix in flight; tiered_storage_local (aarch64); tiered_storage_s3gcs (aarch64)	✅ Not caused by PR
Known flaky stateless	00157_cache_dictionary on arm_binary, sequential (87 fails / 738 OKs across 116 unrelated PRs in 30d)	✅ Not caused by PR
Caused by this PR	0	🟢

Key evidence

Forward-port precedent — The original PR #1529 introduced this feature into antalya-26.1 and was approved by QA on 2026-03-30 (comment). The C++ feature logic is identical; differences vs #1529 are limited to: RestCatalog constructor signature, DatabaseDataLakeSetting member names, Cargo.lock reverted to 26.3, the throwNotImplemented helper, and use of the canonical setting name allow_insert_into_iceberg (the 26.3 base — independently of this PR — registers allow_experimental_insert_into_iceberg as an alias, so the integration test added by this PR using the old name continues to work).

Cross-PR confirmation of infra failures — The mass-fails in settings (1564 failed / 5 xfail on aarch64) and tiered_storage_local (9 fails) appear identically on unrelated PR runs in the same window (e.g., run 25850287638), confirming they are regression-test infrastructure issues rather than caused by this PR.